Diurnal dynamics of oxygen and carbon dioxide concentrations in shoots and rhizomes of a perennial in a constructed wetland indicate down-regulation of below ground oxygen consumption

Plants have evolved mechanisms to provide oxygen to their parts in oxygen-free environments like wetland sediments. We measured the diurnal courses of oxygen supply to rhizomes of the common reed, a widespread wetland plant. During the day the below-ground plant parts can rely on ample oxygen, but during the night its supply to rhizomes and roots as well as to the whole assembly of associated microorganisms is limited. The key finding of the study was that during periods of low oxygen supply the whole below-ground biota reduces its respiration. This regulation mechanism helps the biota survive unfavourable periods.


Introduction
Under strongly anoxic conditions in permanently watersaturated soils of wetland ecosystems (Ponnamperuma 1972), oxygen is a very rare resource with limited occurrence and, if present, it is immediately utilized by biological and chemical processes. Plants growing under such conditions frequently develop an enlarged gas-space continuum which runs from shoots through the extensive rhizome system to the roots (Cardoso et al. 2014;Jackson and Armstrong 1999;Kon calov a 1990). This aerenchyma system enables emergent wetland plants to supply submerged organs with oxygen. In return respiratory gases (e.g. CO 2 ) from below ground aerobic, anaerobic and fermentation processes (e.g. aerobic respiration, denitrification, iron reduction) as well as methane from the anaerobic sediments can ventilate to the atmosphere (Armstrong 1980;Armstrong et al. 2000;Kayranli et al. 2010;Le Mer and Roger 2001;Laanbroek 2009;Maltais-Landry et al. 2009).
Wetland plants enhance their internal ventilation capacity by pressurized gas-flow through the aerenchymatic tissue (Armstrong and Armstrong 1991;Brix et al. 1992). Pressurization is achieved by the generation of gradients of temperature (thermal-transpiration) and/or humidity (humidity-induced convection) over a porous partition. The following free flow of gas inside the aerenchymatic tissue is realized under Knudsen regime between the inner and outer atmosphere (Armstrong et al. 1996). Through stems and rhizomes, gas-flow in Phragmites australis can also occur as a result of wind blowing over the stand by which air pressure gradients (Venturi effects) are created between young and old or dead culms (Armstrong et al. 1992). In P. australis gas through-flow occurs mainly via the pith cavity, as the cortex cylinder has very low porosity (Afreen et al. 2007).
The described ventilation mechanisms can occur simultaneously, but prevailing pressurization types vary between species. Pressurization in emergent wetland macrophytes with graminoid growth type is mainly driven by humidity-induced convection (e.g. Eleocharis spacelata, Typha sp., Juncus ingens, P. australis). In parallel, pressurization efficiency is affected by other factors including light intensity, shoot height, leaf sheath area, photosynthesis and stomatal aperture (Afreen et al. 2007;Armstrong and Armstrong 1991;Bendix et al. 1994;Brix et al. 1992Brix et al. , 1996Tornbjerg et al. 1994). The combined effects of pressurization processes can result in internal overpressure between 4 and 11 hPa (Afreen et al. 2007;Armstrong and Armstrong 1990), which sustains the internal oxygen concentration ([O 2 ]) in submerged organs close to ambient levels (Brix et al. 1996;Konnerup et al. 2011).
Most roots of wetland plants develop barriers to prevent radial loss of oxygen to the soil (ROL) by impeding radial diffusion in the epidermal and hypodermal cell layers (Armstrong et al. 2000;Soukup et al. 2002). Nonetheless, oxygen is released into the rhizosphere through porous regions at the root tips, lateral roots and through gas permeable 'windows' opposite of developing laterals (Armstrong et al. 2000). Thus, partly oxic conditions occur in the rhizosphere (Mainiero and Kazda 2005), which allow oxidation of the phytotoxic compounds, which characterize anaerobic soils. Although aerobic soil microbes benefit from locally oxic conditions in the rhizosphere (Faußer et al. 2012), plants can utilize products of the mineralization processes by the rhizobacterial communities (Radruppa et al. 2008). In such interactions, oxygen supplied below ground is readily consumed and soil borne respiratory gases may enter the internal gas-spaces (Colmer 2003). Consequently, diurnal courses of oxygen and carbon dioxide can be observed in internal plant cavities in both above and below ground plant parts (Brix et al. 1996;Konnerup et al. 2011). It was reported that the internal partial pressure of CO 2 in Typha latifolia leaves is up to 10 times elevated compared with ambient air (Constable et al. 1992). Laboratory experiments on rice plants showed CO 2 concentrations ([CO 2 ]) of above 9 % when growing in CO 2 enriched media (Higuchi 1982).
During periods of limited air ventilation to the submerged organs, e.g. during the night when pressurization mechanisms are unfavourable, the level of internal [O 2 ] may fall in the range of hypoxic conditions (Armstrong et al. 2009;Gupta et al. 2009). It was reported that segments of plant roots are able to down-regulate their cellular rate of respiration if the internal [O 2 ] drop to <3.1 % (i.e. 45 hPa) (Armstrong et al. 2000(Armstrong et al. , 2009Berry and Norris 1949; C ı zkov a and Bauer 1998; Gupta et al. 2009;Zabalza et al. 2009). Armstrong and Beckett (2011b) proposed a model for the mechanism of oxygen conservation in respiring root segments when internal oxygen levels are limited under the hypoxic conditions. In a similar way, soil microbes in the rhizosphere will regulate their cellular rate of aerobic respiration during times of limited oxygen supply (Jin and Bethke 2003).
In this study, diurnal dynamics of [O 2 ] inside rhizomes and culms were connected to plant-internal [CO 2 ]. Please note that a pre-version of this article is included in the doctoral thesis of the first author. At the time of writing, the thesis is not yet published and not citable. Such measurements under field conditions can reveal the whole-system interactions in plant-internal gas transport, gas exchange and oxygen conservation by regulation of its consumption. Based on this, the following three hypotheses were tested: i. Plant-internal concentrations of O 2 and CO 2 are negatively correlated. Despite the aeration in wetland plants is well documented (Afreen et al. 2007;Bendix et al. 1994;Brix et al. 1996;Colmer 2003;Konnerup et al. 2011;Sorrell and Brix 2003) there are only few data on plant-internal [CO 2 ] (c.f. Constable et al. 1992 (Armstrong and Beckett 2011b) and shall here be tested in situ for the whole below ground plant system. The results allow us to interpret the degree of oxygen consumption and conservation by the whole submerged plant associated system and thus to contribute to the recent 'respiratory downregulation debate' (Armstrong and Beckett 2011a).
The results of the presented field study are of added value to previous experiments (Afreen et al. 2007;Armstrong and Armstrong 1991;Armstrong and Armstrong 2005), as they were gathered under field conditions and not in the laboratory, in pots or on excised plants. Perennials in an established constructed wetland (CW) were chosen for the study, as the environmental parameters (e.g. water level, flow rate, water quality) of this system could be well monitored. Furthermore, the growing conditions for the plants are close to those in natural wetlands, while the homogeneous substrate made of gravel is the only noteworthy difference (Du sek et al., 2008;Picek et al. 2007). Studies under undisturbed growing conditions in a wetland ecosystem provide knowledge on interactions between plants and microorganisms (Bodelier 2003) in the rhizosphere competing for resources like nutrients and oxygen. In order to understand the whole ecosystem functioning of wetlands, it is crucial to analyze system interactions of the oxygen cycle under undisturbed conditions.

