A meta-analysis of plant tissue O₂ dynamics

Literature search

We performed a structured and systematic literature search to identify studies reporting O₂ measurements in vegetative plant tissues. The search was conducted using the Web of Science BIOSIS database applying Boolean operators, as explained in detail in the Supplementary materials. The search resulted in a final list of 129 studies published up to September 2022 used for this review. We extracted each observation of plant tissue O₂ status and entered it into a database alongside information of plant species, tissue type, light conditions, submergence status, treatment, O₂ measurement technique, tissue excisions status etc. if provided in the source. The list of studies used for data extraction is available in Supplementary materials. The database of plant tissue O₂ levels can be accessed from the Dryad data repository at https://doi.org/10.5061/dryad.cnp5hqc8v.

Data handling

The studies used for this review provide a wide range of ways to measure and report levels of plant tissue O₂. The following steps describe how we performed data extraction for our meta-analysis.

In a few studies, tissue O₂ was sampled by extracting internal plant gasses in a liquid such as ambient flood water. Such studies where we suspected that air bubbles escaping from plants tissues had been in contact with surrounding medium (thus allowing for O₂ exchange) prior to measurements were discarded (e.g. Setter et al. (1987); Erdmann and Wiedenroth (1988); Stevens et al. (2002)). Studies where gas sampling was performed using syringes, preventing O₂ exchange with the surrounding medium, were included in the analyses.

For correlating submerged plant shoot tissue pO₂ with water column pO₂, we identified 26 studies simultaneously reporting shoot tissue and water column pO₂ levels. Corresponding values of tissue and water column pO₂ were then extracted. For diurnal measurement series, two to three values spread across the period were extracted. When water column levels were manipulated, shoot values for each steady state level of water column pO₂ were extracted. To allow for effects of internal gas transport to be evident in the meta-analysis, only measurements from intact plants were used for this analysis.

Statistical analysis

Data were analysed using GraphPad Prism 9 (GraphPad Software, La Jolla, CA, USA) or R (R Core Team 2022) for Windows statistical software. Normality of distributions was confirmed by visual inspections of residual plots and Shapiro–Wilk normality test, and homogeneity of variance by using F-test or Brown–Forsythe test (P > 0.05). When data failed these assumptions, data was transformed, or a non-parametric test was chosen. Further details on statistical tests are given in the respective figure captions.

Modelling

We evaluated the influence of environmental conditions, plant species and tissue type on plant O₂ status using an information-theoretic model selection approach. This explorative, data-driven approach is increasingly used when several predictors may be associated with a particular trait (Symonds and Moussalli 2011). The information-theoretic model selection approach differs from traditional null-hypothesis significance testing, where a hypothesis is first defined, and then accepted or rejected based on P-values (Symonds and Moussalli 2011). Each competing model predicts the outcome variable (in our case: plant tissue O₂ status) using predictors and their interactions (in our case, submergence, light, tissue type etc.). Thus, each competing model can be viewed as a different hypothesis of how predictors affect plant O₂. Ranking the competing models allowed us to estimate which model best explains tissue O₂ status. We used a type of regression analysis, Linear Mixed-effects Model (LMM), to estimate the model parameters and their importance in controlling plant O₂ status. The use of mixed model allowed us to include plant species as a random factor, taking the variability in plant tissue O₂ related to species specific factors (e.g. aerenchyma formation) into account.

We compared all the possible predictors and combinations of interaction and selected the best model using the Akaike Information Criterion (AIC) (Burnham and Anderson 2004). AIC values derive meaning from comparing with AIC of other generated candidate models. By penalising the number of model predictors, AIC allows for the selection of the ‘best’ model based on fit (e.g. large R²) but also model complexity, and thus presents a compromise between fit and complexity. For details on applied R packages, data categorisation and diagnostic plot (Fig. S1), see Supplementary materials.

‘Submergence’, ‘light’, and ‘tissue type’ significantly affect plant tissue O₂ status

Our information-theoretic model selection identified ‘submergence’, ‘light’, ‘tissue type’ and ‘light × submergence’ interaction as the most important predictors of plant tissue status; i.e. these predictors had a relative variable importance of 1.0 in the only plausible linear mixed-effects model (Table 1). Meanwhile, the ‘tissue type × light’ interaction only had a relative variable importance of 0.8. This indicates that the ‘tissue type × light’ interaction was a weaker predictor of tissue O₂ than the other predictors.

