Root Porosity Contributes to Root Trait Space of Wetland Monocotyledons Independently of Economics Traits.

Aims Root aerenchyma, a key adaptive trait to anoxic soils has rarely been integrated into trait-based plant ecology. This study aims to evaluate the relationship between root porosity and root economics-related traits among wetland plants, focusing on the effect of aerenchyma on root tissue density, a central trait in plant economics spectrum. Methods Root porosity, root tissue density with air-space included or excluded (RTD and RTDA), and other root economics-related traits were measured separately for basal and lateral roots of 16 garden-grown Ontario wetland monocots with contrasting root longevities. Results Interspecic variation in root porosity was unrelated to root economics traits and did not differ between species with long-lived or short-lived roots. Consequently, RTDA better differentiated between species with contrasting root longevities than RTD did, consistently both for basal and lateral roots. Root dry matter content (RDMC) accurately predicted RTDA. A principal component analysis showed that in the root adaptive trait space of wetland plants, the rst dimension is dened by economics-related traits, the second dimension by lateral root porosity and the ratio of lateral to basal root length, and the third dimension by basal root porosity. Conclusions Interspecic variation in the aerenchyma content is independent of root economics: Wetland plants can construct economically conservative or acquisitive roots of any porosity. Consequently, to consistently express root functional relationships among wetland plant species, root tissue density should be expressed with RTDA, i.e., excluding the air space, or with the more easily measured RDMC.