Site description
This study was conducted in a subsurface horizontal flow CW located in Slavo sovice in South Bohemia, Czech Republic in August 2009. This CW was chosen for our study as its development and growing conditions are well documented by previous studies (Du sek et al., 2008;Picek et al. 2007) which guaranties better interpretation of the findings. Situated at 480 m above sea level, the annual average air temperature was 7.9 C and the annual precipitation was 634 mm. Table 1 gives meteorological data at the study site for month August. The facility consisted of a pretreatment (screen, sand trap, sedimentation tank) and two vegetated treatment beds each of 17 m length and 22 m width (total treatment area 374 m 2 ) planted with common reed (P. australis). The CW was created for municipal waste water treatment of 150 person equivalents and started operation in August 2001. The vegetated part of the treatment beds was filled with fine gravel (3-20 mm) while the margins at inlet and outlet (1.5 m each) consisted of coarse gravel (50-100 mm). Water and substrate properties are characterized in detail in earlier studies (Du sek et al. 2008;Picek et al. 2007). Analysis of the final effluent showed a high efficiency of the system in removing organic pollution (reduction of biological oxygen demand, BOD 5 , by 82 % and chemical oxygen demand, COD, by 74 %), total nitrogen (63 %), total phosphorus (75 %) and suspended solids (52 %) (Du sek et al. 2008). On average, the waste water inflow rate was regulated to 0.12 6 0.10 Ls À1 (mean 6 SD), with a maximum of 1.0 Ls À1 during periods of extremely strong precipitation. During the vegetation season, the water level was controlled to 20-30 mm under the gravel surface.
Until this study in August 2009, the common reed successfully established a dense stand covering all available CW area. The stand was estimated as healthy according to the shoot density, average shoot height and appearance (cf. Dickopp et al. 2011). Above and below ground biomass of the common reed stand was evaluated earlier (Picek et al. 2007).
During the period of measurements we recorded microclimatic factors inside the stand in a height of 1 m above the gravel surface. These parameters included photon flux density (PFD) over the waveband 400-700 nm, air temperature and relative air humidity (RH). PFD sensors were produced by LI-COR Biosciences, Lincoln NE, USA; temperature and RH sensors by Delta-T, Burwell, UK. Additionally, substrate water temperature of the reed bed was logged. PFD outside the stand was gathered in a height of 1.5 m. Stable weather conditions prevailed during diurnal oxygen recording (cf. Fig. 1A). In all data, time of day is presented as Central European Time (CET) without summer time adjustment.

Plant selection
Measurements in the internal gas-spaces of reed plants were performed in two regions of the southern For intensive assessment of [O 2 ] in culms and directly adjacent rhizomes, two additional plants (A1, A2) were assigned in the second half of the CW vegetation bed. In this region high amplitudes in internal oxygen were detected in rhizomes (Dickopp et al. 2011). The rhizome systems of the reed plants were gently excavated from leaf litter and gravel (each layer of about 50 mm in thickness). Two dead ending vertical rhizomes, with short stumps of shoots from earlier years (1-2-years old), were chosen in order to connect to directly adjacent this year's culms. On those two plants also gas exchange measurements of top leaf blades and analysis of internal [CO 2 ] were performed.

Oxygen measurement
For the measurement of [O 2 ] in the reed culms (A1, A2, In1, In2, Out1, Out2), optical oxygen sensors (diameter 4 mm, type PSt3, PreSens GmbH, Regensburg, Germany; Klimant et al., 1995) were implanted into internodes (300 mm above ground) through drilled openings. On the plants A1 and A2, the ends of the corresponding rhizomes were cut open 25-55 mm above the water table (i.e. within the gravel layer) and equipped with the same type of oxygen sensor as in the culms. The generated openings in shoots and rhizomes were sealed with several layers of laboratory film (Parafilm, American National Can, Chicago, USA) subsequently after sensor implantation. Oxygen concentrations in In1, In2, Out1 and Out2 were recorded by FIBOX 3 (PreSens GmbH, Regensburg, Germany) every hour on a daily course. The use of a four channel gauge (Oxy-4 mini, PreSens, Regensburg, Germany) connected to the culms and rhizomes of A1 and A2, enabled us to monitor changes in [O 2 ] continuously every five minutes for three complete diurnal courses (24-27 August 2009). During measurements the excavated rhizome systems were covered with aluminium foil to protect them from any direct sunlight, which might penetrate the reed stand. Data recorded from the sensors were recalculated to values of [O 2 ] according to the temperatures of air (for values in culms) or in water (for values in rhizomes), which were assessed close to the implanted optodes. The readings of [O 2 ] were combined with the below described [CO 2 ] observations to test their negative relationship stated in hypothesis 1.