Including species as a random factor increased the model fit. This is evident as marginal R² was 0.40 without species as a random factor, compared with conditional R² of 0.55 with the random factor included (Table 1). Following the data-driven identification of the predictors ‘submergence’, ‘light’, ‘tissue type’ and ‘light × submergence’ controlling plant tissue O₂ status, we continued by exploring our database of published O₂ status for patterns generated by these predictors.

Plant O₂ status depends on plant tissue type

Our meta-analysis showed that O₂ status declined significantly from leaves to stems, and from stems to roots; i.e. median O₂ status was 15.7 kPa in leaves, 13.1 kPa in stems and 9.8 kPa in roots, respectively (Fig. 1). We interpret the significantly higher O₂ status in leaves to be caused in three ways: (1) by leaves producing endogenous O₂ via photosynthesis when in light; (2) by leaf stomata being able to vent leaf gas spaces, thus allowing for leaf O₂ status to remain closer to atmospheric levels than in more dense tissues without stomata; and (3) by leaves normally being in close contact with the atmosphere, in contrast to roots or rhizomes, which are in direct contact with O₂ consuming soils or sediments. Indeed, violin-plots showed a relatively high number of anoxic conditions in root tissues compared with leaves (Fig. 1). It should be noted that distribution of O₂ levels in roots may be influenced by a high number of wetland and aquatic plants in the published data underlying this analysis. In detail, 78% of the observations originated from aquatic and wetland plant species and 21% from terrestrial plants. Plant species inhabiting wetland and aquatic systems possess root traits such as aerenchyma that enhance internal aeration and thus ensure adequate tissue O₂ supply for root functioning, whereas terrestrial plants typically lack such traits leading to O₂ depletion in flooded soils (Armstrong 1979). The distribution of O₂ levels reported in Fig. 1 should therefore be treated with caution when inferring a ‘true’ distribution of O₂ in all plants.

Fig. 1.

O₂ partial pressures (pO₂) in leaf, stem and root tissues. Numbers of replicate values (n): 322, 633 and 404 for leaf, stem, and root tissues, respectively. Letters denote significant difference between medians (non-parametric Kruskal–Wallis test with Dunn’s multiple comparisons, P < 0.05). Truncated violin plot shapes illustrate observation frequency, the solid line the median and dashed line the quartiles. Data from experiments where the O₂ level of the atmosphere/submergence medium around the shoot was manipulated (i.e. atmosphere or submergence solution around shoot was purged with N₂) was excluded in this figure.

The broad categorisation of tissue type into leaf, stem and root in Fig. 1 does not do justice to the vast range in morphology and function of different tissue types within these main categories. For example, the category of roots encompasses diverse tissues such as root tips in anoxic soils, mangrove pneumatophores (acting as pathways for atmospheric O₂ inlet), aquatic adventitious roots and other contrasting root tissues. For example, two very high root tissue O₂ levels (43.2 and 32.5 kPa, Fig. 1) were from chlorophyll containing, photosynthesising aquatic adventitious roots of Meionectes brownii (Rich et al. 2013). In addition to the broad categories in Fig. 1, we therefore also categorised measurements from leaf, shoot and root tissues into several subcategories (Fig. 2).

Fig. 2.

O₂ partial pressures (pO₂) in (a) leaf, (b) stem and (c) root tissue. In (b) ‘Stem tissues’ encompass stems, culms, stem internodes, peduncles, or stolon tissues but not rhizomes, storage organs (tubers and corms) and woody tree stems. Tree stems are woody stems of beech (Fagus sylvatica), Norway spruce (Picea abies), olive (Olea europaea), red maple (Acer rubrum), eastern hemlock (Tsuger canadenis), Northern red oak (Quercus rubra), black poplar (Populus nigra) and American ash (Fraxinus americana). Numbers of replicate values (n): leaf blade/lacunae (102), leaf sheath (13), leaf petiole (125), flower (11), leaf/flower bud (14), ‘Stem’ (306), tree stem (97), potato tubers (28), rhizome (135), root axis (320), aquatic roots (8), pneumatophore (9), root nodule (15). Letters denote significant difference between medians (non-parametric Kruskal–Wallis test with Dunn’s multiple comparisons, P < 0.05). Truncated violin plot shapes illustrate observation frequency, the solid line the median and dashed line the quartiles. Data from experiments where the O₂ level of the atmosphere/submergence medium around the shoot was manipulated (i.e. atmosphere or submergence solution around shoot was purged with N₂) was excluded in this figure.