Introduction
Root economics spectrum (RES), based on morphological and physiological root traits that are associated with resource acquisition and resource conservation, has been suggested to match the aboveground leaf economics spectrum (Mommer and Weemstra 2012; Reich 2014; Roumet et al. 2016). Traits such as speci c root length (SRL), root respiration rate, root tissue density (RTD) and root life span have been shown to co-vary across species and biomes, enabling their potential use to form a synthetic spectrum to represent root economics, i.e., patterns of allocation and turnover of carbon and nutrients (Ryser 1996;Roumet et al. 2016;Liu et al. 2016; Han and Zhu 2020). However, it also has been observed that RES is more complex than above-ground trait relationships within the leaf economics spectrum would indicate (Weemstra et al. 2016(Weemstra et al. , 2021Bergmann et al. 2017Bergmann et al. , 2020Freschet et al. 2020a). There are several possible reasons for this complexity. The multiplicity of belowground environmental constraints could result in a multidimensional root economics space (Weemstra et al. 2016), and the commonly used root traits may be inadequate for disentangling different adaptations to the multiple constraints (Freschet et al. 2020a). Thus, the applicability of RES may be ecosystem-dependent, depending on the speci c below-ground constraints the plants are subjected to in a given ecosystem (Shipley et al. 2016;Carvajal et al. 2018), and, a functional RES may require the use of more speci c traits (Poorter and Ryser, 2015;McCormack et al. 2017).
Wetlands are ecosystems that pose such speci c constraints on roots. Most importantly, wetland soils are saturated, resulting in a reducing hypoxic or anoxic environment (Mitsch and Gosselink 2015). Wetland plants have adapted to such conditions by developing porous roots that allow transport of oxygen-containing air from the atmosphere down into the oxygen-limited environment, enabling root respiration and affecting the biogeochemistry of the rhizosphere (Armstrong 1980;Lamers et al. 2013; Marzocchi et al. 2019). Such air-conducting tissue, root aerenchyma, can also develop in terrestrial plants, especially in habitats prone to occasional ooding (Visser et al. 2000), but in wetland plants it is constitutive (Laan et al. 1989;Jackson and Colmer 2005). Root porosity, i.e., the fractional air space, varies among different species and habitats, reported values ranging from 5-55% (Smirnoff and The aim of the present study is to elucidate the relationship between the interspeci c variations in root porosity and in root economics-related traits among wetland plants. As root porosity is mainly an adaptation to oxygen de ciency, and the economics-related traits are adaptations to nutrient availability and disturbance, the question arises, how these traits interact among wetland plant species (Moor et al. 2017). It has been hypothesized that variation in root porosity might be orthogonal to the expected RES in wetland plants, due to the independent nature of the major selective pressures, i.e., survival under oxygen de ciency and growth under contrasting levels of nutrient supply (Pan et al. 2019). On the other hand, synergy or tradeoffs among the traits could lead to correlations between them. For instance, ooding stress has been shown not only to result in changes in root air space, but also in changes in root diameter and RTD (Visser et al. 2000;Striker et al. 2007;Ryser et al. 2011). Purcell et al. (2019) showed that root porosity correlated with SRL but was independent of dry matter content across a gradient of waterlogging duration.
RTD is considered to be a key root trait in comparative ecology, as it expresses the amount of structural material (dry mass) invested per unit volume of root, thus having consequences both for growth rate and root robustness, and consequently, on life span (Ryser 1996;Birouste et al. 2014). However, variation in the amount of root air space potentially confounds the association of RTD with growth and life span, as gaseous volume does not add to tissue dry mass (Eissenstat et al. 2000;Bouma et al. 2003). The confounding effect would be especially pronounced if root porosity is independent of root economics and if it shows a large interspeci c variation. This problem can be solved by measuring the density of the non-gaseous root tissue, i.e., excluding the air space from the volume measurement. Therefore, we propose a modi ed and more speci c variable: Root tissue density with air excluded (RTDA), i.e., root dry mass per root volume without its air space. We investigate whether this new variable enables a functional improvement of the wetland RES. In addition, because the measurement of root tissue volume is laborious, we also examine the applicability of root dry matter content (RDMC) as a more easily measurable substitute for RTDA. RDMC has previously been recommended as a decent proxy of RTD among non-wetland plants (Birouste et al. 2014).
This study has two primary objectives: 1) to quantify the confounding effect of root aerenchyma in wetland plants on the relationship between RTD and other root traits, including root life span, and 2) to examine the association of root porosity with economics-related root traits in a regional wetland RES. To capture interspeci c variation in root life-span, we chose species based on their contrasting root overwintering habits. Wetland monocots in Northern Ontario with a strongly seasonal climate show two distinct root overwintering habits; they either have roots that all senesce in late autumn, or they have roots that mostly persist to the next growing season (Nieman et al. 2018;Courchesne et al. 2020). This feature allows us to have a binary measure for root life span: short-lived roots with a longevity of one growing season, approximately 5 months at most, or long-lived roots with a longevity of one year, at least. This directly addresses the constraints on RES posed by the trade-off between construction cost and longevity.
Besides the interspeci c variation, root traits also vary within a given root system (Rose 2017;McCormack et al. 2017). Aerenchyma sharply decreases towards the root apex (Armstrong 1980), implying that basal roots are more porous than lateral roots. Therefore, for a comprehensive view of root trait relationships, we investigated the above-mentioned relations separately for basal and for lateral roots. Furthermore, given the various functional associations of root architecture at the root system level (Freschet et al. 2020b), and its phenotypic response to anoxic soil (Pedersen et al. 2021), we characterized root architecture by the ratio of lateral root length to basal root length.