Carbon dioxide measurement
Concentrations of CO 2 were measured with a mobile gas analyzer based on laser absorption technology (Los Gatos Research, Inc., USA, Model: 908-0007). The air samples (10 mL) from inside the reed culms were taken by a gas-tight syringe from the internodes above the oxygen sensor insertion position. The syringe needles were mechanically fixed and sealed with laboratory film to reduce possible leakage of air out of the reed culms. Any manipulation was limited to the sampling procedure only. The needles were closed by empty syringes or by rubber caps during each interval between two sampling times. Immediately after sampling, the air samples were analyzed on the laser gas analyzer. The [CO 2 ] measurements were performed on average every 2 h during daylight.

Photosynthetic gas exchange measurement
In order to observe the activity of the investigated reed plants A1 and A2, the rate of photosynthetic assimilation (A) and stomatal conductance (g s ) were assessed. Correlations between the plant-internal [O 2 ] and the rate of photosynthesis and microclimatic parameters were computed to test the influence of plant activity on ventilation mechanisms (hypothesis 2). Gas exchange measurements were performed every half an hour on leaf blades (second from the top inflorescence) using a LI- where 'n' is the number of instances and 'r' is the correlation coefficient. The P-values were retrieved from a t-distribution, using the df and t-values. In the result tables, the correlation coefficients r, the degrees of freedom df and the P-values are reported. The correlations were calculated for each plant individually and for the whole data set consisting of six plants. We considered a correlation to be significant when the P-value was below 0.05.
Mean values were calculated for minimum and maximum values of [O 2 ] inside the six investigated culms during the measurements. The results are presented with SD and number of included culms (n). A mean value, together with SD and the number of instances (n), is given also for the extent of the sudden oxygen drops recorded during oxygen decline during the night (see below).

Evaluation and modelling of diurnal oxygen courses
In order to evaluate the degree of oxygen conservation mechanisms of the below ground system, the temporal courses of [O 2 ] at nocturnal declines and after cutting of the culms were fitted using the exponential equation: where t is time in seconds and a; b; c 2 R are the parameters to be fitted. Curve fitting was performed by minimizing the least squares with Genetic Algorithms (R, package gafit, www.R-project.org). Intercept 'c' in this formula indicates the minimum levels to which internal [O 2 ] possibly converged when oxygen consumption is regulated. Coefficients of determination (r 2 ) were calculated to give the degree by which the exponential curves represent the declining [O 2 ] in the reed pith cavities. The gradients of the plots given by Equation (1) are expressed by its derivative: rhizosphere-related oxygen demand or loss when main convective ventilation mechanisms are missing overnight or after culm cut-off.

Diurnal oxygen dynamics
The internal oxygen concentration, measured in-situ in culms and rhizomes, showed distinct diurnal patterns (plants A1 and A2 in Fig. 1). On a daily basis, culm-[O 2 ] was correlated strongly with RH (r À0.74; all p < 0.001, see Table 2) and air temperature (r ! 0.76), and moderately with PFD (r ! 0.56). Courses of rhizome-[O 2 ] in the reed plants were similar to those observed in culms and were correlated strongly to RH (r < À0.94) and air temperature (r ! 0.93), and moderate with surface water temperature (r ! 0.50) and PFD (r ! 0.54).
In the morning hours, minimum [O 2 ] ranged in culms from 9.4 to 16.1 % and in rhizomes from 6.7 to 9.0 %. Following sunrise, between 07:25 and 07:40 h, culm-[O 2 ] increased steeply (Fig. 1, Table 3). This pattern was recurrent on each morning at PFD between 200 and 900 mmol m À2 s À1 . At the same time, RH was mainly high but started to decline and the air temperature had risen above 15 C. ] maintained at high levels between 18.8 and 19.9 % for up to 6 h, thus oxygen concentrations were elevated in the rhizomes compared with the culms for some periods (Fig. 1).
In the late afternoon, first rhizome-[O 2 ] (17:00 h) and then culm-[O 2 ] (18:00 h) began to decline rapidly. Meanwhile PFD dropped below 200 mmol m À2 s À1 and RH increased while air temperature was still above 20 C. The [O 2 ] decline was more pronounced in the rhizomes than in the culms, and more distinct in culm A1 compared with culm A2 (Fig. 1)

Plant activity: Gas-exchange of leaf blades
Both Phragmites plants A1 and A2 had positive rates of photosynthetic assimilation (A) in the leaf blades when PFD exceeded 100 mmol m À2 s À1 (Fig. 1). The stomatal conductance (g s ) was constantly high during the measurement. The rise of A rates above 10 mmol(CO 2 )m À2 s À1 coincided mainly with the time of culm-[O 2 ] increase after sunrise (typically 07:30 h), which shows that the start in plant-internal ventilation begins with the photosynthetic activity of the plant in the morning. The rate of A correlates at least moderately strong during the whole period in shoots (r ¼ 0.50, P < 0.001) and rhizomes (r ¼ 0.47, P < 0.001). The daily correlations showed high variations (see Table 2). The observed correlations between [O 2 ] and photosynthesis are in accordance with the moderate correlation to PFD, as photosynthesis is directly linked with the availability of sunlight. During more cloudy and windy conditions on 26 August 2009, rates of photosynthesis above 10 mmol(CO 2 )m À2 s À1 were reached later (10:00 h) and for shorter periods. Rates of A above 15 mmol(CO 2 )m À2 s À1 were recorded when PFD exceeded 1600 mmol m À2 s À1 . After 16:00 h, the photosynthetic rate declined and were below 10 mmol(CO 2 )m À2 s À1 after 18:00 h.

Modelling nocturnal oxygen courses
The coefficients of determination (r 2 ) for the exponential Equation (1) modelling the nocturnal courses of [O 2 ] in culms and rhizomes were above 0.9 (Fig. 3) ] varied between À1.1 and À2.8 % h À1 in the culms and between À3.3 and À4.9 % h À1 in the rhizomes. Thus, the initial decline was twice as steep as in the rhizomes as in the corresponding culms. Rates of oxygen decrease were apparently reduced in the morning hours (05:00 h) to values between À0.02 and À0.12 % h À1 in culm A1, and between À0.01 to À0.02 % h À1 in culm A2. Rhizome-[O 2 ] decrease rates varied in the early morning between À0.04 and À0.09 % h À1 in rhizome A1 and between À0.02 and À0.11 % h À1 in rhizome A2 and were thus in the same range as in the culms. In summary, the twice as high oxygen consumption in rhizomes compared with the culms in the evening levelled down in rhizomes in early morning to the same rates as in the aerial culms. This indicates that oxygen consumption is down-regulated in the submerged plant organs and the rhizosphere at declining oxygen supply.

Experimental limitation of oxygen supply
The analyzed culms A1 and A2 were cut at the internodes above the oxygen sensor insertion on 27 August 2009 at 12:15 h and the stumps were given an airtight seal. The [O 2 ] in both culms and rhizomes showed a sudden and steep decrease after cutting (Fig. 4). Clearly, internal [O 2 ] decreased despite high PFD and low RH. The initiated effect was similar to, but faster than, the evening decreases observed previously. Initial slopes were À8.4 % h À1 in culm A1 and À1.5 % h À1 in culm A2, and from À6.3 to À8.4 % h À1 in the respective rhizomes. In plant A2, [O 2 ] dropped to values in the range of overnight levels 18 h after the culm was cut (culm stump: $16.7 %, rhizome: $8.4 %). About 12 h after the culm was cut, [O 2 ] levels in culm stump and rhizome of plant A1 were also similar to overnight decline. However, during the following 6 h, the internal [O 2 ] decreased further to about 7.3 and 5.2 % in culm stump and rhizome, respectively. Thus, the previously observed mechanisms for downregulation of oxygen consumption were also initiated when oxygen supply was experimentally interrupted but these mechanisms were not sufficient beyond 12 h without new, actively ventilated air.

Oxygen and carbon dioxide dynamics in culms and rhizomes
Dynamics of oxygen concentration inside wetland plant gas-spaces have been previously studied mainly in laboratory or greenhouse experiments or on excavated plants and culms (Armstrong and Armstrong 1990;Armstrong et al. 1992). Studies on intact plant shootrhizome systems in the field are rare (Brix et al. 1996;Konnerup et al. 2011). The diurnal courses of [O 2 ] recorded in this study in culms and rhizomes confirm the strong influence of RH on internal oxygen concentration under undisturbed growing conditions, especially in rhizomes (Table 2). This aligns with previous works demonstrating that internal ventilation is mainly driven by  Armstrong and Armstrong 1991;Armstrong et al. 1992;Bendix et al. 1994;Brix et al. 1992Brix et al. , 1996Dickopp et al. 2011;Tornbjerg et al. 1994). The effect of air temperature can be explained by increasing air temperature resulting in decreasing RH and higher leaf temperatures which elevated the internal water vapour pressures (Armstrong et al. 1992). Thus thermal transpiration supports pressurization efficiency of wetland macrophytes during hours of high PFD (Brix et al. 1996;Konnerup et al. 2011). The culm-[O 2 ] obviously increased earlier than rhizome-[O 2 ], at a similar morning time irrespective of RH or PFD (see Table 3). This shows that the pressurization in the culms started already before low RH was reached and demonstrates the active role of the shoot tissues of P. australis for pressurized gas-flow (Afreen et al. 2009;Armstrong and Armstrong 2005). The steep slopes of [O 2 ] resulted from the widening of stomatal aperture in the early morning, and in the early evening hours from the narrowing of the stomata (cf. Armstrong et al. 1992;Brix et al. 1996; see also Figure 1 for stomatal conductance). The close correlations between [O 2 ] and gas exchange observations confirm that the stable diurnal dynamics of [O 2 ] in shoots and rhizomes were highly influenced by the photosynthetic activity of the plant (hypothesis 2). A close relationship between photosynthesis and ventilation was also found for alder (Armstrong and Armstrong 2005). Especially in the morning hours, the start of the gas exchange initiated the internal pressurized gas ventilation.
The difference in levels of [O 2 ] between rhizomes and culms during the middle of the day indicates that oxygen-rich air is actively pressurized from culm to rhizome gas-spaces during that period (Brix et al. 1996). In the evening rhizome-[O 2 ] decreased as soon as RH increased and PFD declined below 100 mmol m À2 s À1 while culm-[O 2 ] remained high for another hour (Fig. 1). Throughout the measurements, there was a significant lag between the oxygen courses in culms and rhizomes. The above-mentioned observations reveal the importance of pressurized gas-flow for rhizome aeration initiated in the culm.
During the night, micro-climatic parameters were unfavourable for pressurization mechanisms and, as a result, ventilation was dominated by diffusion or, if present, by Venturi effects (Afreen et al. 2007;Armstrong et al. 1992). As pressurization decreased in the evenings, internal [O 2 ] declined rapidly. Aerobic respiration of submerged plant organs ( C ı zkov a and Bauer 1998), radial oxygen loss (Armstrong et al. 2000) and biological and  (1) The coefficients of determination were in all fits above r 2 > 0.9 with P < 0.001, except in (E) culm A2 from 25 to 26 August 2009 (r 2 ¼ 0.76, P < 0.001).
geochemical oxygen demand in the rhizosphere (Du sek et al. 2008) deplete oxygen reserves in the submerged organs. Brix et al.(1996) proposed that 5-20 % of the oxygen vented through below ground organs of P. australis is consumed by oxygen-dependent processes like respiration. The gradual reduction in rhizome-[O 2 ] by as much as 50 % of the maximum values indicates very high oxygen demand in the rhizosphere of Phragmites plants at this study site.
The [O 2 ]-drops observed in the reed culms during the night can be interpreted as sudden, strong effects of convective gas-flow in the culm-rhizome-system. The oxygen optode placed at the end of the rhizome concomitantly registered a slight [O 2 ] increase. The [O 2 ]-drops might be also related to stomatal widening (Armstrong et al. 1992) and narrowing as they occurred in the early morning and evening hours (cf. Fig. 1). Another explanation may be that [O 2 ]-drops were the readings of local effects of a Venturi driven through-flow event (Armstrong et al. 1992). These sudden changes in internal gas concentrations indicate that there are other gas exchange patterns besides of diffusion and convective gas-flow.
High diurnal variation in [O 2 ] was accompanied by changes in [CO 2 ] showing strong negative correlations between the concentrations of both gases (cf. Figs. 1 and 2) and thus confirming the hypothesis (1). Obviously, [CO 2 ] diffused in to plant tissues and accumulated in the pith cavities during the night when pressurization was inactive. The highest [CO 2 ] values were detected at the beginning of the sharp increase in culm- [O 2 ], which resembles the flush of CO 2 -rich air from the rhizome system when the through-flow started in the morning (Konnerup et al. 2011 (Singer et al. 1994) and T. latifolia (Constable and Longstreth 1994) but were not linked to plant aeration. Such high [CO 2 ] can only originate from a combination of all biological processes in the submerged zones including aerobic respiration of rhizomes, roots and rhizosphere microorganisms and anaerobic fermentation processes. Specifically in plant A1, the molar CO 2 /O 2 ratio exceeded 1, which is typical for respiratory quotients of anaerobic systems (Dilly 2003). In contrast, in the plant In1, situated in the inflow zone of the CW, the [CO 2 ] even declined at low culm- [O 2 ]. In this part of the CW with high organic load, Dickopp et al. (2011) observed low diurnal oxygen dynamics and argued that below ground plant organs might be less permeable in order to prevent diffusion of toxic fermentation products.
Our observations show that it in situ measurements may reveal conditions, which are difficult to synthesize to an overall explanation as they reflect highly divers settings influencing the individual plants. The below ground rooting system of P. australis is very complex and alters in age and size. Herbivory or diseases may cause the disconnection of parts of the rhizomes from the ventilation stream and thus result in differences between the responses of the plants.

Oxygen supply and consumption
Steep slopes of [O 2 ] decline in rhizomes began to decelerate when oxygen concentrations dropped below 14.6 %, and later oxygen levels remained above 6.7 % (cf. Figs. 1 and 3). In buried horizontal rhizomes of the same part of the studied CW the mean minimum rhizome-[O 2 ] recorded was around 5.8 6 1.5 % SD (Dickopp et al. 2011). The recorded minimum values in the rhizomes may represent the balance between oxygen supply by diffusion and oxygen demand by submerged plant organs, oxygen loss to the rhizosphere and rhizosphere-related aerobic processes. Also the unmodifiable barriers against radial oxygen loss were presumably more effective at lower oxygen gradients between inner gas lacunae and the surrounding anoxic environment (Armstrong et al. 2000). In accordance with our hypothesis (3), the presented observations suggest that plant organs and rhizosphere microorganisms have mechanisms to reduce their rate of oxygen consumption at low oxygen availability, i.e. during night and after culm removal.
Assuming exponential decline is reasonable for the observed [O 2 ] courses and reproduces the convergence to minimum oxygen values inside the reed pith cavities at reduced ventilation. The declining oxygen curves (cf. Figs. 1 and 3) agree with the modelled curves of internal oxygen concentration in respiring roots during downregulation of respiration (Armstrong and Beckett 2011b). Also the soil microorganisms can be assumed to downregulate their rate of aerobic respiration when the oxygen availability becomes limited (Jin and Bethke 2003), i.e. during periods of reduced plant ventilation. The overall rate of substrate-related oxygen consumption in an undisturbed plant-rhizosphere system is reflecting the plant and microbial oxygen demand, oxygen loss from the roots and biogeochemical oxygen demand in the rhizosphere and may react more sensitively than the above mentioned laboratory studies. Our data indicate that substrate-related oxygen consumption already slowed down when air with around 14 % oxygen was present in the pith cavities.
The excision of the investigated culms at midday, when micro-climatic conditions were generally favourable for pressurization, showed the dependence of oxygen on pressurized gas-flow from upper plant parts (Fig.  4). Removing major parts of the above ground plant organs caused [O 2 ] to drop sharply, although the remaining culm stumps had 30 cm of functional leaf sheath area left. The declining oxygen curves after culm-cut show the submerged system is highly dependent on the active ventilation by the culm. In times of active oxygen supply i.e. during midday, the rhizosphere respiration seems to be highest in the presence of ample molecular oxygen, as shown by twice as high initial oxygen consumption after culm cut (À3.3 to À4.6 % h À1 initial decline in intact plant against À6.3 to À8.4 % h À1 after the culm cut at midday). The rhizosphere of P. australis is intensively inhabited by aerobic methane-oxidizing bacteria (Faußer et al. 2012), which are presumably very active during the period of highest oxygen supply. The minimum levels of [O 2 ] indicate that, additionally to plants' oxygen consumption rate (Armstrong and Beckett 2011b), also rhizobacterial metabolic activity will be triggered by the availability of free oxygen (Jin and Bethke 2003).
The reduction of oxygen levels to about 50 % of predawn values indicates that the rhizome system connected to plant A1 was highly dependent on the experimentally removed shoot. Also the high [CO 2 ] of up to 18 % pointed out that there is a large respiratory turnover in the submerged organs and rhizosphere of plant A1. In the plant A2, the [O 2 ] levels in culm stump and rhizome approached 18 h after the culm removal similar values to previous morning concentrations. This indicated that plant A2 was connected to an extended rhizome system, which had good ventilation properties, and could compensate better for the loss of the current culm A2 (cf. C ı zkov a and Lukavsk a 1999). Hypoxic internal conditions in the submerged organs can be expected when internal [O 2 ] converged to levels <6.3 % in culms and 4.2 % in rhizomes (Armstrong et al. 2009;Gupta et al. 2009). If the root internal oxygen concentrations reach the range of the critical oxygen pressure (i.e. 45 hPa or < 3.1 %) respiration will be down regulated (Armstrong et al. 2000(Armstrong et al. , 2009Berry and Norris 1949; C ı zkov a and Bauer 1998; Gupta et al. 2009;Zabalza et al. 2009). Although the presented pre-dawn oxygen concentrations in the rhizome are above the critical oxygen pressure, down regulation can be expected as oxygen is transported inside the roots by diffusion only. The mechanism of oxygen conservation was modelled for respiring root segments by Armstrong and Beckett 2011b). Down-regulation starts when an anoxic core spreads through the root diameter. Therefore the down-regulation of respiration in apical roots is reflected by slow oxygen decline in rhizome cavities, which serve as oxygen reservoir for root and rhizosphere respiration. These results support the hypothesis (3) that demand for oxygen is regulated during periods of low oxygen availability in submerged regions.

Conclusions
Common reed (P. australis) has a high capacity to actively ventilate oxygen-rich air to submerged organs and to the rhizosphere. The relationship between internal oxygen and carbon dioxide concentrations measured under field conditions reflects the dependence of respiration of the submerged organs and biogeochemical processes in the rhizosphere on plant-internal oxygen supply. Regulation of oxygen demand within an intact rhizosphere was observed for the first time in the field and is highly relevant for the understanding of processes in submerged parts of wetland ecosystems.
The obviously reverse dynamics of oxygen and carbon dioxide concentrations reveal a close relationship between the supply of oxygen by the plant and the rhizosphere-related production of CO 2 . The slopes of declining oxygen concentrations in culms and rhizomes during the night and in particular after culm-cut notably demonstrated the regulation of the oxygen consumption for the complete below ground plant-associated system. This underlines the importance of field studies for a better understanding of sediment aeration processes. The oxygen released by roots of wetland plants is readily utilized for oxidative processes (Becket et al. 2001, Bodelier 2003, Faußer et al. 2012) and the metabolic activity of the rhizospheric microbiota is regulated in dependence on available oxygen. Field research on the oxygen-carbon dioxide interrelationship will provide new and globally important data for carbon balance and methane emissions from wetlands.

Sources of Funding
The authors gratefully acknowledge the financial support provided by project CzechGlobe-Centre for Global Climate Change Impact Studies, Reg. No. CZ.1.05/1.1.00/ 02.0073) and also by the national project for infrastructure support CzeCOS/ICOS Reg.No. LM2010007.

Contributions by the Authors
A.F. performed the measurements of internal oxygen and gas exchange on the P. australis plants in the CW and evaluated the data for the preparation and writing of the article. J.D. analyzed the carbon dioxide content in the gas samples from the reed culms. M.K. and H.
C. were the supervisors of the project. All authors contributed in data evaluation and article preparation.

Conflicts of Interest
None declared