In the leaf tissues subcategories O₂ levels were relatively similar compared with the vast variation in stem and root tissues; i.e. median O₂ status only ranged from 11.4 to 16.8 kPa (Fig. 2a). The only significant difference was found between median leaf blades (16.8 kPa) and leaf petioles (15.2 kPa). There was no significant difference between the remaining leaf categories (leaf sheath, flower, and bud tissues), most likely due to a low number of O₂ observations in these tissues (n < 14) compared with leaf blades and petioles (n > 102).

Tissue O₂ status of stems showed larger variation than leaf tissue and was significantly influenced by stem tissue type. Hence, O₂ status in stem tissues was lower in thick, belowground, and predominantly respiring tissue types, compared with porous and photosynthesising tissue types (Fig. 2b). In stems, median O₂ status in tissues such as culms, peduncles, stolons and stems of herbaceous plants (14.9 kPa, designated as ‘stems’ in Fig. 2b) and tree stems (13.1 kPa) were significantly higher, than in rhizomes (9.4 kPa) and the dense storage organ of Solanum tuberosum (7.6 kPa, designated as potato tuber in Fig. 2b). Thick and dense tissues showing high resistance across the diffusion path are prone to internal O₂ gradients, especially if respiratory demand for O₂ is large (Armstrong 1979; Colmer and Greenway 2005). Indeed, 32 g potato tubers experienced relatively low central O₂ status of 1.5 kPa compared with 7 kPa in 8 g small tubers (Geigenberger et al. 2000). Another example of low internal O₂ status in dense stem tissues results from radial profiling across dense succulent stem tissues of submerged Halosarcia pergranulata using an O₂ microsensor (Pedersen et al. 2006). Here, internal O₂ declined to 1.5 kPa compared with 21 kPa in the submergence solution. Respiration rates also affect plant tissue O₂, exemplified in potatoes stored at 22–25°C showing less than half of the tissue O₂ compared with potatoes stored at 5–11°C (Stiles 1960). In summary, stem tissue O₂ was particularly low in thick, dense, or predominantly respiring tissues such as succulent tissues or belowground rhizomes.

In stark contrast to potato tubers and succulent stems, median O₂ was significantly higher in the other stem categories. Plant stems may contain lysigenous or schizogenous aerenchyma, facilitating gas exchange via interconnected air spaces that enhance O₂ diffusion due to the 10 000-fold higher diffusion coefficients in gas than in liquid phase (Armstrong 1979). Since diffusion is a very slow process over long distances, some specialised wetland plants can induce ‘internal winds’ (pressure driven, convective gas flow) within connected air spaces allowing for effective O₂ transport over long distances (Dacey 1980). For example, O₂ in submerged culms of the sedge Eleocharis sphacelata remained close to atmospheric levels of 20 kPa down to 2.7 m of water depth during daytime where convective flow was operating, compared with <10 kPa during the night where convective flow is absent (Sorrell and Tanner 2000). This was a result of pressurised flow of gas from influx culms to efflux culms of up to 5.8 mL min⁻¹, supplying immersed tissues with atmospheric O₂.

Root tissue O₂ showed even higher range in median O₂ status than leaf and stem tissues. Median O₂ reached 15.2 kPa in pneumatophores but was 0 kPa in root nodules (Fig. 2c). In root nodules, symbiotic bacteria convert N₂ gas into ammonia to be used in amino acid biosynthesis. Since the process of nitrogen fixation is highly O₂ sensitive, root nodules have evolved mechanisms such as leghaemoglobin to regulate free O₂ inside the nodule (Downie 2005). Indeed, the few observations of significant O₂ inside root nodules belong to experiments where nodules were inoculated with mutant bacteria strains, some of which were unable to infect plant cells (Masepohl et al. 1993; Romanov et al. 1995). In contrast, the median O₂ level in pneumatophores was significantly higher (15.2 kPa) than in the root axis or root nodules. Pneumatophores are lateral roots known from plants growing in anoxic mudflats such as mangroves (e.g. Avicennia marina), where they exhibit negative geotropism thereby growing upward. When reaching the sediment surface, O₂ can pass through small bark openings (lenticles) and diffuse into the belowground root system. Other root tissues showing high O₂ status were photosynthesising aquatic roots (Rich et al. 2011). Here, O₂ status were as high as 43.2 kPa (median of 14.2 kPa, Fig. 2c). O₂ may build up to such high levels in submerged, photosynthesising tissues due to physical restrictions in gas diffusion preventing photosynthetic O₂ to diffuse out of the tissues and into the surrounding water.

O₂ gradients within plant tissues

Obtaining O₂ status with high spatial resolution has also revealed that O₂ status vary significantly not only between but also within tissue types. Such data has mainly been obtained using Clark-type O₂ microelectrodes. With a tip diameter of down to 3 μm (Weits et al. 2019) the electrode can be positioned in delicate tissues such as roots, even in field conditions. Our meta-analysis of literature data from studies reporting 40 paired root measurements of stele and cortex O₂ status, reveal that mean O₂ status in the root stele was 66% of the cortex O₂ status (Fig. 3). The root stele is known to be more metabolically active compared with cortex tissues; e.g. for banana (Musa spp.) roots, a 6-fold difference in respiration rate was found (Aguilar et al. 2003). In addition to displaying high respiration rates, stele tissues only contain relatively few air spaces that further increases the O₂ demand per volume of root tissue. In contrast, the root cortex may contain large air spaces caused by aerenchyma formation, allowing for longitudinal flow of O₂ from the shoot or atmosphere toward the root tip (Armstrong 1979). Thus, the resistance to O₂ diffusion increases in the dense stele compared with the cortex. The high respiration rates and low diffusivity in dense tissues result in the lower O₂ status within root steles, as supported by modelling results (Armstrong et al. 2000).

Fig. 3.

O₂ partial pressures (pO₂) in root cortex and stele tissues reported in the literature. Values were extracted from 13 published studies where O₂ microsensors were driven through both tissues (i.e. matched pairs). * denotes statistical significant difference between means (paired t-test on square-root transformed data, P < 0.0001, n = 40). Truncated violin plot shapes illustrate observation frequency, the solid line the median and dashed line the quartiles. Mean pO₂ was 6.2 kPa in the cortex and 4.1 kPa in the stele, respectively.

In addition to root tissues, shoot tissues may also exhibit internal O₂ gradients. Molecular O₂ inside the shoot apical meristem of Arabidopsis thaliana (3.6–8.4 kPa) compared with the surrounding stem tissue (13–19 kPa) regulated leaf production, with hypoxia inhibiting proteolysis of an N-degron-pathway (Weits et al. 2019). The observed gradient in the shoot apex of 15 kPa developed over a distance less than 50 μm. Meanwhile, O₂ within the 2-mm thick leaf petioles of Rumex palustris was much more homogeneous, probably owing to the large number of internal gas spaces (aerenchyma content of >20%) and underwater photosynthesis of the green petioles (Mommer et al. 2004). Corresponding to a longitudinal decline of O₂ in roots (discussed below), profiles along the leaf petioles of R. palustris exhibited ~3 kPa longitudinal O₂ decline from a position close to the O₂ source (leaf) to O₂ sink (roots) (Mommer et al. 2004).

Plant tissues also show significant longitudinal variation in internal O₂ status. In roots, the degree of longitudinal O₂ diffusion from the shoot (O₂ source) via the root base to the root tip (O₂ sink) is crucial for plant functioning when roots grow in O₂ free media such as soils with excess water (Armstrong 1979). Plant adaptations and acclimation enabling a continuous supply of shoot-derived molecular O₂ to the metabolically active root tip have therefore been thoroughly studied. Key traits such as aerenchyma formation (constitutive as well as inducible) and induction of barriers to radial O₂ loss, which prevent O₂ leaking from the root axis into anoxic soils, allow for continued root aeration despite the anoxic soil environment (Colmer 2003). To evaluate the effect on root aeration of a barrier to radial O₂ loss and aerenchyma formation, one can compare O₂ status of roots grown in a stagnant, deoxygenated nutrient solution mimicking a flooded soil (Wiengweera et al. 1997) with roots grown in aerated nutrient solution (mimicking a drained soil). Most wetland species will further induce aerenchyma and a barrier to radial O₂ loss only in the former condition. For example, in Zea nicaraguensis, root apical O₂ status was 8.7 kPa in roots formed in stagnant, deoxygenated nutrient solutions compared with 0.47 kPa in roots from aerated nutrient solution when measured in roots in a severely hypoxic solution (Pedersen et al. 2021). In the roots formed in stagnant, deoxygenated nutrient solution, a barrier to radial O₂ loss was induced, and root aerenchyma was 1.4–fold of that under aerated conditions (Pedersen et al. 2021). Thus, in this concrete example, aerenchyma and the barrier to radial O₂ loss resulted in better O₂ status of the roots formed in the stagnant, deoxygenated nutrient solution.

In order to isolate the effect of the barrier to radial O₂ loss on root aeration without shoot-derived O₂ and varying root lengths impeding interpretations, a recent study reported radial O₂ intrusion into submerged root segments of rice (Oryza sativa) (Peralta Ogorek et al. 2021). Here, maximum O₂ status within the root segments with a weak barrier to radial O₂ loss differed 54-fold from roots with a tight barrier, when incubated with 21 kPa O₂ in the surrounding water (Peralta Ogorek et al. 2021). The strong effect of the barrier to radial O₂ loss on root O₂ status underlines the importance of species-specific adaptations and/or acclimations for O₂ status in plant tissues.

In conclusion, both root and shoot tissues may show substantial radial and longitudinal internal O₂ gradients. The gradients depend on tissue density, respiration activity, plant adaptations and/or acclimations such as aerenchyma and radial O₂ loss barrier formation, and environmental conditions.

O₂ status in submerged plant tissues – the importance of light

Submergence is a compound stress to plants, brought on by several effects. For one, the 10 000–fold slower gas diffusion in liquid compared with gas impedes the diffusive supply from the environment of O₂ for respiration and CO₂ for photosynthesis (Jackson and Drew 1984). As floodwaters absorb and scatter incoming light, photosynthesis may be hampered even further (Colmer et al. 2011). Build up of the volatile phytohormone ethylene may induce leaf senescence as chlorophyll is degraded (Voesenek and Bailey-Serres 2015). Thus, the capacity for underwater photosynthesis and therefore O₂ production generally declines with time of submergence for terrestrial vegetation (Colmer et al. 2011). Indeed, our data-driven model-selection approach identified both ‘submergence’ and ‘submergence × light’ interaction as factors that significantly affect plant tissue O₂ status (Table 1). It should be noted that the identification of ‘submergence’ over other environmental conditions could be due to a relatively large number of studies focusing on plant tissue O₂ in submerged plants (52% of observations used for model selection), compared with plants subject to drought, nutrient limitations, or pathogen infections. Nevertheless, the effect of submergence on plant tissue O₂ per se was validated by the model selection. To further quantify the effect of light and submergence on plant tissue O₂ status, we filtered data from the literature where O₂ was measured inside intact plants during complete submergence or with atmospheric contact (drained with shoot in air, waterlogged with shoot in air, or partial submerged shoot), and in light or dark conditions. We excluded data from experiments where the O₂ level of the atmosphere/submergence medium around the shoot was manipulated (i.e. atmosphere or submergence solution around the shoot was purged with N₂), as this would otherwise mask any light or submergence effects. This analysis underlined the vast importance of light particularly for shoot tissues (Fig. 4). Shoots in air contained 24% less O₂ in dark than in light, most likely owing to the lack of photosynthetic O₂ under dark conditions. The effect of light on tissue O₂ was also evident in root tissues, where mean O₂ status declined from 10.6 kPa in roots of plants in light to 7.9 kPa in roots of plants under dark conditions.

Fig. 4.

O₂ partial pressures (pO₂) in (a) shoot tissues and (b) root tissues during complete submergence (‘shoot submerged’, filled bars) or the shoot in air (‘shoot in air’, i.e. drained conditions, waterlogging and partial submergence, open bars) in light or dark conditions. Data were extracted from 117 studies where O₂ was measured inside intact plants (i.e. excluding studies where O₂ was measured within excised tissues) and where the shoot was clearly stated as being in light or dark conditions. Data was excluded when the O₂ level of the atmosphere/submergence medium around the shoot was manipulated (i.e. atmosphere or submergence solution around the shoot was purged with N₂). (a) Letters denote statistically significant differences between means (P < 0.05, Tukey post hoc test; n = 201, 226, 106 and 221 for columns from left to right) following 2-way ANOVA showing a significant light (P < 0.0001) and submergence × light interaction (P < 0.0001). (b) Letters denote statistically significant differences between means (P < 0.05, Tukey post hoc test; n = 115, 48, 56 and 31 for columns from left to right) following 2-way ANOVA showing a significant light (P < 0.0001), submergence (P = 0.0001) and submergence × light interaction (P = 0.0078). Bars show mean and error bars s.e.m.

The effect of complete submergence on shoot tissue O₂ status was strongly light dependent. In the presence of light, submergence resulted in a significant increase (44%) in shoot tissue O₂ status, compared with shoots with air contact (Fig. 4a). In contrast, when the entire shoot was submerged in dark conditions, submergence resulted in a significant decline (42%) in O₂ status. Hence two-way ANOVA showed a significant ‘light’ and ‘submergence × light’ interaction. We interpret the submergence-induced increase of shoot tissue in light as a build up of O₂ deriving endogenously from underwater photosynthesis, or to a lesser extent, from floodwaters containing high levels of O₂. O₂ produced in shoot tissues by underwater photosynthesis is likely to build up as stomatal closure, physical boundary layers and slow outward diffusion of O₂ into the aquatic environment hampers gas exchange between the shoot and its surroundings (Sand-Jensen et al. 2005; Pedersen et al. 2006, 2016). In contrast, in shoots with atmospheric contact, O₂ deriving from photosynthesis would be vented into the surrounding air. The importance of light on plant tissue O₂ in situ is nicely exemplified by diurnal leaf O₂ and irradiance measurements in seagrass (Thalassia testidinum) (Koch et al. 2022a). Here, a decline in daytime irradiance of ~1000 μmol photons m⁻² s⁻¹ from passage of clouds resulted in a rapid decrease in leaf O₂ status from 45 to 27 kPa.

The submergence-induced increase in shoot tissue O₂ observed in Fig. 4a could also be due to high levels of O₂ in floodwaters. High levels (>30 kPa) of O₂ in natural floodwaters have been observed in marine (Pedersen et al. 2016) and freshwater (Rich et al. 2013) ecosystems. To test how floodwater O₂ levels affect plant tissue O₂, we performed a correlation analysis of data from studies simultaneously measuring plant and floodwater O₂ levels (Fig. 5). Correlation between shoot tissue O₂ levels and water column O₂ levels was significant in both cases but stronger in the dark (r = 0.6808, Spearman, P < 0.0001) than in light (r = 0.1925, Pearson, P = 0.0384). The weaker correlation in light than in the dark indicates that production of O₂ in underwater photosynthesis provides the plant with an alternative O₂ source. As net underwater photosynthesis produces excess endogenous O₂, the plant is less likely to experience the same O₂ level as the surrounding water column. In contrast, plants submerged in the dark rely on their immediate surroundings as O₂ source, and hence the correlation is stronger under dark conditions than in the light. Since plant respiration consumes O₂, points cluster below the line of equality in the dark (Fig. 5), but not in light where photosynthesis can elevate tissue O₂ to above water column O₂.

Fig. 5.

O₂ partial pressures (pO₂) in submerged tissues vs O₂ levels in the surrounding water column under (a) dark or (b) light conditions with the line of equality (identity line). Values were extracted from 26 studies. O₂ values are from freshwater plants (Lobelia dortmanna, Litorella uniflora and Isoetes autralis), seagrasses (Cymodocea nodosa, Thalassia testudinum, Zostera marina, Zostera muelleri, Syringodium foliforme, Posidonia sinuosa, Thalassia hemprichii and Enhalus acoroides) and wetland plants (Suaeda maritima, Glyceria fluitans, Rumex palustris, Halosarcia pergranulata, Meinocetes brownii, Spartina anglica and Oryza sativa) where O₂ in submerged tissues was measured simultaneously with water column O₂. In experiments where diurnal variations were recorded, diurnal minimum and maximum values were extracted. For Pedersen et al. (2016), tidal waters caused the water column O₂ to fluctuate so O₂ measurements at the start, middle and end of the dark and light period, respectively, were extracted. Tissue types were leaves, leaf sheaths, rhizomes, stems, leaf meristems and leaf petioles but not roots. (a) Tissue and water column pO₂ recorded under dark conditions (at night in field experiments, or without day light/artificial light in laboratory experiments). Spearman correlation coefficient = 0.6808, P < 0.0001 (n = 325). (b) Tissue and water column pO₂ recorded under light conditions (during daytime in field experiments, or with artificial light in laboratory experiments). Pearson correlation coefficient = 0.1925, P = 0.0384 (n = 116).

Interestingly, three measurements from the leaf lacunae of the small isoetid Lobelia dortmanna (Møller and Sand-Jensen 2011) showed anoxic conditions in the shoot but high levels of floodwater O₂ (21–22 kPa) (Fig. 5). The reason why floodwater O₂ did not diffuse into the leaves of L. dortmanna most likely relates to its strategy for CO₂ acquisition. L. dortmanna is found in oligotrophic soft water lakes, where underwater photosynthesis is often carbon limited. Thick leaf cuticles therefore reduce permeability to CO₂ (and O₂), helping L. dortmanna to acquire sediment-derived CO₂ for underwater photosynthesis via its root system. In the experiments by Møller and Sand-Jensen (2011), the sediment was enriched with organic matter, resulting in sediment O₂ depletion. Due to the lack of a barrier to radial O₂ loss in the root, and continuous internal air lacunae, O₂ was readily lost from the plant to the sediment. Meanwhile, the thick leaf cuticle restricted O₂ exchange with oxygenated floodwater. These observations nicely underline how plant tissue O₂ dynamics follow general trends but may deviate if habitats require special adaptations.

Continuous O₂ measurements using microelectrodes have revealed severe hyperoxia of >40 kPa O₂ (Sand-Jensen et al. 2005; Pedersen et al. 2006, 2016; Rich et al. 2013; Koch et al. 2022a) in submerged wetland and aquatic plants. The highest non-manipulated plant tissue O₂ level reported in the literature (53.1 kPa) was measured in Thalassia hemprichii subjected to submergence by tidal waters and high radiation (2000 μmol photons m⁻² s⁻¹) (Pedersen et al. 2016). Hyperoxia in submerged vegetation coincides with high irradiation and may even intermit with night time hypoxia (Sand-Jensen et al. 2005; Pedersen et al. 2006, 2016). High levels of O₂ may result in the formation of ROS and increased photorespiration; however, the degree of tissue damage and plant mechanisms to tolerate hyperoxia remain much less studied than for hypoxia.

In our meta-analysis of tissue O₂ status, we have attempted to summarise the current state of knowledge on plant O₂ dynamics. Much knowledge on plant tissue O₂ dynamics has been gained due to ongoing development of new O₂ measuring techniques. In the following, we therefore present an overview of the progress in determining plant tissue O₂ status. We then use this overview to identify future opportunities to obtain further insight of plant O₂ dynamics.

Measuring O₂ in plant tissues: past, present, and future approaches

The earliest measurements of O₂ in plant tissues made use of various forms of gas analysers. Gases sampled from tissues using syringes (or by crushing in mercury) was drawn into a glass chamber, which was then brought into contact with an O₂ absorbing solution, allowing for the decrease in gas volume to be monitored (Thoday 1913; Scholander 1947). The earliest methods to our knowledge was the Bonnier-Mangin microtechnique (Thoday 1913) with adaptations by Scholander (1947), used as late as 1994 on tomato (Solanum lycopersicum) and potato tubers (Nery and Calbo 1994). Other studies (Laing 1940) applied the Henderson-Haldane gas analyser, widely used in human physiology research, for measuring O₂ in aquatic plants. Finally, gas sampled from tissues with syringes has been analysed using gas chromatographs (Spalding et al. 1979; Tjepkema and Cartica 1982). To our knowledge, the results obtained using gas analysers have not been subject to a direct comparison with other methods such as microsensors.

The development of microsensors allowed for O₂ measurements with improved spatial and temporal resolution (Revsbech 1989). The most popular sensors are miniaturised, stiff and narrow (down to 3 μm tip diameter) Clark-type sensors, ideal for penetrating plant tissue and measuring O₂ within plants (Pedersen et al. 2020). Optical O₂ sensors (O₂ optodes) can also be positioned within plant tissues, especially if built into a syringe. Foils coated with O₂-sensitive fluorescent dyes and optical O₂ sensor nanoparticles are often used for mapping O₂ on plant surfaces and their surroundings (Rudolph et al. 2012; Brodersen et al. 2014; Koren et al. 2015). However, optical O₂ sensor nanoparticles have only seen limited application inside plants tissues due to their size of up to 600 nm (Schmälzlin et al. 2005; Shaw and Honeychurch 2022). Because of microsensor popularity in plant research, >70% of the 1567 observations of plant tissue O₂ used in this study were obtained using an O₂ microsensor. For a more detailed review on use of microsensors in plant biology, see Pedersen et al. (2020).

While the abovementioned methods allow for direct O₂ measurements within plant tissues, more indirect methods have also been applied. Measurements of radial O₂ loss from roots using cylindrical platinum electrodes have been used to model root internal O₂ status (Armstrong and Wright 1975); see also Jiménez et al. (2021) for a recent review on methods on measurement of radial O₂ loss from roots. Another indirect measurement of tissue O₂ status is related to the expression of genes strongly upregulated upon low O₂ (referred to as hypoxic reporters) (Mustroph et al. 2010).

In the present meta-analysis, we identified ‘tissue type’, ‘submergence’, ‘light’ and ‘light × submergence’ interaction as important environmental factors for plant O₂ status. When conducting meta-analysis, researchers should consider the risk of ‘publication bias’, a term originating from health sciences describing information suppression mechanisms (Dickersin 2005). This includes language bias or accession bias (selective inclusion of studies that are accessible, and published in English), and outcome bias (selective reporting of statistically significant results) among others (Dickersin 2005). The latter would translate to studies reporting surprisingly low or high plant tissue O₂ levels, or significant treatment effects of for example plant submergence, being more likely to reach publication than negative results, either due to bias at the investigator or editorial level. The concern that O₂ measurements of experiments revealing no treatment effects are under-reported cannot be excluded. However, considering the large number of studies across time, space, species, and methodology describing significant effects of light and submergence on plant tissue O₂ levels, we consider their potential importance as validated by this study. We also believe that our data gives an impression of the range and variability in plant tissue O₂ levels, although random sampling both at community, species and tissue level would be required to achieve a distribution resembling the ‘true’ distribution of O₂ in all plants.

Methods to conduct non-invasive O₂ measurements at high spatial and temporal resolution should promote expansion of plant tissue O₂ research from predominantly studying the effect of submergence into other areas. Areas already under investigation include studying plant O₂ levels in relation to plant development (Weits et al. 2019), pathology (Kumari et al. 2017) and plant signalling (Weits et al. 2021). It has been noted that unravelling tissue O₂ dynamics in plants still lacks behind the mammalian field (Ast et al. 2012; Schmidt et al. 2018). This is underlined by the search term ‘tissue hypoxia’ in the Web of Science BIOSIS Previews database resulting in 34 970 records belonging to the ‘Mammalia’ taxa compared with 1041 in ‘Plantae’ (performed in November 2022). In mammals, cancer research resulted in the development of methods such as positron emission tomography (PET), genetically encoded sensors, and ¹⁸F NMR for non-invasive O₂ measurements (Schmidt et al. 2018). Challenges in translating these methods to plant biology include plant tissue autofluorescence (Ast et al. 2012). However, PET scans of plants tissues used to study carbon dynamics illustrate the possible future application of this technique in plant science (Mincke et al. 2021).