Skip to main content
NCBI home page
Annals of Botany logoLink to Annals of Botany
. 2018 Dec 14;123(5):783–792. doi: 10.1093/aob/mcy209

Disentangling the effects of disturbance, climate and tree age on xylem hydraulic conductivity of Betula pendula

Jan Tumajer 1,✉, Václav Treml 1
PMCID: PMC6526315  PMID: 30551134

Abstract

Background and Aims

The increasing frequency of disturbances in temperate forests is responsible for the greater numbers of trees with mechanically damaged cambial zones. Adjustment of wood anatomical structure to balance between safe and efficient water conductivity is one mechanism trees employ to cope with mechanical damage. The relative role of disturbances, tree age and climate in shaping xylem conduits and affecting xylem hydraulic conductivity remains unknown.

Methods

We performed an experiment with five different mechanical treatments simulating natural disturbances of juvenile Betula pendula trees (stem scarring, tilting, decapitation, root exposure and stem-base burial). After 3 years, trees were cut down, conduit size and density were measured, and specific hydraulic conductivity of each tree ring was calculated. Between-tree and between-year variability in xylem conductivity was decomposed into effects of tree age, climate and disturbances using linear mixed-effects models.

Key Results

Xylem-specific hydraulic conductivity decreased significantly after treatment in decapitated, tilted and scarred trees. In the last treatment, wood anatomical adjustment was restricted to the area next to the callus tissue zone; in contrast, specific hydraulic conductivity declined over the entire stem circumference after tilting or decapitation. The response of trees with buried stems and exposed roots was generally weak. The overall effect of disturbances on inter-annual variability of wood anatomical structure was greater than the contribution of tree age and climate.

Conclusions

The results indicate that disturbances are important drivers of xylem hydraulic conductivity. Expected increases in the frequency and intensity of disturbances may alter the theoretical capacity of forest stands for water conductance with a feedback to climate.

Keywords: Cambial injury; disturbance experiment; dendrochronology; mixed-effects model; quantitative wood anatomy; silver birch; tree-ring, vessel; vessel lumen area

INTRODUCTION

An increasing frequency of forest disturbances has been observed in recent decades and is predicted for the near future in temperate and boreal forests (Schelhaas et al., 2003; Thom and Seidl, 2016; Seidl et al., 2017). Many disturbance agents (e.g. windstorms, landslides, rockfall, floods) mechanically attack tree stems, crowns and roots. To survive, trees have developed mechanisms allowing them to cope with partly damaged segments. Among these, plastic adjustment of conduit size and the proportions of conduits, parenchyma and fibres represents an efficient strategy to maintain mechanical support as well as a safe and efficient vascular system (Sperry et al., 2008; Hacke et al., 2017). The imprint of past disturbances is therefore visible in conduit parameters, allowing dating of disturbance events (Heinrich and Gärtner, 2008; Arbellay et al., 2012b; Copini et al., 2015). As conduits largely affect the capacity of xylem to transport water, disturbances might influence overall forest transpiration and thus feed back into climatic conditions (Matheny et al., 2014). However, the relative influence of disturbing events on xylem conduit traits in comparison with other factors (tree ageing, climate) is poorly understood.

Conduit size increases ontogenetically with increasing height of the stem due to physical constraints (Olson et al., 2014; Carrer et al., 2015) and associated reduction in the polar transport of auxins (Aloni, 2007; Hacke et al., 2017); thus, ontogeny is often believed to explain the largest proportion of variability in mean conduit size (Zhao, 2015; Rosell et al., 2017). However, recent studies have reported conduit size not to be solely a function of tree size, but also to contain a significant climatic signal (Hacke et al., 2017), which may be used as a valuable palaeoenvironmental proxy (Fonti et al., 2009) or as a tool for determination of tree response to climatic constraints (Eilmann et al., 2009; Anderegg et al., 2016; Castagneri et al., 2017). Indeed, climate influences cell size in all the following ways: (1) climatically driven water potential influences the rate of cell expansion via cell turgor (Schreiber et al., 2015); (2) climate during the growing season modifies the length of the expansion period (Cuny et al., 2014); and (3) climate affects the amount of carbohydrates available for xylogenesis in the cambial zone (Simard et al., 2013). Obtaining a meaningful climatic signal from a series of wood anatomical parameters often requires large sample depth and an appropriate standardization approach to filter out the effect of increasing tree size over time (Matisons et al., 2012; Carrer et al., 2015).

Mechanically damaged trees often abruptly adjust conduit size in response to wound-related shifts in phytohormonal concentrations (Aloni, 2007). Disturbances that mechanically damage bark and cambium (e.g. mass-movements, floods) usually lead to decreased conduit size in an area adjacent to damaged tissues, which might be partly compensated for by an increase in conduit number (Arbellay et al., 2012b; Tumajer et al., 2015). These responses may result in a total reduction of up to 60 % of pre-disturbance mean conduit lumen area, overriding both long-term age trends and climatically governed inter-annual variability (Tumajer et al., 2015). A similar decrease in conduit dimension, together with increased cell-wall thickness, occurs after stem tilting due to formation of reaction-wood (Heinrich and Gärtner, 2008; Groover, 2016). In addition, the wood structure of roots exposed due to loss of the soil in an erosion event adjusts to the lack of mechanical support by decreasing conduit size by up to 50 % compared to tree-rings formed in the soil prior to exposure (Gärtner, 2007; Hitz et al., 2008). In contrast, after stem base burial, an abrupt increase in conduit size may occur (Friedman et al., 2005). Precise quantification of the contribution of disturbances to conduit dimensions is, however, complicated, because most of the above-mentioned studies were: (1) focused exclusively on the disturbance signal in tree-rings, while considering age trend and climate as noise; and (2) based on analysis of trees without precisely knowing the timing and extent of disturbance events. Field experiments with trees in controlled conditions have only rarely been performed (Heinrich and Gärtner, 2008; Copini et al., 2015).

Because forest transpiration is extremely important for forest–climate interactions, and disturbances might affect the efficiency of trees in conducting water from roots to crown (Matheny et al., 2014), studies analysing the contribution of tree age, climatic factors and disturbances to variability in conduit size are needed. We aim to fill this gap using data on inter-annual variation of xylem structure of Betula pendula individuals that were experimentally mechanically treated 3 years before sampling. The anatomical structure of tree rings formed before and after the treatment was described using the most common definition of hydraulic efficiency of xylem, namely specific hydraulic conductivity (Gleason et al., 2016). Interannual variability in this parameter was modelled using linear mixed-effects models to quantify the contributions of climate, tree age and disturbances to modelled values. We hypothesize that disturbance is a more important driver of specific hydraulic conductivity than tree age and climatic effects.

MATERIALS AND METHODS

Study area and experimental design

A monospecific unmanaged forest stand of Betula pendula Roth located in north-west Czech Republic (50°32.041′N, 13°41.740′E, 315 m a.s.l.) was selected for this experimental study. The stand originated through natural recolonization of a former spoil heap in a post-mining (brown coal) locality. The mean (±s.d.) tree age estimated by counting tree rings in 2015 was 14.6 (±2.0) years at sampling height. The mean diameter at breast height was 11.8 (±2.4) cm, and mean tree height was 14 (±0.7) m.

Randomly selected trees were mechanically treated in five different ways (Fig. 1) to simulate events frequently appearing as consequences of mass-movements (including stem tilting, decapitation, stem scarring, stem aggradation by deposits, root exposure; Shroder, 1980; Stoffel and Corona, 2014), windstorms (stem tilting, decapitation, stem scarring; Gardiner et al., 2016) or floods (stem tilting, stem scarring, stem aggradation by deposits, root exposure; Ballesteros-Cánovas et al., 2015). The first deformation represented stem tilting to 45° from the vertical. We used 30-cm-long segments of plastic pipes to protect tree stems in places of contact with the rope used for tilting. In this way, pressure from the rope was distributed over a 30-cm-long stem segment. Another group of treated trees was mechanically damaged using a hammer to abrade bark and cambium in an area of approx. 5 × 10 cm on one side of the stem at a height of 100 cm above the ground. In the third treatment group, we decapitated trees at a height of 6 m above the ground, but with some branches remaining below the cut. In the fourth treatment group, basal parts of stems up to a height of approx. 0.8 m were buried in local clayey soil. In the last treatment group, trees were subjected to having their roots exposed. The depth of root exposure balanced between affecting as many roots as possible and retaining sufficient root stability to hold the stem in vertical orientation. Due to the generally shallow root system of B. pendula at our study site, we excavated only the upper 0.5–0.8 m of soil.

Fig. 1.

Fig. 1.

Open in a new tab

Schematic representation of five different experimental treatments, as well as control reference trees (n = replicates per treatment). Dotted lines indicate sampling positions (centimetres above the ground). Colour rectangles over cross-sections indicate positions of individual microsamples prepared from each cross-section.

To determine the effect of the timing of disturbance on wood anatomical anomalies, we performed treatments in two different periods relative to the growing season. Thus, half of the trees were subjected to treatments performed in March 2013 (before the beginning of the growing season) and the other half were treated in May 2014 (during the growing season). The treatments were performed on different sets of trees from areas adjacent to each other within the same forest stand. Each type of treatment was represented by nine (tilting, root exposure) to 13 (decapitation) replicates per treatment; 53 trees in total were treated. Similar numbers of trees were subjected to each treatment in each of the treatment periods.

Trees remained growing in situ for 3 years after the treatment. At the end of the third growing season after the mechanical damage (i.e. November 2015 and 2016), we cut discs (i.e. cross-sections) from all experimentally treated trees. At least two discs were cut from each stem at fixed levels above the ground (Fig. 1). Additionally, four control reference trees (one disc per reference tree) were sampled in 2016.

Laboratory processing and wood-anatomical measurements

Discs were split into three or four segments regularly distributed around the circumference (Fig. 1), and from each segment a 1-cm-wide block spanning the bark to the pith was extracted. If the length of the block exceeded 6 cm, it was split into two or three segments to enable it to be mounted in a sample holder on a Leica RM2125 RTS rotary microtome. The risk of wall deformation during the cutting process was reduced by boiling the sample in water and applying a non-Newtonian fluid to the sample surface (Schneider and Gärtner, 2013). Transverse microsections of 25 µm thickness were cut. Each microsection was rinsed with sodium hypochlorite solution and distilled water, stained with AstraBlue and Safranin solution to enhance wall–lumen contrast, dehydrated using ethanol and Diasolve and, finally, mounted using Canada balsam (Gärtner and Schweingruber, 2013).

Each microsample was scanned using a Nikon Eclipse Ni motorized microscope and NIS-Elements Ar 4.40.00 software (Nikon Corporation). Images were captured at a resolution of 1.7 µm per pixel with 10 % overlap of adjacent pictures, which allowed automatic geometric stitching of the whole microimage. Subsequently, we measured various parameters of vessel lumina (area, diameter, shape index, density and others) for each sample and each tree-ring using WinCell 2011Pro (Régent Instruments, 2011). Measurement was performed in a rectangular area manually delimited to cover the whole tree-ring width. The width of the rectangle varied slightly among tree rings to follow recent advice on the minimum number of measured vessels per tree ring of ring-porous species; thus, care was given to use rectangles wider than 2.5 mm (Diaconu et al., 2017) and to measure more than 60 vessels per tree ring (Arbellay et al., 2012a). To focus solely on the contribution of vessels to the hydraulic structure of xylem, we excluded all objects with lumina smaller than 300 µm2, as most of them probably represent fibres. No manual editing of the microimages was performed; however, vessels with broken walls, which appeared rarely and with similar frequency in all treatments, were excluded from the analysis.

In addition to measuring various parameters of each vessel lumen, we also estimated the total area of the measured segment of xylem in each tree ring. Although WinCell reports the total area of each measured rectangle, this value is strongly biased in juvenile tree rings with small radii, due to frequent overlaps of the rectangles with previous and next tree rings, such that vessels need to be manually excluded from the analysis. Thus, we estimated the precise area in each tree ring by modelling a concave hull around vessel positions using R (R Core Team, 2017) and the packages ‘concaveman’ (Gombin et al., 2017) and ‘sp’ (Bivand et al., 2013).

Specific hydraulic conductivity calculation

The lumen area of each vessel was used to calculate its theoretical hydraulic conductivity [modified from Scholz et al. (2013) and Gebauer and Volařík (2013)], with vessels assumed to be circular in cross-section, using the following equation:

Khi=ρ. Ai28. η .π

where Khi is the theoretical hydraulic conductivity of vessel i (kg.m s−1 MPa−1), ρ is water density at 20 °C (998.205 kg m−3), Ai is lumen area of vessel i (m2) and η is the viscosity of water at 20 °C (1.002.10−9 MPa.s). Theoretical hydraulic conductivity of vessels reflects the Hagen–Poiseuille law regarding the greater contribution of large vessels to water transport than that of small ones (Tyree and Zimmermann, 2002).

Subsequently, Kh was converted to specific hydraulic conductivity of each tree ring using the following equation:

Ks = ∑i=1nKhiAxyl

where Ks is specific hydraulic conductivity of tree-ring xylem (kg m−1 s−1 MPa−1) and Axyl is the measured area of the respective tree ring (Tyree and Zimmermann, 2002). Ks, which represents hydraulic conductivity per unit conductive area across a given pressure gradient (Gleason et al. 2016), tends to increase in tree rings with (1) abundance of large vessels and (2) high vessel density. The above-described calculation of Ks does not consider the effects of vessel grouping (Hacke et al., 2017), vessel axial length (Jacobsen et al., 2012), or ratio of leaf area to xylem area (Gleason et al., 2012). Nevertheless, Ks does represent a useful approximation of whole-plant conductance (Tyree and Zimmermann, 2002). Missing rings identified by visual comparison of tree-ring width curves of sampled trees were incorporated into the dataset with Ks = 0.

Linear mixed-effects models of specific hydraulic conductivity

In the next step, inter-annual variability in Ks was modelled using linear mixed-effects models. Different random and fixed effects were selected to represent parts of the variability in Ks given by (1) tree ageing, (2) climate, (3) disturbances and (4) between-tree variability. More specifically, predictors were as follows:

  1. Tree ageing is represented by the cambial age of each tree ring (fixed effect). Approximation of a long-term trend in vessel dimensions by cambial age was previously shown as appropriate for short (<20–25 years) time series (Zhao, 2015; Anadon-Rosell et al., 2018).

  2. Climate is represented by mean temperature from previous August to previous October, temperature in current May and precipitation in current March (fixed effects). These months and period were previously identified as the only climatic variables with significant (P < 0.05) correlation with measured values of Ks in the pre-experimental period (n = 4254 tree rings). Monthly climatic data from Kopisty climate station (Institute of Atmospheric Physics CAS) located 5 km from our study site at 240 m a.s.l. were used.

  3. The effect of disturbances is represented by four categorical variables (factors; random effects). The first one (phase) indicates whether the tree ring was formed before the experiment (PRE), in the year of the experiment (EY) or later (POST+1 or POST+2). Three different levels were used for each of the three years after the treatment, because we previously observed 3 years of continuous recovery of vessel size of B. pendula after cambial injury caused by rockfall (Tumajer et al., 2015). Next, orientation represents the designation for sample orientation around the stem circumference (A–D). Sampling level indicates position along the stem (High, Low and Middle in the case of injured trees). Finally, timing indicates whether the tree was damaged during the first (March 2013) or the second field season (May 2014). We nested orientation, level and timing within phase, because we expect an increasing variability of specific hydraulic conductivity along (root exhumation, stem burial) and around (scarring, tilting) the stem after the experiment than during the pre-experimental period.

  4. The last predictor captures between-tree variability in specific hydraulic conductivity and is represented by tree identifier used as a random effect.

The above described structure of the linear mixed-effects model can be summarized as follows:

Kst = β0+ β1.CambialAget+β2.Temp(pAUG_pOCT)t+β3.Temp(MAY)t+β4.Prec(MAR)t+U1t.Phaset+ U2t.(Phaset:Orientation)+ U3t.(Phaset:Level)+U4.(Phaset:Experiment)+U5.Tree

where t represents calendar years, β0 represents a fixed intercept, β1–4 represent coefficients of fixed effects and U1–5 represent coefficients of random effects. The model was fit separately but with the same structure for each type of treatment. Model structure was partly modified in the case of reference trees, as all predictors related to disturbance except for orientation were not available and were therefore omitted from the model. Fitting of linear mixed-effects models was performed in R (R Core Team, 2017) using the ‘lme4’ package (Bates et al., 2015).

The quality of the model was quantified using marginal and conditional R2 as proposed by Nakagawa and Schielzeth (2013) and by simple pseudo-R2 defined as a squared correlation between observed and modelled values. The effect of individual predictors on predictability of the model was estimated by quantifying the decrease in the above-defined pseudo-R2 due to omitting that predictor, i.e. single variable or variable groups (all climatic variables, all disturbance-related variables), from the model.

RESULTS

In total, more than 1.3 million vessels were measured, and Ks was calculated for 5796 tree rings (i.e. on average, there were 228 vessels measured per tree-ring segment). There were slightly increasing trends of Ks with age regardless of treatment group, and non-significant differences in mean Ks of trees treated before and during the growing season in the pre-experimental period, with the exception of buried trees in the period 2007–2013 (Fig. 2). The strongest reductions in Ks were observed after the decapitation and in tilting treatments (regardless of sample position along and around the stem; Fig. 3; Supplementary Data Fig. S1) as well as scarred trees (mainly in segments adjacent to the scar at its level; Fig. 3; Supplementary Data Fig. S2). In contrast, no abrupt response in Ks was observed for trees with stem-base burial or root exposure at either sampled height level (Supplementary Data Fig. S3).

Fig. 2.

Fig. 2.

Open in a new tab

Observed (solid line) and modelled (dotted line) specific hydraulic conductivity averaged for each treatment category and treatment timing (before growing season, red; during growing season, blue). Shaded areas represent respective sample depths, and stippled areas represent missing tree rings with Ks = 0. Vertical dotted lines indicate years in which the treatments were performed.

Fig. 3.

Fig. 3.

Open in a new tab

Anatomical responses of vessels to the most serious types of treatments: stem tilting (before growing season), decapitation (during growing season) and stem scarring (during growing season). Triangles denote tree-ring borders, with the red triangle indicating the tree ring formed in the treatment year. Scale bars represent 250 µm. Blue colour of tree rings formed after stem tilting on the upper side of the stem is due to occurrence of gelatinous layers inside lumina of tension-wood cells. Line charts show standardized variability (such that 1 equals mean value of three pre-treatment tree rings) in average vessel lumen area (blue), vessel density (green) and specific hydraulic conductivity (red) of three pre-treatment tree rings and up to three post-treatment tree rings.

An abrupt decline in Ks in the tension-wood side of the tilted stem is driven by a simultaneous decrease in vessel size and density (Fig. 3). However, the decrease in vessel size is partly compensated for by increasing density in samples that are perpendicular to or opposite the scar, resulting in smaller declines of Ks. A similar pattern of increasing density and decreasing size of vessels appears also in scarred and decapitated trees (Fig. 3). There is a close correlation between Ks and average vessel lumen area of the tree ring; in contrast, correlation between Ks and vessel density is not significant (Supplementary Data, Fig. S4).

Decapitation resulted in missing tree rings in 49 % of analysed samples from the year of the treatment, and in 99 % of tree rings 1 year after the treatment. None of the decapitated trees formed a tree ring in the sampled parts of the stem 2 years after the treatment, although trees were still living (extensive sprouting from basal segments of the stem and live branches). Tilted trees exhibited missing tree rings in 7 % of the segments analysed in the year of tilting, and this value increased to 70 % 2 years afterwards. Other treatments did not increase the frequency of missing rings (Fig. 2).

Linear mixed-effects models successfully predicted the long-term trend in Ks during the pre-experimental period as well as abrupt anomalies after the treatment. Graphical comparison of modelled and measured values (Fig. 2) reveals that their concordance tends to be better after the treatment than before. Values of marginal R2 do not exceed 0.23; however, conditional R2 varies between 0.59 and 0.78 for treatment categories (Table 1). The highest correlation between observed and modelled values of Ks was observed in trees with decapitated crowns or tilted stems and the lowest in reference trees. Modelled values of Ks increase with increasing cambial age (on average by 0.71 kg m−1 s−1 MPa−1 per cambial year) and March precipitation (0.01 kg m−1 s−1 MPa−1 per mm) and decrease with both previous August–October temperature and current May temperature (0.68 and 0.81 kg m−1 s−1 MPa−1 per °C, respectively) (Table 2).

Table 1.

Quality assessment statistics of linear mixed-effects models

Treatment Marginal R2 Conditional R2 Pseudo-R2
Scarring 0.209 0.592 0.442
Tilting 0.080 0.777 0.642
Decapitation 0.057 0.778 0.635
Root exposure 0.225 0.673 0.535
Stem-base burial 0.211 0.680 0.593
Reference 0.185 0.369 0.354
Open in a new tab

Table 2.

Coefficients of fixed effects and range of coefficients of random effects estimated by linear mixed-effects models

Treatment Fixed effects Random effects
(Interc.) Cambial age Precip. MAR Temp. pAUG-pOCT Temp. MAY Phase Phase:
Orient.		 Phase:
Level		 Phase:
Timing		 Tree
Scarring 43.267 0.498 0.004 −1.142 −1.272 −0.358
0.242		 −3.060
1.663		 −2.478
2.381		 −3.773
1.867		 −2.578
3. 161
Tilting 25.701 0.461 0.016 −0.763 −1.067 −2.869
8.168		 −0.138
0.208		 0
0		 −3.269
2.650		 −2.493
7.070
Decapitation 21.111 0.437 0.021 −0.514 −1.113 −4.648
9.468		 −0.133
0.211		 0
0		 −4.317
4.181		 −3.563
4.916
Root exposure 17.827 1.049 0.023 −0.554 −0.617 0
0		 −2.260
1.312		 −4.043
4.921		 −1.532
1.461		 −7.265
5.588
Stem-base burial 22.056 1.505 0.001 −0.880 −0.220 0
0		 −1.131
1.391		 −2.149
2.269		 −9.553
6.561		 −6.533
8.078
Reference 23.651 0.330 −0.016 −0.226 −0.593 NA −0.297
0.616		 NA NA −2.803
1.518
Open in a new tab

For decapitated, tilted and root-exposed trees, omitting all disturbance-related predictors has the strongest effect on the correlation between observed and modelled values (pseudo-R2 decreases by 37–48 %; Fig. 4). Moderate importance of disturbance parameters is also apparent for scarred trees (25 %) and stem-base-buried individuals (17 %). The importance of individual disturbance predictors varies between treatments – phase is dominant in scarred, tilted, stem-buried and decapitated trees; however, sampling level is dominant in trees with exposed roots. The effects of disturbance timing and sample orientation on stem circumference appear to be marginal and do not exceed 6 % change in pseudo-R2.

Fig. 4.

Fig. 4.

Open in a new tab

The effect sizes of individual predictors on correlations between modelled–observed values of specific hydraulic conductivity. The effect sizes were estimated based on decrease of pseudo-R2 after omitting of given predictor.

In scarred, stem-base-buried and reference trees, the random parameter tree has the largest influence on pseudo-R2 (Fig. 4). If tree is not used as a predictor in the model, pseudo-R2 decreases by up to 41 % for reference trees and between 13–32 % for different experimental treatments. The relative decrease in pseudo-R2 yielded by omitting all climatic factors is lower than 17 % for all treatments, always being lower than the combined effect of disturbance parameters. Cambial age is the second-most important predictor of Ks in reference trees (25 % decrease of pseudo-R2 arising from its omission); however, in mechanically damaged individuals its importance decreases significantly: moderate-to-high values of importance are shown for root exposure, stem-base burial and stem scarring (27, 18 and 12 % decrease in pseudo-R2, respectively), and very small importance is shown for tilting and decapitation (5 and 4 % decrease in pseudo-R2, respectively) (Fig. 4).

Histograms of residuals resemble the normal distribution, and the mean absolute value of residuals is 3.5 kg m−1 s−1 MPa−1 (24 % of mean Ks). There appears to be an increasing trend in residuals with observed values of Ks (Supplementary Data Fig. S5), which is driven by weaker predictive skills of linear models in the case of missing tree rings (Ks = 0) and tree rings with extremely low or high Ks.

DISCUSSION

Disturbance effect on specific hydraulic conductivity

Specific hydraulic conductivity of juvenile B. pendula was proven to be sensitive to both ontogenetic (internal) and environmental (external) factors. For all treatments, disturbances significantly influenced Ks, and leaving them out of the model yields a 17–48 % decrease in the squared correlation between modelled and observed values. The combined effect of all disturbance predictors is comparable to (stem-base burial) or stronger than (all other treatments) tree age. In addition, it also outweighs climatic forcing in all treatment groups. Abrupt adjustment of wood anatomical structure in response to mechanical damage has previously been observed in various species (Heinrich and Gärtner, 2008; Arbellay et al., 2012a, b) and attributed to reduced transpiration after crown damage, shifts in phytohormonal concentrations and/or balancing between safety and efficiency of water transport (Aloni, 2007; Sperry et al., 2008; Hacke et al., 2017). Stem tilting and decapitation caused the highest intensity of wood anatomical adjustment due to direct mechanical impact on the stem; in contrast, stem-base burial represents a rather indirect effect on wood anatomy due to soil hypoxia, causing less-intensive wood anatomical anomalies. Moreover, a period of stem burial longer than 3 years would probably be necessary to induce anatomical anomaly and, possibly, adventitious root formation (Copini et al., 2015).

Considering disturbance-related predictors, correlation between observed and modelled Ks was more sensitive to phase than to orientation and level, indicating that differences in wood anatomical structure between pre-experiment and post-experiment periods are larger than those along and around tree stems. Surprisingly, we observed low variability of Ks around the stem circumference in tilted or scarred trees. The abrupt decrease in vessel size and Ks due to stem tilting was observed around the entire stem with only slightly higher intensity on the tension wood side (Supplementary Data Fig. S1). Similarly, a significant decrease in mean vessel dimensions was observed for both the tension wood side and the opposite side of the stem in two different diffuse porous broadleaved species after tilting to 45° (Heinrich and Gärtner, 2008). This indicates that xylem conductivity after tilting is not primarily restricted by tension wood formation (influencing only the upper part of an inclined stem), but rather by gravitational forcing affecting the whole cambium. For instance, tilting of broadleaved seedlings of Quercus ilex L. to angles smaller than 30° successfully induced the formation of tension wood, but had no significant effect on xylem conductivity or embolism vulnerability (Gartner et al., 2003). In contrast, greater tilting (e.g. to 80° from the vertical) could cause divergence of Ks around the stem circumference by reducing both vessel size and density on the tension wood side (Heinrich and Gärtner, 2008). In scarred trees, the observed decrease in Ks being restricted only to the area adjacent to callus tissue is in line with the previously observed fact that anomalies in vessel size and density occur only in a sector of 30° (Arbellay et al., 2012b) to 90° (Tumajer et al., 2015) from the scar axis.

Our experiment simulated one disturbance event per approx. 15 years of juvenile stand existence (mean tree age at sampling year). The real contribution of disturbances to variability in specific hydraulic conductivity of forest ecosystems in long time series (decades to centuries) is dependent on the site-specific frequency of disturbances. A 15-year rotation of disturbance lies in the interval of recurrence of large-scale disturbance (e.g. windstorms) estimated for highly disturbed European mountain forests (Zielonka et al., 2010; Svoboda et al., 2014; Holeksa et al., 2017; Panayotov et al., 2017). However, the frequency of disturbances is expected to increase in the future (Seidl et al., 2017), and thus the effect of mechanical disturbances on transpiration via modifications of wood anatomical structure will increase in lowland forests as well. Although the disturbance sensitivity of mature trees decreases slightly with age due to increasing bark thickness (Šilhán and Stoffel, 2015), the relative effect of disturbances vs. other predictors on Ks would probably be larger in adult trees than juveniles due to the less pronounced age trend of vessel parameters (Olson et al., 2014).

Quantification of a model’s sensitivity to disturbance predictors might be also influenced by the number of available tree rings formed after the treatment, as experimental design should include not only the period immediately following the disturbance, but also several years afterwards to allow sufficient time for both adjustment and recovery. In our study, responses to most of the treatments (tilting, scarring, root exposure) are considered immediate (Hitz et al., 2008; Heinrich and Gärtner, 2008; Arbellay et al., 2013; Groover, 2016), and the graphical representation in Fig. 2 shows that our results are in agreement with this. Previous studies have found varying recovery speeds of wood anatomical parameters after strong mechanical events causing cambial damage (flood, rockfall, avalanche), ranging from 1 to 5 years (Arbellay et al., 2012b, 2013; Tumajer et al., 2015). Based on this, our experimental design included three analysed post-treatment years. For stem-base burial, a longer experiment would probably be needed to quantify its true effect on wood anatomy, as wood anatomical response to this phenomenon is often delayed (Copini et al., 2015).

Effects of tree age, climate and between-tree variability on specific hydraulic conductivity

Cambial age is responsible for a long-term increase in Ks of 0.71 kg m−1 s−1 MPa−1 per year (average across treatments; Table 2). This trend is driven by hydraulic scaling of vessel size due to increasing height of the tree (Olson et al., 2014; Rosell et al., 2017) and is a dominant continuous predictor of Ks if the tree was not mechanically damaged. However, if the tree is stressed by mechanical treatment, sensitivity of the model to cambial age decreases significantly to moderate values for treatments influencing roots or stem base (e.g. stem-base burial) or to very low values in the case of strong mechanical treatments applied to the stem. This suggests that the effect of wood anatomical scaling along the stem axis is outweighed by the anatomical response to the treatment (Tumajer et al., 2015).

The effect of climatic factors on the predictive ability of the models was significantly weaker than that of tree age, disturbances and between-tree variability in two types of treatments affecting basal parts of the tree (root exposure and stem-base burial). Only in strongly damaged stems did climate have a similar importance as tree age (due to strong disruption of the increasing trend in Ks with cambial age after the treatment), although its effect on modelled–observed concordance did not exceed 17 %. This indicates that climatic signal in quantitative wood anatomical series derived from stands subjected to periodical disturbances is largely masked by other effects. Additionally, ideally undisturbed stands should be used to extract any climatic signal, as their wood anatomical series would be free of disturbance effects (reflecting only an age trend and climatic signal). However, it is usually not possible to detect whether a tree was subjected to old disturbances, so possible disturbance signals should be taken into account for long chronologies. Before using quantitative wood anatomical series as a paleoclimatological proxy, it is necessary to: (1) reduce between-tree variability through sampling and analyse a sufficient sample depth to yield robust chronologies; (2) achieve a proper selection of undisturbed trees, in which climatic signal is not overshadowed by disturbance effects; and (3) appropriately remove age trends from the series (Carrer et al., 2015).

A significant proportion of the variability in Ks is driven by between-tree differences, which is in line with studies reporting naturally lower coherency statistics [e.g. between-tree correlation, expressed population signal sensuWigley et al. (1984)] for anatomical time-series comparing other dendrochronological proxies (Matisons et al., 2015; García-González et al., 2016). It is important to note that individual trees in the present study tend to have different mean values of Ks, but to show similar long-term trends and between-year variability (there is a random intercept but fixed slope of continuous predictors among the trees in the linear mixed-effects models). This indicates that wood anatomical structure of trees with different capacity for water transport respond in a similar manner to environmental influences and tree ageing. Drivers of the observed high between-tree variability are difficult to identify; however, as we were not able to reconstruct series of apical growth (and thus tree height) in the past, it is possible that at least some of this variability may be due to differences in tree size (Olson et al., 2014; Carrer et al., 2015). In addition, microsite effects, local soil thermal conditions or genetics may influence the water-conducting capacity of individual trees (Tombesi et al., 2010; Schreiber et al., 2015; Anadon-Rosell et al., 2018).

CONCLUSION

The results of this study indicate that inter-annual variability in specific hydraulic conductivity must be seen as a parameter driven by multiple factors, including disturbances, climate, tree age and tree-specific conditions. Mechanical treatment affected the anatomical structure of whole juvenile trees (regardless of the position of the sample along or around the stem) in tilted and decapitated individuals; scarred trees adjusted their anatomical structure only near callus tissue. In contrast, stem-base burial and root exposure induced less-pronounced anomalies. Mechanical treatments represent crucial predictors of variability in specific hydraulic conductivity, whose effects outweigh age trend and climate. This indicates that future variations in disturbance frequency may alter water conductivity and transpiration of forest ecosystems.

The multi-source-driven character of quantitative wood anatomical time series described here represents a challenge for their future use in environmental studies. Specific approaches (e.g. standardization, reduction of between-tree variability) need to be applied to extract the desired part of time series variability for studies on past climate, geomorphological processes or stand competition.

SUPPLEMENTARY MATERIAL

Supplementary data are available online at https://academic.oup.com/aob and consist of the following. Fig. S1: Observed values of specific hydraulic conductivity of tilted and decapitated trees. Fig. S2: Observed values of specific hydraulic conductivity of scarred trees. Fig. S3: Observed values of specific hydraulic conductivity of trees with buried-stem base and exposed roots. Fig. S4: Relationship between specific hydraulic conductivity, average vessel lumen area and vessel density. Fig. S5: Charts of residuals of linear mixed-effects models of xylem-specific hydraulic conductivity

mcy209_suppl_Supplementary_Material
Click here for additional data file. (1.3MB, docx)

ACKNOWLEDGEMENTS

We thank the Institute of Atmospheric Physics CAS for providing climatic data, J. Kašpar and J. Burda for their help in the field, J. Šobotník for providing access to laboratory techniques and J. Rosenthal for improving the English text. We are also grateful to two anonymous reviewers for their valuable comments and suggestions to previous versions of the manuscript. This work was supported by the Grant Agency of Charles University [GAUK: 174214] and the Ministry of Education, Youth and Sports of the Czech Republic [SVV: 260438].

LITERATURE CITED

  1. Aloni R. 2007. Phytohormonal mechanisms that control wood quality formation in young and mature trees. In: Entwistle K, Harris P, Walker J, eds. The Compromised Wood Workshop 2007. University of Canterbury, Christchurch, New Zealand: The Wood Technology Research Centre, 1–22. [Google Scholar]
  2. Anadon-Rosell A, Dawes MA, Fonti P, Hagedorn F, Rixen C, von Arx G. 2018. Xylem anatomical and growth responses of the dwarf shrub Vaccinium myrtillus to experimental CO2 enrichment and soil warming at treeline. Science of The Total Environment 642: 1172–1183. [DOI] [PubMed] [Google Scholar]
  3. Anderegg WRL, Klein T, Bartlett M, et al. 2016. Meta-analysis reveals that hydraulic traits explain cross-species patterns of drought-induced tree mortality across the globe. Proceedings of the National Academy of Sciences of the United States of America 113: 5024–5029. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Arbellay E, Corona C, Stoffel M, Fonti P, Decaulne A. 2012a Defining an adequate sample of earlywood vessels for retrospective injury detection in diffuse-porous species. PLoS ONE 7: e38824. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Arbellay E, Fonti P, Stoffel M. 2012b Duration and extension of anatomical changes in wood structure after cambial injury. Journal of Experimental Botany 63: 3271–3277. [DOI] [PubMed] [Google Scholar]
  6. Arbellay E, Stoffel M, Decaulne A. 2013. Dating of snow avalanches by means of wound-induced vessel anomalies in sub-arctic Betula pubescens. Boreas 42: 568–574. [Google Scholar]
  7. Ballesteros-Cánovas JA, Stoffel M, St George S, Hirschboeck K. 2015. A review of flood records from tree rings. Progress in Physical Geography 39: 794–816. [Google Scholar]
  8. Bates D, Mächler M, Bolker B, Walker S. 2015. Fitting linear mixed-effects models using lme4. Journal of Statistical Software 67: 1–48. [Google Scholar]
  9. Bivand RS, Pebesma E, Gomez-Rubio V. 2013. Applied spatial data analysis with R. New York: Springer. [Google Scholar]
  10. Carrer M, von Arx G, Castagneri D, Petit G. 2015. Distilling allometric and environmental information from time series of conduit size: the standardization issue and its relationship to tree hydraulic architecture. Tree Physiology 35: 27–33. [DOI] [PubMed] [Google Scholar]
  11. Castagneri D, Regev L, Boaretto E, Carrer M. 2017. Xylem anatomical traits reveal different strategies of two Mediterranean oaks to cope with drought and warming. Environmental and Experimental Botany 133: 128–138. [Google Scholar]
  12. Copini P, Decuyper M, Sass-Klaassen U, Gärtner H, Mohren F, den Ouden J. 2015. Effects of experimental stem burial on radial growth and wood anatomy of pedunculate oak. Dendrochronologia 33: 54–60. [Google Scholar]
  13. Cuny HE, Rathgeber CBK, Frank D, Fonti P, Fournier M. 2014. Kinetics of tracheid development explain conifer tree-ring structure. New Phytologist 203: 1231–1241. [DOI] [PubMed] [Google Scholar]
  14. Diaconu D, Hackenberg J, Stangler DF, Kahle H, Spiecker H. 2017. Simulation study to determine necessary sample sizes for image analysis-based quantitative wood anatomy of vessels of beech (Fagus sylvatica). Dendrochronologia 45: 35–38. [Google Scholar]
  15. Eilmann B, Zweifel R, Buchmann N, Fonti P, Rigling A. 2009. Drought-induced adaptation of the xylem in Scots pine and pubescent oak. Tree Physiology 29: 1011–1020. [DOI] [PubMed] [Google Scholar]
  16. Fonti P, Treydte K, Osenstetter S, Frank D, Esper J. 2009. Frequency-dependent signals in multi-centennial oak vessel data. Palaeogeography, Palaeoclimatology, Palaeoecology 275: 92–99. [Google Scholar]
  17. Friedman JM, Vincent KR, Shafroth PB. 2005 Dating floodplain sediments using tree-ring response to burial. Earth Surface Processes and Landforms 30: 1077–1091. [Google Scholar]
  18. García-González I, Souto-Herrero M, Campelo F. 2016. Ring-porosity and earlywood vessels: a review on extracting environmental information through time. IAWA Journal 37: 295–314. [Google Scholar]
  19. Gardiner B, Berry P, Moulia B. 2016. Review: Wind impacts on plant growth, mechanics and damage. Plant Science 245: 94–118. [DOI] [PubMed] [Google Scholar]
  20. Gartner BL, Roy J, Huc R. 2003. Effects of tension wood on specific conductivity and vulnerability to embolism of Quercus ilex seedlings grown at two atmospheric CO2 concentrations. Tree Physiology 23: 387–395. [DOI] [PubMed] [Google Scholar]
  21. Gärtner H. 2007. Tree roots — Methodological review and new development in dating and quantifying erosive processes. Geomorphology 86: 243–251. [Google Scholar]
  22. Gärtner H, Schweingruber FH. 2013. Microscopic preparation techniques for plant stem analysis. Remagen-Oberwinter: Verlag Dr. Kessel. [Google Scholar]
  23. Gebauer R, Volařík D. 2013. Root hydraulic conductivity and vessel structure modification with increasing soil depth of two oak species: Quercus pubescens and Quercus robur. Trees 27: 523–531. [Google Scholar]
  24. Gleason SM, Butler DW, Ziemińska K, Waryszak P, Westoby M. 2012. Stem xylem conductivity is key to plant water balance across Australian angiosperm species. Functional Ecology 26: 343–352. [Google Scholar]
  25. