The role of imperforate tracheary elements and narrow vessels in wood capacitance of angiosperm trees

Summary – There is a broad diversity of imperforate tracheary elements (ITEs) — libriform fibers, fiber-tracheids, true tracheids andvasicentric/vasculartracheids—describedthoroughlybySherwinCarlquist.However,inaquantitativesense,thefunctional meaningofdifferentITEtypespresentinthewoodofvessel-bearingangiospermsremainsunclearbecauseveryfewstructure– functionstudiesmeasureITEs’properties.ITEswithabundantpitsandwidepitborders—vasculartracheids,vasicentric tracheids,andtruetracheids sensu Carlquist — have been shown to conduct water and, thanks to this conductive ability and the multitude of pits, they could also contribute to wood capacitance. A dataset of 30 temperate angiosperm tree species was reanalysed to record the presence/absence of true, vasicentric, and vascular tracheids including data on conduits 15 fraction and vessel-conduit 15 contact fraction (conduits 15 were defined as cells resembling vessels and with a maximum lumen diameter of 15 μm. They encompassed narrow vessels, vasicentric tracheids, and vessel tails). The presence of tracheids, conduits 15 fraction, and contact fraction had no effect on wood capacitance, except, per given wood volumetric lumen water content, species with true tracheids tended to have lower capacitance. These results suggest that the presence of tracheids or conduits 15 properties do not limit wood capacitance, but the results do not exclude the potential role these cells may play in internal water dynamics


Introduction
Properties of wood anatomical structure underpin tree functions, for instance, vessel diameter determines water conductivity (Tyree & Zimmermann 2013).Capacitance -the amount of water released from storage into the transpiration stream -is considered to be an important component of tree hydraulic strategies because it allows stems to buffer transient imbalance between water supply and demand on a daily (Goldstein et al. 1998;Meinzer et al. 2004;Scholz et al. 2007;Carrasco et al. 2015) and seasonal basis (Hao et al. 2013).Capacitance might also aid in drought coping strategies but this remains to be determined experimentally (Hölttä et al. 2009).It has been commonly assumed that wood parenchyma is a primary capacitance reservoir (Borchert & Pockman 2005;Scholz et al. 2011;Morris et al. 2016;Pratt & Jacobsen 2017;Nardini et al. 2018) but more recent studies showed a lack of or weak negative (opposite to the expected) correlation between parenchyma fraction and capacitance (Pratt et al. 2007(Pratt et al. , 2021;;Jupa et al. 2016;Fu et al. 2019;Ziemińska et al. 2020), suggesting that parenchyma abundance does not limit capacitance.So what anatomical properties are the primary determinants of wood capacitance?
Most insightful and progressive structure-function studies rely on careful anatomical observations.Sherwin Carlquist was a champion of scrupulous and detailed descriptions of plant anatomical structure in diverse species, demonstrated in his multitude of papers and compiled in his book (Carlquist 2001).Carlquist used these observations and other authors' physiological work to develop structure-function hypotheses.Capitalising on Carlquist's detailed descriptions of anatomical structure the following questions can be asked: (1) what did Carlquist say about the anatomy-capacitance link? ( 2) what is currently known about this link?(3) what Carlquist's other observations could be relevant to the anatomy-capacitance link, particularly, observations of angiosperm tracheids, and (4) the subsequent hypotheses can be tested across 30 angiosperm tree species (Ziemińska et al. 2020).
What did Sherwin Carlquist say about the anatomy-capacitance relationship?
Carlquist uses the term "water storage" meaning water stored in cells that could be utilized for transpiration or some form of water stress survival strategy.Here, it is referred to as "capacitance" because this term is used in current literature and, more precisely, it indicates the amount of water released from a given wood volume per water potential change.It needs to be noted that Carlquist's perspective encompasses a wide range of species from cacti and perennial herbs to woody shrubs, and emergent trees, and across both gymnosperms and angiosperms.However, this work focuses on wood capacitance in vessel-bearing angiosperm shrubs and trees, primarily because these organisms are crucial components of forest ecosystems worldwide and understanding capacitance may improve our predictions of forest responses to drought, and, also, angiosperm shrubs and trees have been the focus of recent literature on capacitance.
Carlquist identified several cell types that could store and release water: axial and ray parenchyma; wide, thinwalled libriform fibers; living fibers (when devoid of starch); and the gelatinous layer in tension wood fibers (Carlquist 2012b(Carlquist , 2015(Carlquist , 2018)).Higher capacitance is expected in species with abundant axial and ray parenchyma, large diameter cells, thin cell walls, and more isodiametric cell shape (particularly, ray cell, but not applicable to fibers).In vesselbearing woods, these parenchyma cells would be devoid of cell content such as starch and less densely pitted than cells whose primary role is photosynthate storage and transport.Carlquist recognized, however, that water content in stems of "typical" woody angiosperms is relatively small in comparison with succulent woody angiosperms, and that capacitance strategies may vary across species and growth forms.That is, in some species axial parenchyma may play an important role, in others -ray parenchyma (Carlquist 2015(Carlquist , 2018)).
What do we currently know about the wood anatomy-capacitance link?
In general, experimental studies linking wood anatomy with capacitance in angiosperm shrubs and trees are not common and, together, cover 78 species (Pratt et al. 2007(Pratt et al. , 2021;;Jupa et al. 2016;Fu et al. 2019;Ziemińska et al. 2020).Contrary to Carlquist's hypothesis, a fraction of axial and/or ray parenchyma was not correlated or was negatively correlated with capacitance (Pratt et al. 2007(Pratt et al. , 2021;;Jupa et al. 2016;Fu et al. 2019;Ziemińska et al. 2020).However, these studies encompassed the lower half of the worldwide parenchyma fraction variation in angiosperm trees (Morris et al. 2016).So, perhaps it is not implausible that in species with more abundant parenchyma (up to 0.9 fraction in Baobabs, and up to 0.7 in other tropical trees), parenchyma may indeed contribute capacitance water.Instead of parenchyma, vessel lumen fraction showed a positive, although weak, correlation with capacitance (Fu et al. 2019;Ziemińska et al. 2020;Pratt et al. 2021) suggesting that perhaps cavitating vessels may partake in capacitance to a larger extent than parenchyma fraction (Hölttä et al. 2009;Vergeynst et al. 2015;Knipfer et al. 2019;Yazaki et al. 2020).In terms of the size of water storage cells, Jupa et al. (2016) found a positive relationship between fiber/tracheid lumen area and capacitance across four angiosperms and one gymnosperm species (and across both branches and roots).In principle, the size of the cell lumen should influence capillary water release as the larger the lumen the smaller the capillary tension (Tyree & Yang 1990;Hölttä et al. 2009).The hypothesis linking cell size and capacitance is therefore plausible, at least for capillary water, but remains to be tested on a larger species set and other cell types (e.g., parenchyma cells, living fibers, and tracheids).(Jupa et al. 2016) also found that axial parenchyma double wall thickness (thickness of two adjacent parenchyma cells) was negatively correlated with capacitance, supporting the Carlquist hypothesis.Yet, similar to the previous hypothesis, it needs to be tested on a larger number of species.It is not clear, however, how high parenchyma wall elasticity (presumed to be associated with thinner walls) would contribute to capacitance.Cells in wood are densely packed and joined by a middle lamella, so shrinkage of living cells due to water release would need to occur concurrently with changes of volume in surrounding tissues (Holbrook 1995).This synchronous shrinkage, higher in more elastic stems, may indeed be the strongest driver of stem capacitance, and not traits of any individual tissue (but see earlier about vessel fraction).This hypothesis is supported by frequent reports that capacitance is negatively correlated with wood density, more strongly than with tissue fractions, in nature (Wolfe & Kursar 2015;Li et al. 2018;Ziemińska et al. 2020) and laboratory observations (Meinzer et al. 2003;Richards et al. 2014;Jupa et al. 2016;Pratt et al. 2021).Taking into account that stems indeed can shrink/expand on a daily basis (Irvine & Grace 1997;Scholz et al. 2008;Sevanto et al. 2011;Lintunen et al. 2017;Hölttä et al. 2018), it is reasonable to suggest that capacitance is not limited by tissue fractions but rather by the emergent property of wood structure, i.e., wood density.The amount of contact between vessels and other cell types may have an additional but minor effect on capacitance through the facilitation of horizontal movement of water between the different cell types (Ziemińska et al. 2020).
A lack of or negative relationship between parenchyma fraction and capacitance in the studies conducted to date, however, does not exclude a hypothesis that parenchyma may play a role in capacitance.It has been suggested that parenchyma generates osmotic gradients which could promote the refilling of embolized vessels (Secchi & Zwieniecki 2011;Brodersen et al. 2013b;Knipfer et al. 2016;Pagliarani et al. 2019).Perhaps a similar mechanism could be utilized for capacitance water movement between different cells.This hypothesis is also supported by the finding that more vessel-axial parenchyma contact fraction contributed to higher capacitance (Ziemińska et al. 2020).Alternatively, or in addition, ray parenchyma may transport water from bark or pith via rays (Goldstein et al. 1984;Cochard et al. 2001;Pfautsch et al. 2015a, b;Mason Earles et al. 2016).The contribution of pith and bark to stem capacitance remains to be quantified.
In parallel with the structure-function correlative studies, in vivo visualizations using high-resolution computed tomography (microCT), nuclear magnetic resonance (NMR), and cryo-scanning electron microscopy (cryo-SEM) provide unique insights into the movement and spatial distribution of water in the wood.However, only a handful of species have been examined thus far and so there is no consensus on the generality of the evidence.MicroCT and MNR studies showed that in samples (seedlings or cut shoots) experiencing decreasing water potential (more negative) water was released, first, from embolizing vessels and concurrently or afterward -from surrounding imperforate tracheary elements (ITEs) -a cell group encompassing fibers and vascular/vasicentric tracheids (see below for definitions); or ITEs did not empty at all or were already empty before the experiment (Fukuda et al. 2015;Umebayashi et al. 2016;Knipfer et al. 2017Knipfer et al. , 2019;;Yazaki et al. 2020;Baer et al. 2021).This evidence suggests that: (1) capacitance water comes simultaneously from both vessels and ITEs on a daily basis (if it can be refilled overnight) and/or (2) water stored outside vessels is not used for quick, daily release and prevention of embolism in conduits but, instead, may be used during drought or gradually, throughout the seasons.The first idea aligns with the finding of a positive correlation between vessel fraction and capacitance discussed above and is supported by a modeling study too (Hölttä et al. 2009).Additionally, the examination of capacitance and conductivity loss curves in Frangula californica (Eschsch.)A.Gray (Rhamnaceae) showed that considerable cavitation occurs in the initial phases of water release (Pratt & Jacobsen 2017) further supporting the idea that capacitance water comes primarily from vessels.However, this is in contrast with several other studies which show water present in vessels and absent in surrounding ITEs in samples in their native state (without experimental drought) (Utsumi et al. 1996(Utsumi et al. , 1998;;Fukuda et al. 2015;Yazaki et al. 2015Yazaki et al. , 2019;;Umebayashi et al. 2016;Ogasa et al. 2019).Sap flow studies also provide contrary evidence.A comparison of sap flow in branches and trunks showed that trunks lagged behind the branches at the beginning of the day.Sap flow evened up around midday and then branch sap flow decreased in comparison to trunks (Goldstein et al. 1998;Carrasco et al. 2015).This higher, initial sap flow in branches supposedly occurred due to water being drawn from storage outside vessels.Otherwise, embolized vessels would presumably cut off the water supply from trunks.Overall, the several in vivo visualizations, cryo-SEM, and sap flow studies seem to be in contradiction as to where capacitance water comes from on a diurnal time scale.The idea that stored water is used on a seasonal basis is supported by the measurement of water content changes throughout seasons (Hao et al. 2013) and by cryo-SEM and dye perfusion studies, showing diverse water distribution patterns within growth rings (Utsumi et al. 1996(Utsumi et al. , 1998;;Umebayashi et al. 2007Umebayashi et al. , 2008Umebayashi et al. , 2010Umebayashi et al. , 2016)).For example, in Fraxinus mandshurica Rupr.(Oleaceae) vessels remained water filled throughout the growing season, while fibers gradually emptied starting from earlywood and progressing into latewood (Utsumi et al. 1996).The authors suggested that emptying fibers were due to fiber maturation, but that does not exclude the possibility that water left from disintegrating cytoplasm could be used for seasonal capacitance.
Altogether, the evidence suggests that capacitance water in wood originates from dead cells -vessels and possibly ITEs, but there seems to be a disparity as to which one of these is the main source of capacitance water and at what time scales they would be operating.There is one anatomical feature, however, that has not been quantified by any (with one exception) of the above-mentioned work and which may aid in clarifying these disparities, namely, the type of ITE.
What other anatomical structures described by Carlquist could play a role in wood capacitance?
The wood anatomical diversity associated with ITEs is large and described in detail, although not quantitatively, by Carlquist.He defined ITE as "a cell with a secondary wall, derived from a fusiform cambial initial (in secondary xylem; derived from procambium in primary xylem) that neither has perforations (or a single perforation) nor is subdivided into a strand of cells each surrounded by a secondary wall.The last item in this definition permits strand parenchyma to be distinguished from septate fibers" (Carlquist 2001).An additional diagnostic feature implemented by Carlquist which helps to tell apart ITEs from non-stranded axial parenchyma is that ITEs typically elongate intrusively beyond the length of the fusiform cambial initial they originated from, in contrast to axial parenchyma cells, which do not elongate intrusively.
In short, ITEs encompass fibers and tracheids.More specifically, Carlquist recognized five ITE types (Carlquist 1986a(Carlquist , b, 2001)): (1) vasicentric tracheids resemble small vessels but have no perforation plate; they only occur in adjoining vessels, their pit border diameter and pit density are similar to those in vessels, (2) vascular tracheids are anatomically the same as vasicentric tracheids but occur at the end of growth ring (in species with growth rings present) and do not adjoin vessels (but, for a discrepancy, see Fig. 6 in Carlquist (2012a) which shows vascular tracheids in contact with small vessels), (3) true tracheids make up ground tissue (the bulk of ITEs outside vessels, irrespective of growth ring presence/absence) and have pit border diameter and frequency similar to those in vessels, (4) fiber-tracheids-ground tissue ITEs which either have infrequent pits with border diameter similar to that of vessels or small-border pits and (5) libriform fibers with simple pits.It needs to be noted that the literature recognizes different definitions of ITEs (Baas 1986;Carlquist 1986a, b;IAWA Committee 1989;Olson et al. 2020).The most commonly used terminology follows the IAWA list of microscopic features for hardwood identification (IAWA Committee 1989).According to IAWA terminology, vascular and vasicentric tracheids are imperforate tracheary elements with pits resembling that of small vessels, where vascular tracheids intergrade with narrow vessels while vasicentric tracheids surround the vessel circumference.Next, there are ground tissue fibers: (1) with simple to minutely bordered pits (border diameter <3 μm), (2) with distinctly bordered pits (border diameter >3 μm), (3) with pits common in both radial and tangential walls.In the present study, Carlquist's classification is applied for the following reasons: (1) measurements of pit properties were not available and (2) the functional meaning for the IAWA's border diameter cut-off at 3 μm is unclear and at odds with a previous study which found that conductive ITEs had a pit diameter of at least 4 μm (Sano et al. 2011).In the author's view, neither of these classifications is perfect because they are subject to either observer bias or arbitrarily chosen numerical thresholds which may not be functionally relevant.Ideally, ITE types would be replaced by continuous variables (e.g., pit border diameter, pit density) and linked with the measurement of a function, as utilized by (Sano et al. 2011).This approach should be a goal for future studies.Nevertheless, in the absence of unbiased, quantitative data, relying on categories is a reasonable starting point.

4
IAWA Journal Because of frequent and large pits, Carlquist suggested that tracheids (true, vasicentric and vascular) play a part in water transport, particularly as substitute conductive tissue when vessels embolize (Carlquist 1980(Carlquist , 1984(Carlquist , 1985(Carlquist , 1988(Carlquist , 2001;;Carlquist & Hoekman 1985).While fiber-tracheids and libriform fibers would primarily contribute to mechanical strength.The functional meaning of the different ITEs, however, has not been quantified because the overwhelming majority of current structure-function studies group all ITEs -particularly true tracheids, fibertracheids, and libriform fibers -into one category of "fibers".Living vs dead ITEs (fiber-tracheids or libriform fibers) are typically not identified either, despite their very different roles: living fibers may store starch, and dead fibers may not.
Limited work carried out on ITE types showed that tracheids (vascular, vasicentric and true) can conduct water, but their contribution to the overall conductivity in vessel-bearing angiosperms is minimal (<2.2% of total conductivity) at least in the handful of species studied (Sano et al. 2011;Cai et al. 2014;Pan & Tyree 2019), suggesting that the tracheid bridges, connections between vessels via tracheids, could play a role in the sideways movement of water between vessels and perhaps decrease the probability of air seeding and vessel embolism spread (by separating vessels from each other).Based on these reports, it can be hypothesized that due to tracheids' capacity for sideways water transport, they may play a role in water capacitance by releasing water into the transpiration stream.Very narrow and short vessels could potentially contribute to the sideways water movement, too (Brodersen et al. 2013a).To test this hypothesis, the dataset of anatomical and capacitance traits for 30 angiosperm tree species was re-examined (Ziemińska et al. 2020).Specifically, the following question was addressed, do species with tracheids and/or abundant narrow vessels have higher capacitance?

Materials and methods
Transverse and longitudinal sections of the 30 species studied in Ziemińska et al. (2020) were re-analysed and the presence/absence of true tracheids (TT) and vasicentric tracheids following Carlquist's definitions (see above, Carlquist 2002) were recorded.
The presence/absence of true tracheids (TT) and vasicentric tracheids following Carlquist's definitions (see above, Carlquist 2001) were examined.Tracheids adjoining vessels were classified as vasicentric tracheids.In several cases, vasicentric tracheids were at the end of a radial vessel multiple in latewood.Vascular tracheids -not adjoining vessels -were not observed in any of the studied species.While true tracheids were relatively easy to identify (Fig. 1), it was more difficult to unequivocally confirm the presence/absence of vasicentric tracheids because the presence/absence of perforation plate can only be reliably identified on macerations.Hence, additional criteria were developed for these tracheids: (1) wall thickness similar to a nearby vessel and (2) outer circumference similar to or smaller than the circumference of the nearest ground tissue ITEs from the same growth ring (fiber-tracheids or libriform fibers).In reality, these cells could be either tracheids or small vessels and are therefore referred to as vasicentric tracheids/vessels (VTV).In addition to these categorical variables, data for the conduits 15 fraction and contact fraction between vessels and conduits 15 (the proportion of vessel circumference in contact with conduits 15 called hereafter vessel-conduits 15 contact fraction) were reanalysed.Conduits 15 were defined as conduits that were either narrow vessels, vessel tails, or vasicentric tracheids (based on their large pit border and high pit density), and with maximum diameter <15 μm (hence conduit 15 ).Although this data was included in (Ziemińska et al. 2020), their relationship to capacitance was not described in that work.
Variance homogeneity and normality in the four groups (TTs present, TTs absent, VTVs present, and VTVs absent) were tested using Leven's and Shapiro-Wilk tests, respectively (α = 0.05).Based on these results, t-tests were utilized to compare capacitance between the groups.Next, multiple regression models were performed.The independent variables in the models included: tracheid presence (TTs and VTVs), conduits 15 fraction, vessel-conduits 15 contact fraction, wood density (WD), and wood lumen volumetric water content at predawn (VWC L-pd ).WD and VWC L-pd were used here because they were the strongest predictors of capacitance in Ziemińska et al. (2020).The structure of the models and results are presented in Table 2.All statistical analyses excluded Paulownia tomentosa which was an outlier in capacitance value.It did not have TTs, VTVs or conduts 15 .All analyses were done in R (R Core Team 2018).
The measurements were carried out in the peak summer of 2017 in the Arnold Arboretum of Harvard University, Boston, MA, USA.All sampled species were temperate, winter-deciduous, vessel-bearing angiosperm trees.Three individuals per species were sampled and data analysis was run on species average values.All measurements were done on approx.5 mm wood diameter twigs (excluding bark and pith).In contrast to most wood capacitance studies, the capacitance was measured in nature and calculated as the amount of water released from wood tissues per change in stem water potential between predawn and midday (kg/m 3 per MPa) referred to as day capacitance.Detailed information about the study location, climate, material, and methodology for anatomy, water content, and capacitance measurements is described in Ziemińska et al. (2020).

Results
Seven out of 30 species had TTs (23%) and 20 species had VTVs (77%, Table 1).Four species had both TTs and VTVs and seven species had neither of these cell types.In comparison, in Californian flora, TTs were present in 24% of the analyzed species (similar to the present study), vasicentric tracheids in 33%, vascular tracheids in 13%, and 30% had none of these ITEs (Rosell et al. 2007).According to InsideWood descriptions (InsideWood 2004;Wheeler 2011), only four species from my species set had vascular/vasicentric tracheids (description for one species was absent in InsideWood).This disparity in vasicentric tracheid presence between the present and other records implies that the tracheid circumference size criterion for VTVs applied in this study was inappropriate and/or InsideWood descriptions misidentified the presence/absence of tracheids.While the Californian dataset represented a warmer and drier climate than the one in North-East USA where the Ziemińska et al. (2020) study was conducted, and the species set was primarily composed of subshrubs (two-thirds) followed by other growth forms (shrubs, trees, herbs, and vines).
There was no difference in mean day capacitance between the species with and without TTs, or with and without VTVs (Fig. 2).Linear regression models (Table 2) showed the presence of TTs or VTVs did not affect day capacitance.Only per given VWC L-pd , species with TTs tended to have lower capacitance (r 2 adj = 0.37, p < 0.01, the TTs presence explained an additional 8% variation in capacitance, Fig. 3).Conduits 15 fraction and vessel-conduits 15 contact fraction were not correlated with day capacitance either in pairwise relationships or as independent variables in the regression models.protect them from embolism.This interpretation, however, is counter to microCT and some cryo-SEM studies which found that ITEs embolize simultaneously or shortly after vessels do (Fukuda et al. 2015;Umebayashi et al. 2016;Knipfer et al. 2017Knipfer et al. , 2019;;Yazaki et al. 2020;Baer et al. 2021).However, thus far these studies encompassed a handful of species only and they did not quantify the type of ITEs and, instead, combined them into a "fibers" category.One study (Yazaki et al. 2020) identified ITE types and found that in species with true tracheids (Cercidyphyllum japonicum, also studied here) vessels embolized as the water potential became more negative, but ITEs did not, which rejects my hypothesis (note that the authors identified ITEs as libriform fibers, while according to my observations, C. japonicum ITEs are composed of TTs and fiber-tracheids).The third possibility is that the TTs were emptied throughout the season and devoid of water by the time the day capacitance was measured, while species without TTs still had reservoirs of stored water.This would align with the evidence showing species-specific seasonal changes in water distribution within a growth ring such as, for instance, emptying of ITEs throughout the growing season (Utsumi et al. 1996(Utsumi et al. , 1998;;Umebayashi et al. 2016).This is also supported by observations that in some species, in their native state water is   present in vessels, and absent in the surrounding ITEs.Altogether, the available evidence does not exclude the role of TTs in capacitance.The major task for the future would be to quantify the type of ITEs present and their properties.
The reason why VTVs were not correlated with day capacitance might be related to the diverse spatial distribution and abundance of these cells across the growth ring.In some species, VTVs surrounded vessels and were present across the entire growth ring (Q.muehlenbergii, see Table 1 for full names).In other ring-porous species, VTVs occurred in groups in latewood (e.g., C. speciosa, C. kentukea, M. pomifera, M. alba, P. amurense, T. danieli), and could also be present in earlywood in contact with vessels (Z.sinica).While in diffuse-porous species, VTVs were either intermixed with earlywood vessels (e.g., F. grandifolia, L. styraciflua, O. arboretum, S. pseudocamellia) or only appeared at the very end of latewood (e.g., B. dahurica, A. saccharum, L. tulipifera, S. obassia, T. japonica).It is possible that the latewood VTVs primarily function as water-conducting cells, while in the earlywood they could contribute to water storage.Yazaki et al. (2020) found that as water potential decreased, vessels and libriform fibers embolized while vasicentric tracheids retained water for longer in Quercus serrata Roxb., Fagaceae.This implies that vasicentric tracheids may chiefly function as an alternative water-conductive pathway.This evidence, together with the observed differences in VTVs distribution and abundance suggest their disparate functional roles and quantitative contribution to water movement in the wood and could explain the lack of relationship between VTVs' presence and day capacitance.Also, as mentioned earlier, the size threshold for VTV identification used in the present study could obscure the possible patterns.Altogether, the present study does not exclude the role of VTVs in capacitance.A promising path forward would be to quantify the different characteristics of VTVs as well as TTs (spatial distribution, abundance) concurrently with seasonal and spatial variation of water content and distribution.This would help clarify the diversity of tree hydraulic strategies and the functional role of ITEs.
The lack of relationships between day capacitance and the presence of tracheids/small vessels (TTs or VTVs) could also be due to the crude nature of the "presence/absence" metric as alluded to in the previous paragraph.Similar to VTVs, variation in TT's spatial distribution, abundance, lumen area, or wall thickness was observed but not quantified.For example, F. grandifolia and C. kousa had abundant TTs and VTVs in earlywood, but were scarcer in latewood while the abundance of fiber-tracheids (ITEs with infrequent bordered pits) increased.The size of the pit border and the density of pits in tracheids are examples of continuous variables that could also influence capacitance, as they did conductivity (Sano et al. 2011).Identifying ITE types and their presence/absence is an essential starting point implemented in the current study.Yet, to clearly identify ITE functions, continuous ITE properties need to be measured, such as border diameter, pit frequency, wall thickness, cell spatial distribution, abundance, and amount of contact between them and other cell types.Because identifying tracheid presence (especially, vascular and vasicentric tracheids because they are so similar to small vessels) on a cross-section is very difficult, establishing some criteria, like for the conduits 15 allows for unbiased quantification of the anatomical structure.Yet, ideally, this criterion would be developed based on observations of cross-and longitudinal sections, and, importantly, on macerations, because they most clearly show the presence/absence of tracheids.Quantification of these characteristics would allow a robust data analysis.And it would also circumvent disagreements in the definitions of the different cell categories and researcher's biases when identifying them.
The continuous variables -conduits 15 fraction and vessel-conduits 15 contact fraction -measured in the current study were not related to capacitance.The reasons might be the same as discussed above for TTs and VTVs.It is possible, that they contribute water to capacitance, but the amount of water is not limited by either their fraction or vessel-tissue contact fraction.The inaccuracy resulting from the grid method might be at play, too.This method counts several grid points that fall in a given tissue.That number divided by the total number of grid points analyzed per area gives tissue fraction.This technique provides reliable results for abundant tissues, but less so for scarce tissues such as conduits 15 .

Conclusions
Based on the present analyses and previous structure-function studies, there is no convincing evidence whether TTs, VTVs, or conduits 15 contribute to day capacitance, but this possibility cannot be excluded either.To answer the question of the functional role of ITEs types in tree hydraulics, future work needs to combine detailed anatomical quantification of vessels and ITEs properties with in vivo (microCT, cryo-SEM, NMR) and physiological observations across variable time scales (daily and seasonal).
The present study was an exercise in understanding the function of ITEs, which are often neglected in current quantitative anatomical studies and merged into a single category of "fibers".Although for some studies, depending on the question, this approach is satisfactory, for others, it simplifies the anatomical structure to the point where we fail to see the structure-function link, while one might really exist.Highly detailed studies of comparative anatomy, like the ones carried out by Carlquist provide a crucial wealth of knowledge on anatomical diversity and can generate better-informed functional hypotheses.The way forward would be to harness the wealth of comparative anatomical descriptions, translate them into numerical variables, and measure them in association with physiological function.

Fig. 1 .
Fig. 1.Examples of true tracheids (a-c) and vasicentric tracheids (d-f) in six species as seen on twig wood cross-sections.The examples of species with vasicentric tracheids shown here were also reported to have tracheids by InsideWood.The smaller photo is a twice-enlarged fragment of the photo above it.The scale bar for the large photos corresponds to 25 μm, and in the small photos to 10 μm.All large photos are at the same magnification and all small photos are at the same magnification.Arrowheads outside the photos indicate a growth ring boundary, with earlywood at the top and latewood at the bottom.Arrowheads in the large photos: true tracheids, vasicentric tracheids, or small vessels.Arrowheads in the small photos: bordered pits.A, Eucommia ulmoides; B, Oxydendrum arboretum; C, Liquidambar styraciflua; D, Zelkova sinica; E, Quercus muehlenbergii; F, Betula dahurica.

Fig. 2 .
Fig.2.Boxplots showing differences in day capacitance between the groups of species with/without true tracheids (a) and with/without vasicentric tracheids (b).Symbols correspond to species mean.Groups were compared using t-tests and the differences were not statistically significant (ns).

Fig. 3 .
Fig. 3. Relationship between day capacitance and volumetric lumen water content at predawn (VWC L-pd ).Symbols correspond to the species' average value and error bars represent one standard deviation.**p < 0.01.

Table 2 .
Models predicting day capacitance, their r 2 adj and p values.