Species selection
For this study, sixteen monocotyledonous plant species from Northern Ontario wetlands were selected (Table 1). These species represent two different root overwintering habits. Roots of ten of the species overwinter, and roots of six of the species completely senesce in late autumn (Nieman et al. 2018). Given the differences in root overwintering habit, roots of these species can be considered to be either shortlived (<5 months) or long-lived (>1 year). Trait measurement Seventy-one plants from sixteen species in total were harvested, of which eight species were harvested in September 2019 and eight species in August 2020. For each species, there were three to ve replicate plants (Table S2, in Supplementary Information). For each individual plant, ten to twenty root segments consisting of basal root and the connected lateral roots were sampled from a depth of 10 to 20 cm for the taller pots, and 7 to 13 cm for the shallower pots. Basal roots and lateral roots were separated using a razor blade while oated in a water-lled dish. To calculate root porosity and root economics traits, the following variables were measured: root length, root fresh mass (m fresh ), root dry mass (m dry ), total root volume (V root ), and the air-excluded root volume (V air-excluded ) ( Table 2). All variables were measured separately for basal and lateral roots.
The m fresh was weighed with a microbalance (MX5; Mettler-Toledo, Greifensee, Switzerland), immediately after removing root surface water by carefully blotting with paper tissues (Visser and Bögemann, 2003). The m dry was measured after drying in the oven for 48 h at 70℃. Root length was determined using the grid-intersection method (Newman 1966;Tennant 1975).
The V root and V air-excluded were measured with the pycnometer method (Vernescu and Ryser 2009). This method is a kind of Archimedes' method as described in Birouste et al. (2014), the most direct measurement to V root based on displacement of water by submersed roots. To be speci c, we measured the mass of the pycnometer lled with water (m 1 ), lled with water and the roots (m 2 ), and lled with water and roots after a vacuum treatment at about 7 kPa three times for 5 mins in a desiccator (m 3 ).
During the vacuum treatment roots were kept submerged with a weighted mesh, resulting in the vacuumed intercellular space of roots to be lled by water when the air pressure was released. The absence of gas bubbles on the surface of vacuumed roots con rmed the e ciency of the air evacuation.
Hence, V root and V air-excluded can be calculated with the equations: V root = (m 1 -m 2 + m fresh )/ρ and V airexcluded = (m 1 -m 3 + m fresh )/ρ, respectively, in which ρ is the density of water, 1.00 g cm -3 at 25°C.
In addition, to test the values of V root obtained with the pycnometer method, diameter of 100 randomly sampled positions along a root system was measured for eight species using a microscope with an ocular micrometer and the root volume calculated assuming a cylindrical form (Ryser and Lambers, 1995). The strong correlation between the two volume measurements validated the pycnometer method as a reliable way to measure V root for basal roots (R 2 =0.78) and for lateral roots (R 2 =0.73).

Calculated traits
Based on the variables measured, root porosity, speci c root length (SRL), root average cross-sectional area (RCSA) and root density-associated traits were calculated ( Table 2). Root tissue is physically composed of three phases: solid, liquid and air (Roderick et al. 1999a); root cellular tissue mainly contains the solid and liquid phase but root aerenchyma mainly the air phases. Density of root tissue can be calculated as the ratio of mass and volume, based on different combinations of root phases.  Table 2).
We also calculated the lateral to basal root length ratio, as a trait describing root architecture in terms of branching density and elongation of the lower-order roots (

Statistical analysis
Most statistical analyses were conducted using average values of each of the 16 species. A two-way ANOVA was used to test the interaction between root order (basal/lateral roots) and root life span (shortlived/long-lived roots) on root traits. The difference between RTD and RTDA was examined with a paired t-test separately for basal and lateral roots. Pairwise trait relationships were assessed using Pearson's correlations for basal and lateral roots respectively. The dominant dimensions of the trait space at the root system level were analyzed with a Principal Component Analysis (PCA), including root porosity and key root economics traits of both basal and lateral roots, as well as the lateral to basal root length ratio. From all measurement data of basal and lateral roots, major axis regression were performed using individual plant values for assessing the relationship of RDMC as proxy to RTDA or RTD (Warton et al.

Results
Trait differences and correlations between basal and lateral roots Across the 16 studied wetland species, aerenchyma occupied on average 44% of the root volume in the basal roots, and 13% of the lateral roots (Tables 3, 4). Porosities of basal and lateral roots did not correlate with each other (Table 5). With a CV of 56% the interspeci c variation of root porosity was larger for lateral roots than for basal roots with a CV of 21.5% (Table S3, in Supplementary Information).
In traits re ecting root density, RTDA, RDMC and fresh root cellular density did not differ between basal and lateral roots, but RTD was 41% lower in basal roots than in lateral roots (Tables 3, 4). Unsurprisingly, lateral roots were signi cantly thinner than basal roots, but basal and lateral root RCSA did not correlate among the species. In contrast, species averages of RTD, RTDA and RDMC correlated between basal and lateral roots (Table 5). Fresh root cellular density varied only little, ranging for all species and roots between 1.01 and 1.15 g cm -3 (Table S3, in Supplementary Information).
Trait differences between long-lived and short-lived roots Root porosity did not differ between species with contrasting root longevities (Tables 3, 4). In contrast, traits re ecting root density (RTD, RTDA, fresh root cellular density and RDMC) were signi cantly lower and SRL higher in species with short-lived roots compared to species with long-lived roots (Tables 3, 4).
RCSA was smaller in species with short-lived roots than in those with long-lived roots, but the difference was only weakly signi cant ( Table 4). The interaction between root order (basal/lateral) and root life span (short-lived/long-lived) was non-signi cant for all root traits, indicating that trait differences between long-lived roots and short-lived roots were not affected by the root order (Table 4).
In uence of root aerenchyma on measures of root tissue density (RTDA vs. RTD).
Root air content causes RTD to be an underestimation of the production costs per unit root tissue, in basal roots by 44% and in lateral roots by 13% ( Fig. 1; Table 3); these differences are signi cant (student t-test; P<0.05). As a result, although both RTD and RTDA were on average different for species with the different root life spans, this difference was more consistent for RTDA (Fig. 1). In case of RTD, the threshold separating the species with short-lived and long-lived roots was different for basal and lateral roots, whereas for RTDA the threshold was consistent (0.15 g cm -3 ) (Fig. 1b). In addition, RTDA was more closely correlated with SRL than RTD was, both in basal roots as well as in lateral roots (Table 5).

Trait correlations
Root porosity did not show statistically signi cant relationships with root traits related to resource economics, neither in basal nor in lateral roots (Table 5). On the other hand, traits re ecting root density (RTD, RTDA, fresh root cellular density and RDMC) were all positively correlated with each other and negatively with SRL in both root orders. Additionally, RCSA correlated with SRL, both in basal and in lateral roots (Table 5).

Multidimensional 'root adaptive space' of wetland plants
To compose a trait variation space of root systems, a PCA was conducted using nine traits of basal and lateral roots of the 16 studied wetland species with contrasting root life spans. The included traits were root porosity, root economics-related traits (SRL, RCSA and RTDA) and lateral to basal root length ratio as a root architectural trait. The rst dimension (46% of the total variation) was mainly represented by the covariation of RTDA and SRL, in which species with short-lived roots and long-lived roots were clearly clustered to opposite ends along this axis (Fig. 2). The second dimension (22% of the total variation) was dominated by lateral root porosity and the lateral to basal root length ratio, two variables that showed a negative trend with each other (Table 5). In addition, basal root porosity dominates the third dimension (12% of the total variation). The loading value of RCSAs of basal roots and lateral roots were not speci c to any dimension (Table S4, in Supplementary Information).
RDMC as the proxy of RTDA Among the 140 paired measurements for RTDA and RDMC, there was a strong positive correlation between these two variables, both for basal and lateral roots (Fig. 3). Their numeric values were almost identical with an R 2 of 0.98 in the linear correlation (Fig. 3), RDMC less than RTDA by 6% on average ( Table 3). The relationship of RDMC with RTD also showed a signi cant positive correlation, but with a lower correlation coe cient than the relationship with RTDA, especially for basal roots (R 2 =0.83; Fig. S1, in Supplementary Information).

Discussion
Our results demonstrate that interspeci c variation in root porosity, an adaptation to oxygen-de cient soils (Justin and Armstrong 1987) is largely independent of root economics traits such as tissue density and root longevity (Ryser 1996 RES in wetland plants has rarely been addressed (Moor et al. 2017). Among the wetland plants of the present study, the rst dimension of the root trait space was de ned by traits related to economics, both of basal and lateral roots, supporting an extrapolation of RES, as described by Roumet et al. (2016), into wetlands. The second dimension was determined by lateral root porosity, accompanied by root architecture expressed as the ratio of lateral to basal root length. A relationship between root porosity and root architecture to match the oxygen-conducting structure with oxygen requirements has been postulated by Pedersen et al. (2021). Hence, this dimension seems to be associated with tolerance to oxygen de ciency. Interestingly, there was no signi cant correlation between lateral and basal root porosities, the latter one being dominant in the third PCA dimension, and varying within a relatively narrow range of high values. Large aerenchyma is required in axial roots to conduct air to lateral roots (Armstrong 1980), but the porosity is constrained by requirements for mechanical strength (Striker et al. 2007).
Root diameter, often considered an important trait for root economics, did not show a clear functional association among the studied wetland plants: Variation in RCSA contributed to several dimensions of the trait space, but in a different manner for basal and lateral roots, and with no crucial role on any of the dimensions. Constraints of root diameter may be order-speci c, with independent involvement of the functions of cortical parenchyma, aerenchyma, and stele (Yamauchi et al. 2021), and related to plant size (Ryser 1998) and temperature (Weemstra et al. 2021). Furthermore, the size of cortex is known to be associated with the degree of mycorrhizal infection, which has been suggested to form its own dimension in root economics space (Bergmann et al. 2020). We did not investigate this aspect, but published literature indicates that mycorrhizal infection in submerged wetland plants, especially among species of Cyperaceae is mostly absent or very weak (Peat and Fitter 1993;Miller and Bever 1999). Eissenstat et al. (2000) and Bouma et al. (2003) point out that aerenchyma formation in wetland plants may confound relationships between tissue density and root life span. The observed lack of correlation between root porosity and root economics shows that this concern is real and that an exclusion of root air space from calculation of root tissue density improves our understanding of the relationships among the economics traits of aerenchymatous roots. Not only does a large interspeci c variation in root air space blur relations between the structural costs of root tissue and the species' physiology and ecology, but also, the difference in root porosity between basal and lateral roots creates a difference in RTD between these root orders, whereas RTDA is independent of the order. The latter comparison seems to be more meaningful, given the modular production and senescence of the root systems of herbaceous species, i.e., a basal root produces only one cohort of lateral roots (Sun et al. 2016). Consequently, minimal life-span differences between basal and lateral roots was observed for Ontario wetland monocots (Nieman et al. 2018). Hence, we propose RTDA, tissue density excluding air space, as the functionally more meaningful variable in context of root economics.
Direct determination of root volume by the buoyancy-based method is laborious and time-consuming (Biroueste et al. 2014). Our results con rmed that the more easily measurable fresh mass can accurately approximate root tissue volume (Shipley and Vu 2002; Biroueste et al. 2014), resulting in RDMC being almost identical to RTDA. This is because at least in case of herbaceous tissues the speci c weight of fresh plant tissue, varies within a very limited range, only slightly above the speci c weight of water 1 g cm − 3 (Roderick et al. 1999b). Moreover, because the measurement of fresh mass is insensitive to tissue air content, the fresh mass is for porous roots a better equivalent to tissue volume than root volume is.
Hence, we suggest that RDMC is preferable to RTD to quantify the actual root tissue structural costs. Nonetheless, also fresh mass measurement requires great care, with a delicate balance between removal of surface water, and avoiding tissue dehydration due to evaporation (Freschet et al. 2020b).
Our data on interspeci c variation of root traits in herbaceous wetland species within a region shows a multidimensional trait space with the main adaptation to oxygen-de cient wetland soils, root aerenchyma, being independent of traits related to economics spectrum. As aerenchyma can occupy more than half of the root volume, for a representative assessment of trait-based wetland root economics it is advisable to exclude this air space from calculation of tissue density. This can be done by using RTDA or RDMC, traits that show a close association with root life span, irrespectively of root order. Other root traits, such as porosity and diameter vary independently for basal and lateral roots, indicating that constraints on those traits depend on root order.     An equivalent for R 2 was calculated as the sum of the effect in proportion to the total sum of squares. SRL, speci c root length; RCSA, root cross-sectional area; RTD, root tissue density (air-included); RTDA, root tissue density air-excluded; RDMC, root dry matter content. *** P<0.001, * P < 0.05 Correlations were signi cant at: **, P < 0.01; *, P < 0.05; +, P < 0. 1. SRL, speci c root length; RCSA, mean root cross-sectional area; RTD, root tissue density; RTDA, root tissue density air-excluded; RDMC, root dry matter content. Figure 1 The variation pattern in (a) root tissue density (RTD) and (b) root tissue density air-excluded (RTDA) of basal and lateral roots among 16 plant species grouped by their root life span, i.e., species with long-lived roots (root longevity more than one year; n=10) or species with short-lived roots (only one growing season of about 5 months; n=6). Each dot represents a species mean value with 3-5 replicate plants.