How plants cope with complete submergence.

Flooding is a widespread phenomenon that drastically reduces the growth and survival of terrestrial plants. The dramatic decrease of gas diffusion in water compared with in air is a major problem for terrestrial plants and limits the entry of CO(2) for photosynthesis and of O(2) for respiration. Responses to avoid the adverse effects of submergence are the central theme in this review. These include underwater photosynthesis, aerenchyma formation and enhanced shoot elongation. Aerenchyma facilitates gas diffusion inside plants so that shoot-derived O(2) can diffuse to O(2)-deprived plant parts, such as the roots. The underwater gas-exchange capacity of leaves can be greatly enhanced by a thinner cuticle, reorientation of the chloroplasts towards the epidermis and increased specific leaf area (i.e. thinner leaves). At the same time, plants can outgrow the water through increased shoot elongation, which in some species is preceded by an adjustment of leaf angle to a more vertical position. The molecular regulatory networks involved in these responses, including the putative signals to sense submergence, are discussed and suggestions made on how to unravel the mechanistic basis of the induced expression of various adaptations that alleviate O(2) shortage underwater.


I. Introduction
Flooding is one of the most common and widespread of all natural disasters. It is detrimental for nearly all terrestrial plants and results in hampered growth and ultimately death of many plant species.
Flooding can have catastrophic impacts on the productivity of arable farmland, as most crops are intolerant to excess water (http:// www.plantstress.com/Articles/waterlogging_i/waterlog_i.htm).
The negative impact of flooding on terrestrial plant life is a consequence of the slow diffusion rates of gases in water compared with in air and the relatively low solubility of O 2 in water ( Jackson, 1985;Armstrong & Drew, 2002). In nonphotosynthesising plant tissues, such as roots, O 2 concentrations decline strongly upon submergence (Gaynard & Armstrong, 1987;Armstrong et al ., 1994), whereas CO 2 produced during respiration and fermentation probably accumulates (Crawford, 1992). Submerged green tissues, on the other hand, are characterized by a significant diurnal variation in gas composition (Raskin & Kende, 1984a). During daytime, photosynthesis depletes endogenous CO 2 concentrations, whereas O 2 concentrations can be maintained close to 21%. During the night, however, endogenous O 2 concentrations are reduced as a result of respiration, but complete depletion within shoot tissues should be rare because the surrounding water generally contains O 2 (Mommer et al ., 2004). A third, endogenously produced, gas, which changes in concentration when plant tissues are submerged, is the plant hormone ethylene. Physical entrapment, combined with a very low rate of catabolism (Hall, 1991), leads to an increase of ethylene inside the tissues of submerged plants. A physiologically relevant increase in ethylene concentration can occur within minutes after submergence (Vreeburg et al ., 2005). Despite these dramatic changes in gas composition and the stress it imposes, some plant species, especially those growing in flood-prone areas, have a remarkable tolerance and can grow vigorously under flooded conditions. However, species from flood-exposed environments are not always flood tolerant. Some avoid the adverse effects of flooding by timing of life cycle events (Blom & Voesenek, 1996). Here, we review how plants regulate morphological adjustments that help to prevent severe O 2 shortages during submergence.
It is generally accepted that energy deficit, caused by a shortage of O 2 , and thus inhibition of respiration, is one of the most severe problems encountered by plants when subjected to flooding. Oxygen is the terminal acceptor of electrons in the oxidative phosphorylation that indirectly provides the plant with ATP. Metabolic acclimations that aim to continue energy production in plants during anoxia were reviewed recently by Sauter (2000), Dolferus et al . (2001), Geigenberger (2003),  and . When O 2 concentrations decline (e.g. ≤ 5%), glycolysis and fermentation pathways are stimulated in Zea mays  -the classic 'Pasteur' effect. Consistent with this biochemical observation at low conentrations of O 2 ( ≤ 5%) are the results of several microarray studies with Arabidopsis that showed an up-regulation of genes coding for enzymes of sugar metabolism, glycolysis and fermentation upon low O 2 concentrations (Klok et al ., 2002;Liu et al ., 2005;Loreti et al ., 2005). The induction of these pathways allows continuation of at least some energy production when anoxia starts. Low O 2 also induces the expression of nonsymbiotic hemoglobin genes (Dordas et al ., 2004). It has been suggested for Z. mays that hemoglobin can detoxify NO produced during hypoxia (Dordas et al ., 2003) and/or play a role in the regeneration of NAD + necessary to maintain glycolysis (Sowa et al ., 1998;Igamberdiev et al ., 2005). Another group of genes that are up-regulated at low O 2 concentrations are associated with the detoxification of reactive oxygen species (peroxidase, ascorbate peroxidase, monohydroascorbate, gluthatione reductase and superoxide dismutase) (Klok et al ., 2002). It is assumed that the products of these genes play roles in the protection against postanoxia injury (Monk et al ., 1989).
A second strategy in plants that inhabit areas prone to submergence is functionally associated with the long-term avoidance of low internal O 2 partial pressures. For example, a major adjustment induced by complete submergence is a change in leaf morphology and anatomy. Leaves of Rumex palustris , R. crispus and Phalaris arundinacea developed underwater have a higher specific leaf area (SLA; m 2 kg − 1 ) (Vervuren et al ., 1999;Mommer et al ., 2005a), and R. palustris leaves also have a thinner epidermal cell wall and cuticle, and the chloroplasts are oriented towards the epidermis instead of the intercellular spaces (Mommer et al ., 2005b). These acclimations result in higher underwater rates of photosynthesis and lower CO 2 compensation points, ultimately generating higher internal O 2 concentrations in submerged plants (Mommer et al ., 2004(Mommer et al ., , 2005a. In addition, many species show stimulated elongation of shoot organs; a trait that can restore contact of the leaves with air. Aerenchyma formation is also common in wetland species. The aerenchyma creates longitudinally interconnected pathways of gas spaces that, together with the renewed contact of leaf tips with the atmosphere through enhanced shoot elongation, enable diffusion of O 2 from the air to the root tips. Ethylene is a key regulator of the formation of lysigenous aerenchyma, at least in Z. mays , through programmed cell death (Drew et al ., 2000), and stimulated shoot elongation mediated by cell-wall loosening in rice and Rumex (Voesenek & Blom, 1999;Sauter, 2000;Vreeburg et al ., 2005). In Arabidopsis , genes involved in the regulation of ethylene synthesis, ethylene signaling, programmed cell death and cell-wall loosening are up-regulated upon low O 2 conditions (Klok et al ., 2002;Liu et al ., 2005).
The present review summarizes the recent progress in the molecular and physiological regulation of mechanisms that contribute to long-term anoxia avoidance, namely stimulated shoot elongation, aerenchyma formation and underwater photosynthesis, which ultimately enhance the internal O 2 status of plants in flooded environments. In addition, the importance of understanding submergence sensing, and the opportunities for rapid progress in submergence research by the use of Arabidopsis , will be highlighted.

II. Sensing of submergence
In order to react appropriately to the flooded environment, plants have to sense the changes that take place. The changes in endogenous gas composition make O 2 , CO 2 and C 2 H 4 potential molecules to signal submergence. However, the availability of light complicates the matter with respect to CO 2 and O 2 because their internal concentrations are strongly influenced by both photosynthesis and respiration. For this reason, C 2 H 4 is considered to be a far more reliable indicator for complete submergence in green shoot tissues (Voesenek & Blom, 1999). In addition to these gases, reactive oxygen species (ROS), which may accumulate during flooding, have also been proposed to have a regulatory role during submergence (Bailey-Serres & Chang, 2005;Blokhina et al ., 2003). Furthermore, we cannot exclude a direct role for light quality and quantity as submergence signals as both can change upon submergence (Holmes & Klein, 1987).

Ethylene
In R. palustris , ethylene concentrations increase 20-fold within the first hour of submergence as a result of physical entrapment (Voesenek et al ., 1993;Banga et al ., 1996). The production rate of ethylene during submergence is maintained at the same level in Rumex (Voesenek et al ., 1993;Vriezen et al ., 1999), or increased, as in the internode tissue of deepwater rice (Kende et al ., 1998). Biosynthesis of ethylene is a wellcharacterized process (Wang et al ., 2002) that is dependent on the presence of O 2 (Kende, 1993). In R. palustris , it was shown that ethylene plays a pivotal role in initiating responses to submergence, although it is not fully clear whether it is the sensing signal. In flooding-tolerant Rumex species, the 1aminocyclopropane-1-carboxylic acid concentration (ACC, a precurser of ethylene) strongly increased upon submergence, whereas the conversion to ethylene was inhibited. At the gene expression and protein levels ( in vitro assay), increased expression of ACC synthase genes, and resulting enzyme activity, were observed (Vriezen et al ., 1999;Voesenek et al ., 2003;Rieu et al ., 2005). In R. palustris , ACC oxidases are strongly up-regulated during submergence and this correlates with an increase in enzyme activity, measured in vitro (Vriezen et al ., 1999;Voesenek et al., 2003). The increased concentration of ACC oxidase enzyme during submergence possibly counterbalances the reduced enzyme activity at low O 2 concentrations, thus sustaining ethylene production during submergence (Vriezen et al., 1999).
In plants, ethylene is detected by a set of receptors. These and the subsequent signal-transduction cascade have been elegantly clarified over the last decade in A. thaliana, using genetic techniques (Chang, 2003;Alonso & Stepanova, 2004;Guo & Ecker, 2004;Stepanova & Alonso, 2005). An R. palustris orthologue of the ethylene response sensor (ERS) ethylene receptor gene was isolated and shown to be strongly up-regulated upon submergence. This desensitizes the tissue towards ethylene , which results from the fact that ethylene receptors actively suppress a downstream regulator of ethylene responses when the receptors have not bound ethylene. This desensitization has probably no effect on the ethylene-induced responses, as ethylene concentrations were found to be saturated in the shoots of submerged plants (Voesenek et al., 1993;Banga et al., 1996).

Carbon dioxide
The role of CO 2 as a sensing molecule in submergence research has received limited attention. High concentrations of CO 2 do not affect petiole elongation in R. palustris  but it was shown, in rice, that elevated CO 2 concentrations (6%) could induce elongation of internodes, albeit to a much lesser extent than ethylene (Raskin & Kende, 1984a). Recently, Mommer et al. (2005b) showed, in the leaves of submerged plants, reorientation of chloroplasts towards the leaf exterior where CO 2 diffuses in through the cuticle and cell wall. This, and other changes, led to a decreased diffusion resistance for CO 2 to reach Rubisco, resulting in higher underwater assimilation rates at low external CO 2 concentrations. These data suggest that plant cells, in addition to guard cells, are able to sense the CO 2 concentration and respond to it accordingly. Whether these alterations are caused by CO 2 sensing, possibly through a mechanism as in guard cells, or by other mechanisms (perhaps even independent of CO 2 ), is still to be elucidated.
Much work has been carried out on the sensing of CO 2 in relation to stomatal opening and closure. However, despite the numerous publications on this subject, the CO 2 -sensing pathway(s) of guard cells remain poorly understood and may involve cytosolic pH and malate, zeaxanthin, Ca 2+ , or anion channels in the short-term response to CO 2 (Assmann, 1999;Schroeder et al., 2001).
Elevated concentrations of CO 2 also affect metabolic rates of plant species, as reviewed by (Gonzàles-Meler et al., 1996). Mangrove seedlings demonstrated a marked reduction in the rate of respiration when exposed to low O 2 concentrations in the presence of external CO 2 (Brown et al., 1969). By contrast, exposure of dryland species, such as Cicer ariefinum, Tussilago farfara and Elymus repens, to elevated concentrations of CO 2 up-regulated the glycolytic activity and reduced the viability (Crawford, 1992).

Oxygen
The O 2 concentration in plant tissues decreases upon submergence (Rijnders et al., 2000). Furthermore, it has been shown that low O 2 could sensitize petiole tissues to ethylene, thus suggesting an interaction between ethylene and O 2 . Another indication of the interaction between ethylene and O 2 was found by Peng et al. (2001), who showed that ethylene signaling is required for hypoxia-induced gene expression. Furthermore, Kamaluddin & Zwiazek (2002) showed that enhanced ethylene concentrations can overcome the hampered hydraulic conductivity in roots caused by low O 2 concentrations (Tournaire-Roux et al., 2003).
Decreases of O 2 , ranging from low concentrations (hypoxia) to complete absence (anoxia), in plants have been associated with changes in the expression of genes related to carbohydrate and lipid metabolism, glycolysis, fermentation pathways, ethylene synthesis, auxin-mediated processes, and calcium-and ROS-mediated signal transduction pathways (Klok et al., 2002;Paul et al., 2003;Branco-Price et al., 2005;Liu et al., 2005;Loreti et al., 2005). Moreover, morphological responses, such as aerenchyma and adventitious root formation, and cell and organ elongation of shoots, are induced in some species. To date, no direct O 2 sensor in plants has been identified (Drew, 1997;Geigenberger, 2003). This is in contrast to animal cells where the hypoxia-inducible heterodimeric transcription factor (HIF, reviewed by Bruick, 2003) subunit, HIF1α, is regulated by an O 2 -dependent HIF1α prolyl hydroxylase. It was shown that the activity of this enzyme in Drosophila melanogaster and Caenorhabditis elegans is proportional to the O 2 availability and, as a consequence, this system might be considered as a direct O 2 -sensing system (Bruick, 2003). Another (albeit indirect) mechanism by which O 2 deprivation might be sensed is the rapid acidification of the cytosol upon O 2 deprivation (Roberts et al., 1984). This is a quick response to the energy crisis experienced by cells under anoxia. The lower pH is relatively stable in anoxia-tolerant tissues, but can decline further in less tolerant tissues (Felle, 2005). There has been much debate about the precise role of cytosolic acidification, but it can affect cellular metabolism (Geigenberger, 2003) and water conductivity across membranes (Tournaire-Roux et al., 2003), and could thus play a role in indirect O 2 sensing (Bailey-Serres & Chang, 2005; reviewed in Felle, 2005).

Hydrogen peroxide
It is generally accepted that biotic and abiotic stresses, among these hypoxia and anoxia, lead to a change in H 2 O 2 production (Apel & Hirt, 2004;Mittler et al., 2004). In particular, H 2 O 2 produced by cytosolic membrane-bound NADPH oxidases has been associated with stress responses (Laloi et al., 2004). As H 2 O 2 is potentially damaging, there are cellular antioxidant mechanisms that remove it very efficiently (reviewed in Noctor & Foyer, 1998;Corpas et al., 2001).
Recent reviews by Neill et al. (2002), Laloi et al. (2004), Apel & Hirt (2004) and Mittler et al. (2004) elaborate on the progress made on various aspects in ROS research. More specifically, Bailey-Serres & Chang (2005) discuss the current state-of-the-art about the role of ROS in sensing O 2 deprivation. There is evidence that the change in ROS in the particular cases of hypoxia and anoxia in plants are mediated by a group of compounds called the RHO-like GTPases (ROP). Activation of ROP after treatment with a low concentration of O 2 induces H 2 O 2 production and this seems necessary for low O 2 signaling (Baxter-Burrell et al., 2002, 2003Bailey-Serres & Chang, 2005). Components of the signal-transduction pathway of ROS itself have been revealed by recent studies (reviewed in Apel & Hirt, 2004;Mittler et al., 2004). ROS might be sensed by two-component sensors possessing a histidine kinase, subsequently inducing a mitogen-activated protein kinase (MAPK) signaling pathway, and it has been speculated that ROS may inhibit protein phosphatases, and with some evidence also that ROS can activate transcription factors directly. All of these actions lead to changed gene expression.
Interestingly, it has been shown that ETR1, one of the receptors in the ethylene signaling pathway, mediates stomatal closure in response to H 2 O 2 (Desikan et al., 2005). A cysteine at position 65 was essential for this possible interaction between ethylene and H 2 O 2 (Desikan et al., 2005). Moreover, it was shown by D' Haeze et al. (2003) that ROS production may precede ethylene production during Nodulation factorstimulated root hair growth, suggesting another interaction between ROS and ethylene.

Light
Light quantity attenuates strongly in water as a result of its absorptive properties and of the absorptive, reflective and scattering properties of suspended material within the water (Holmes & Klein, 1987). Vervuren et al. (2003) analysed the light transmission of river Rhine floods and showed that inundations with a depth of 50 cm already have a transmission reduction of 90%, whereas deep floods (> 1.5 m) have light transmissions of < 1%.
Reduced intensities of light and submergence with normal light conditions are both able to induce a similar set of traits in Rumex species, such as an increase in petiole length and SLA, and a decline in the underwater compensation point for light (Mommer et al., 2005a). Moreover, both low light and submergence can induce hyponastic growth in A. thaliana Col Pierik et al., 2005). The resemblance in the morphological and biochemical responses to low light and submergence might simply be explained by separate signal-transduction cascades that both affect a common downstream component. Alternatively, it is conceivable that these two pathways operate synergistically or additively when both signals are present at the same time (flooding with a strong attenuation of light). Recently, an additive effect was observed for petiole elongation in R. palustris: submerged plants in low light had significantly longer petioles than plants submerged in high light (Mommer et al., 2005a).
In addition to decreases in light quantity, the spectral composition of light changes when it passes through water. Water specifically absorbs the longer wavelength, like far-red (FR), which results in an increase of the red (R) : FR ratio (reviewed in Holmes & Klein, 1987). In turbid water, caused by, for example, silt or clay, the light is scattered and also the shorter wavelengths are absorbed (ultraviolet and blue light). Therefore, in turbid water, the green wavelengths penetrate the furthest (reviewed in Holmes & Klein, 1987). New Phytologist (2006)  Elongation of plant organs, such as petioles, is stimulated by low intensities of blue light and a low R : FR ratio (Pierik et al., 2004;Pierik et al., 2005). Therefore, because the R : FR ratio increases in submerged environments, we conclude that submergence-induced elongation is not induced by this change in the R : FR ratio. However, a stimulating contribution of low blue light intensities can occur as this light signal is attenuated in water under turbid field conditions.

Summary of possible submergence signals
Based on the information presented above there is no clearcut conclusion to be drawn about how submergence is sensed. There is ample evidence that in R. palustris, ethylene is responsible for very rapid responses, although the exact role of ethylene needs to be further elucidated. There are indications for possible interactions between ethylene and low O 2 , and ethylene signaling and H 2 O 2 , but these have to be studied further. At this point we do not know whether submergenceinduced responses are influenced by direct or indirect sensing of changes in O 2 concentration, or changes in ROS and/or other second messengers, such as NO and Ca 2+ . The effect of light is probably restricted to conditions in which the floodwater has a high turbidity.

III. Internal aeration and underwater photosynthesis
Internal aeration provides O 2 to submerged organs of plants in flooded environments; enabling aerobic respiration even when the adjacent external environment limits O 2 entrance (Jackson & Armstrong, 1999). In addition, a portion of the O 2 transported to roots is lost radially [radial O 2 loss (ROL)], particularly from laterals and tips of the main axes in the case of many wetland species (Armstrong, 1979;Colmer, 2003). ROL re-oxygenates the rhizosphere (Pedersen et al., 1995;Revsbech et al., 1999), thereby protecting roots from the reduced substances (e.g. Fe 2+ , Mn 2+ , S 2-, organic acids) commonly present in flooded soils (Ponnamperuma, 1984).
In emergent wetland plants, atmospheric O 2 enters the shoot and moves, via aerenchyma, to the tips of the roots. Gas transport in shoots is mostly diffusive, but can (in some species) occur through pressure-driven flows that force gases along the lacunae in stems or petioles and rhizomes, at least when environmental conditions are suitable (Dacey, 1980;Armstrong & Armstrong, 1991, reviewed in Colmer, 2003. In contrast to throughflow convection, demonstrated elegantly in water lilies (Dacey, 1980) and reed (Brix, 1988;Armstrong & Armstrong, 1991), no significant aeration contribution can be expected from a second type of convective gas flow, the so-called nonthroughflow convection (Beckett et al., 1988).
When plants are completely submerged, however, direct access to atmospheric O 2 is abolished. The sources of O 2 to completely submerged plants are those produced during photosynthesis and present in the floodwaters (Sand-Jensen et al., 2005). Thus, submerged plants can experience marked diurnal changes in tissue O 2 concentrations, the O 2 concentration being highest during the day and lowest during the night (e.g. for rice; Waters et al., 1989). The consequences of distance within the plant body along the diffusion path from an O 2 source, and the combined resistances (physical and O 2 consumption) to diffusion along the pathway, have been reviewed in detail (Armstrong & Drew, 2002;Colmer & Greenway, 2005). The remainder of this section will focus on environmental factors and plant traits, determining: (i) aerenchyma formation; (ii) rates of photosynthesis by submerged plants; and (iii) O 2 exchange between the shoot and floodwater.

Aerenchyma formation
Aerenchyma is tissue containing large gas-filled spaces (lacunae) that interconnect longitudinally to provide a lowresistance pathway for long-distance gas transport along plant organs. The presence of lacunae provides increased porosity (gas volume per unit tissue volume) of tissues to levels much higher than the inherent values resulting from the usual intercellular gas-filled spaces formed as a constitutive part of development. In roots, aerenchyma usually forms in the cortex, whereas in stems (including rhizomes) aerenchyma can occur in the cortex and in the pith cavity (Armstrong, 1979). The volume of aerenchyma formed depends on genotype (species, as well as cultivar/accession) and environmental conditions (Colmer, 2003).
Two basic forms of aerenchyma have been recognized (Sachs, 1882), namely that which develops by the separation of cells ('schizogenous' aerenchyma) or that which develops by the collapse of cells ('lysigenous' aerenchyma) (Fig. 1), although within both of these main 'types', several different processes can form the resulting aerenchyma ( Justin & Longstreth & Borkhsenious, 2000;Seago et al., 2005). In many wetland plants, aerenchyma forms constitutively, but is further enhanced in response to flooding. In most nonwetland species, aerenchyma is largely absent until it is induced by environmental conditions, such as flooding (Armstrong, 1979;Colmer, 2003), and also, in some species, by soil compaction or nutrient deficiencies (Drew et al., 1989;He et al., 1996). Information on the regulation of formation of schizogenous types of aerenchyma is scant, whereas the induction of lysigenous aerenchyma has been studied in some detail, at least in roots of Z. mays.
The formation of lysigenous aerenchyma involves the removal of files of cortical cells, leaving enlarged gas-filled spaces, although the remnants of cell walls persist (Fig. 1b). Ethylene is the primary signal that initiates programmed cell death in the cortical cells that are removed (Drew et al., 2000). Numerous files of cells die, but some files remain intact and these maintain symplastic connections from the stele to the rhizodermis. The mechanisms by which some radial files respond to the ethylene signal (and die), whilst tangential neighbours do not, is unknown (Evans, 2004). Nevertheless, some aspects of the signaling pathway, details of the ultrastructural events during the programmed cell death, and several biochemical aspects (particularly related to cell-wall degradation), have been described. Drew et al. (2000) proposed a hypothetical working model for ethylene signaling, ultimately resulting in the phosphorylation of (unidentified) downstream proteins, which then presumably regulate the required changes in gene expression to control the programmed cell death in these target cells. Stages of the ultrastructural changes in cells destined to die to form lysigenous aerenchyma are described in detail for the roots of Z. mays (Gunawardena et al. 2001), as are changes in expression of enzymes (cellulases and XET) proposed to have roles in cell-wall degradation during aerenchyma formation (He et al., 1994;Saab & Sachs, 1996). As also indicated in other reviews ( Jackson & Armstrong, 1999;Evans, 2004), further studies are needed to elucidate the molecular basis of aerenchyma formation in plants in greater detail.

Underwater photosynthesis
Photosynthesis in submerged plants can commonly be restricted by low availability of light and/or CO 2 (Sand- Jensen, 1989;Maberly & Spence, 1989). Light reaching the leaves of submerged plants is attenuated by water, dissolved organic matter, particles (e.g. silt) and/or phytoplankton suspended in the water column (Bach et al., 1998;Vervuren et al., 2003). It was recently demonstrated for the river Rhine system in the Netherlands that floods with a water depth of 1 m can have a light transmission below 1% (Vervuren et al., 2003). On the other hand, light intensities are not necessarily always low underwater, as demonstrated in a study where a photosynthetically active radiation (PAR) value of > 600 µmol m −2 s −1 was measured at a flooding depth of 30 cm in an abandoned clay pit in the Netherlands, the typical environment of R. palustris (Mommer et al., 2006). CO 2 concentrations in most water bodies are much lower than the concentrations required to saturate photosynthesis by leaves of vascular plants (under saturating light and well-mixed conditions), and even that by aquatic species (Sand-Jensen, 1989). Diurnal fluctuations in dissolved CO 2 commonly occur in floodwaters, with a build-up of CO 2 during darkness and depletion to concentrations well below those in air-saturated water when light is available for photosynthesis (e.g. to 0.003 mol m −3 in flooded rice fields, Setter et al., 1987). Rates of CO 2 uptake by leaves can limit photosynthesis, owing to the already mentioned 10 4 -times slower diffusion of gases in water than in air, resulting in the movement of CO 2 across boundary layers adjacent to leaves becoming rate-limiting, even in submergence-acclimated leaves of aquatic species (Smith & Walker, 1980;Maberly & Spence, 1983;Raven et al., 1985). Furthermore, for leaves of terrestrial species that become submerged, the cuticle is a major component of the resistance to CO 2 entry, further impeding underwater photosynthesis (Mommer et al., 2005b). In addition, low concentrations of CO 2 , relative to O 2 , increase photorespiration in submerged leaves (Mommer & Visser, 2005;Mommer et al., 2005b).
Physiological and morphological adaptations in leaves of aquatic and amphibious macrophytes have been reviewed in detail (Maberly & Spence, 1983, 1989Raven et al., 1985;Bowes & Salvucci, 1989;Madsen & Sand-Jensen, 1991). CO 2 -concentrating mechanisms, the capacity to utilize HCO 3 − and a leaf morphology to reduce boundary layer thickness (e.g. dissected leaves), were identified as important adaptations to access CO 2 in aquatic environments. Thin cutical layers that are relatively permeable to CO 2 are also important (Frost-Christensen et al., 2003;Mommer et al., 2005b). Other strategies to access CO 2 were noted, such as, in many species, the development of floating leaves or aerial leaves to access atmospheric CO 2 (and O 2 ), or the more specialized strategy by certain plants that exploit CO 2 from the sediment (Raven et al., 1988).

Leaf morphology and anatomy
Diffusion across the cuticle might become the major pathway for CO 2 entry into submerged leaves (Frost-Christensen et al., 2003), even for leaves with stomata (i.e. aerial leaves of amphibious species and leaves of emergent wetland plants); although information on whether the stomata remain open, or close, during submergence appears to be scarce. Most stomata of R. palustris are closed when this plant is completely submerged (Mommer et al., 2005b). Resistance to gas movement across the cuticle can be less in water-grown leaves than in air-grown leaves of several species (Frost-Christensen et al., 2003;Mommer et al., 2005b).
In response to long-term (e.g. more than several days) submergence, some plants can produce new acclimated leaves underwater that have an enhanced gas exchange with the water column (Mommer & Visser, 2005). For example, young leaves of Rumex species elongate upon submergence, and although most of the elongation growth occurs in the petiole, the lamina also elongates and becomes narrower (e.g. for R. palustris see Fig. 2 in Mommer & Visser, 2005). This elongated leaf morphology would be expected to result in smaller boundary layers than present for the broader lamina of nonelongated leaves. Next to becoming narrower, the leaves of submergence-tolerant R. palustris and R. crispus also become thinner (i.e. increased SLA), which can enhance gas exchange between the leaf and the surrounding water (Vervuren et al., 1999;Mommer et al., 2005a). Furthermore, submergenceacclimated leaves of R. palustris had thinner epidermal cell walls, a thinner cuticle and the chloroplasts were orientated towards the epidermis, all of which would decrease diffusion resistance between the leaf surface and chloroplasts (Mommer et al., 2005b). Thus, changes in morphology and anatomy in submergence-acclimated leaves enhance CO 2 entry.
The enhanced capacity for underwater gas exchange by new leaves of R. palustris produced following submergence was evidenced by higher internal O 2 concentrations in petioles of these leaves during periods in light, when compared with O 2 concentrations in plants grown in air prior to submergence (Mommer et al., 2004). Higher internal O 2 presumably resulted from improved rates of CO 2 entry from the water column and thus enhanced rates of photosynthesis, and this was evident across a range of dissolved CO 2 concentrations (up to ≈ 1 mol m −3 ) in a turbulent water column ( Fig. 2b; Mommer et al., 2004;2005a). Improved uptake of dissolved gases by the acclimated leaves was also demonstrated by manipulations of water column O 2 concentrations in the dark, whilst monitoring O 2 concentrations in the petiole. In darkness, acclimated leaves displayed higher internal O 2 concentrations at any given external O 2 concentration, indicating a greater entry of O 2 ( Fig. 2a; Mommer et al., 2004).
In summary, leaves of submergence-tolerant plant species (e.g. R. palustris) can acclimate to an underwater environment, resulting in enhanced CO 2 uptake for photosynthesis and therefore providing a supply of sugars to the plant. In addition, O 2 status is improved not only during the daytime (O 2 derived from photosynthesis), but also during the night (enhanced entry of O 2 from the water column). An improved O 2 status in the shoots would benefit the roots, as internal O 2 diffusion via the aerenchyma to the roots would increase, not only enabling respiration within the roots, but also oxygenation of the rhizosphere. Despite these recent insights, much remains to be elucidated regarding the physiological and molecular regulation of the acclimations. As suggested by Mommer & Visser (2005), the two hormones ethylene and abscisic acid (ABA) would, however, be good starting points as these are known to be involved in the switch between aquatic and terrestrial leaf type formation (heterophylly) in aquatic plants (Lin & Yang, 1999;Kuwabara et al., 2003).

IV. Stimulated shoot elongation
Several plant species from wetland environments demonstrate dramatic changes in shoot anatomy, morphology and growth pattern when submerged partially or completely (Ridge, 1987). Spectacular submergence responses are the altered orientation of petioles from prostrate to vertical (hyponastic growth) and the strongly stimulated petiole and internode elongation (Fig. 3). Not all petioles and internodes within one plant are equally responsive, as demonstrated in a study on R. palustris . In this study, both the very young and the oldest petioles possess only a very minimal competence to elongate, whereas intermediately aged petioles are characterized by an elongation response of up to 300%, measured over 10 d of submergence. Differentiation in responsiveness was also described for internode elongation in deepwater rice where only the youngest internodes respond to submergence (Sauter, 2000).
Shoot elongation underwater requires energy and carbohydrates for cell divisions and the synthesis of new cell-wall material. Evidence for these costs of elongation growth came from work on rice in which elongation underwater without reaching the water surface occurs at the expense of survival (Setter & Laureles, 1996). Lowland rice cultivars demonstrated a negative correlation between survival and elongation growth. Moreover, enhancing shoot-elongation capacity by the addition of the growth-promoting hormone, gibberellin (GA), decreased survival, whereas an inhibitor (paclobutrazol) of GA biosynthesis, and therefore shoot elongation, enhanced survival. It is suggested that the nonelongating strategy is more suitable for short-lasting and/or deep-flooding events that cannot be outgrown, whereas fast underwater elongation increases fitness mainly under prolonged, but relatively shallow, floods (Setter & Laureles, 1996;Sauter, 2000). This hypothesis was recently confirmed in a study that combined field surveys in floodplains of the river Rhine (the Netherlands) with experiments on the shoot-elongation capacity of 22 selected plant species. Fast-elongating species were exclusively found in flood-prone environments characterized by slow drainage and shallow temporary pools. Under these environmental conditions, the costs of elongation are probably outweighed by benefits such as improved aeration and restored rates of photosynthesis when the leaves emerge. Nonelongating species were only found on rarely flooded sites and in environments with frequent, but short-lasting, floods .

Hyponastic growth interacts with shoot elongation
One of the first phenotypic modifications that takes place upon submergence in some plant species is an increase in the angle, compared with the horizontal, of mainly the younger petioles, thus decreasing the distance between the tips of the leaf blade and the water surface (Banga et al., 1997). This flooding-induced response has been described, to date, for Ranunculus repens, Caltha palustris, Leontodon taraxacoides, Paspalum dilatatum, R. palustris and A. thaliana (Ridge, 1987;Voesenek & Blom, 1989;Grimoldi et al., 1999;Insausti et al., 2001;Millenaar et al., 2005). Interestingly, hyponastic growth responses are also described in response to other environmental cues, such as shading and high temperature (Ballaré, 1999;Pierik et al., 2004;Millenaar et al., 2005). In R. palustris, hyponastic growth is caused by differential cell elongation across the petiole base, with epidermal cells on the abaxial surface elongating faster than cells on the adaxial surface (Cox et al., 2004). Furthermore, it was discovered that, in this species, hyponastic growth interacts with nondifferential petiole elongation. A prerequisite for flooding-induced petiole elongation is that a minimal petiole angle of 40-50° (from horizontal) has to be reached. When the initial petiole angle is lower than this threshold, no petiole elongation will take place until hyponastic growth has increased the petiole angle to 40-50° (Cox et al., 2003).

Regulation of submergence-induced shoot elongation: hormone interactions and cell-wall events
Accumulated ethylene is probably the primary signal that triggers the plant to start a cascade of reactions ultimately leading to enhanced cell elongation in submerged shoot organs. A direct effect of this increased ethylene concentration is a fast decrease in the ratio between two other plant hormones: ABA and GA. This is related to a strong and fast reduction of the endogenous ABA concentration (Hoffmann-Benning & Kende, 1992;Benschop et al., 2005) and an increase of the endogenous GA concentration (Rijnders et al., 1997;Raskin & Kende, 1984b). The decrease of ABA Fig. 3 Upon submergence, Rumex palustris quickly re-orientates its leaves into a more vertical position (hyponastic growth) followed by strong petiole elongation towards the water surface. Note that petiole elongation without the preceding hyponastic growth would not bring the leaf tips much closer to the water surface.
in petioles of R. palustris during submergence and treatment with ethylene is regulated via a fast down-regulation of several members of the 9-cis-epoxycarotenoid dioxygenase (NCED) gene family, a group of genes that limit ABA biosynthesis in higher plants (Benschop et al., 2005). Furthermore, an increase (of 10 -25%) of a product of ABA catabolism, phaseic acid, was observed in the time frame in which ABA decreased by 70% (Benschop et al., 2005), suggesting increased ABA breakdown. Interestingly, application of the ABA biosynthesis inhibitor, fluridone, reduced the lag phase of submergenceinduced petiole elongation in R. palustris by 50%, resulting in a very rapid elongation response. However, when this fluridone treatment was combined with a 1-MCP pretreatment (inhibitor of ethylene action), elongation growth underwater was completely abolished. This result indicates that a reduction of ABA by itself is insufficient to induce fast petiole elongation; the presence of elevated concentrations of ethylene is required (Benschop et al., 2005).
It is clear that in both Rumex and rice, ABA acts as a negative regulator of underwater elongation. This raises the question as to how this is regulated at a physiological level. Recent data indicate that ABA directly interferes with GA biosynthesis. When submerged R. palustris plants were additionally treated with 10 µM ABA, no increase was observed in the endogenous GA1 concentration; this result was in contrast to submerged controls. Therefore, ABA probably inhibits the up-regulation of the growth-promoting hormone GA (Benschop et al., 2005). In this respect, it is interesting to note that GA-induced growth stimulation in roots and elongation of shoots in Arabidopsis seedlings (Fu & Harberd, 2003;Vriezen et al., 2004) is associated with the destruction of DELLA proteins in the 26S proteasome. These DELLA proteins, localized in the nucleus of plant cells, function as repressors of growth (Alvey & Harberd, 2004). In rice, GA binds to a soluble GA receptor (GID1) present in the nucleus. Upon GA binding, this receptor molecule interacts with DELLA proteins, resulting in the degradation of these growth-repressing DELLA proteins (Ueguchi-Tanaka et al., 2005). It is highly possible that DELLA degradation is a prerequisite for ethylene-induced shoot elongation in submerged plants.
Ultimately, the hormonal changes described above result in cell elongation within the elongating organ (e.g. petioles, internodes). This requires loosening of cell walls, uptake of water and synthesis of cell-wall polysaccharides (Cosgrove, 1999). Elongation of internodal cells in deepwater rice is primarily related to increased cell-wall extensibility, and not to decreases in osmotic potential or increases in hydraulic conductance (Kutschera & Kende, 1988). Cell-wall extensibility is thought to be associated with cell-wall loosening proteins, such as those belonging to the family of expansins (McQueen-Mason et al., 1992). Expansin activity has a pH optimum between 3.5 and 4.5 (McQueen-Mason et al., 1992) and is hypothesized to break H-bonds between hemicelluloses and cellulose microfibrils (McQueen-Mason & Cosgrove, 1994).
The expression of expansin genes and the protein activity correlate strongly with submergence-induced elongation in several plant species (Cho & Kende, 1997;Huang et al., 2000;Kim et al., 2000;Vreeburg et al., 2005). Increased expression of expansins in deepwater rice was not only induced by submergence treatments, but also by GA application, indicating that expansin genes are a downstream target of GA action (Cho & Kende, 1997). However, in R. palustris, GA application cannot mimic the induction of RpEXPA1 by submergence. This gene was the only expansin, out of a group of 13 R. palustris expansin genes, that was up-regulated in petioles upon submergence (Vreeburg et al., 2005). The increased expression of RpEXPA1 could be mimicked by ethylene treatment and repressed during submergence when plants were pretreated with the specific ethylene action inhibitor, 1-MCP (Vreeburg et al., 2005). These results suggest that in Rumex, ethylene is at the base of at least two parallel operating signal transduction pathways: one leading to a decrease of the ABA: GA ratio and the other resulting in increased expression of the expansin gene and subseqent concentration of expansin protein (Vreeburg et al., 2005). Ethylene also seems to be the triggering agent in expansin gene expression in the semiaquatic fern Regnellidium diphyllum, a species characterized by fast underwater elongation of the rachis (Kim et al., 2000).
Moreover, ethylene plays a role in a process leading to cell-wall acidification in submerged shoots. The importance of cell-wall pH for petioles of R. palustris was recently illustrated in experiments in which buffers with a low (4.0) and more neutral (6.0) pH were injected into petioles (Vreeburg et al., 2005). The low pH strongly stimulated elongation growth, whereas a temporary inhibition of underwater growth was observed when the neutral buffer was injected. An increased net efflux of H + was measured within minutes after starting a submergence treatment; a response that could be abolished completely when the petiole tissue was pretreated with 1-MCP (Vreeburg et al., 2005).

V. Perspectives
The suite of submergence-induced avoidance responses described in this review all are functionally associated with preventing the adverse consequences of the very slow gas diffusion in water. The regulatory mechanisms leading to these responses upon detection of submergence are, however, with the exception of shoot elongation, not well understood. There is, for example, much to be learned about the regulatory mechanisms behind aerenchyma formation and the reduced gas diffusion resistance in leaves formed underwater. Ethylene is not only the key factor in enhanced shoot elongation, but is also a major player in the formation of aquatic leaves in aquatic plants and in lysigenous aerenchyma formation, at least in Z. mays. Many physiological and molecular processes interacting with ethylene during submergence have become known from work on stimulated shoot elongation. This information could serve as a starting point for unravelling signaling cascades involved in the submergence-induced regulation of aerenchyma formation and reduced gas-diffusion resistance in leaves.
Although there is extensive information on signal transduction in submergence-induced hyponasty and shoot elongation, considerable progress is to be expected from the recent introduction of the model plant Arabidopsis in submergence research . The genetic variation in Arabidopsis for hyponasty , ethylene sensitivity (Peeters et al., 2002), and tolerance to full submergence (D. Tholen & L.A.C.J. Voesenek, unpublished) and soil flooding (Kolodynska & Pigliucci, 2003), is currently being exploited in quantitative trait loci (QTL) analyses. These analyses should enable the identification of genomic loci, and ultimately the underlying genes (Maloof et al., 2001;El Din El Assal et al., 2001) involved in hyponastic growth B. Snoek & A. J. M. Peeters, unpublished). The kinetics of hyponastic growth in Arabidopsis, upon submergence and ethylene exposure, are remarkably similar to those of R. palustris (compare Cox et al., 2003 Fig. 4), making it feasible to translate insights obtained from Arabidopsis to flood-tolerant species, such as R. palustris. Recent microarray studies on Arabidopsis have identified a number of candidate genes involved in hyponastic growth F. F. Millenaar & A. J. M. Peeters, unpublished), and an ongoing mutant screen in our laboratory has already yielded several Arabidopsis mutants that fail to show hyponastic growth (M. van Zanten & F. F. Millenaar, unpublished).
Such molecular and genetic research strategies could also identify the regulatory networks that enhance internal aeration during submergence through changes in aerenchyma formation and underwater gas-exchange properties of leaves. These traits may not be well developed in Arabidopsis but can probably be investigated in rice, for which extensive genomic information is available and which shows a high degree of genetic variation for submergence tolerance ( Jackson & Ram, 2003). Such studies would provide detailed knowledge on the physiological and molecular regulation of the plant responses to submergence, as highlighted in this review, adding to present knowledge on submergence tolerance in this important crop species ( Jackson & Ram, 2003;Toojinda et al., 2003). Such mechanistic knowledge is required for direct experimental tests of the costs and benefits of these responses (e.g. Voesenek et al., 2004). This will ultimately elucidate how the different traits are inter-related and how the interactions between these traits determine submergence tolerance.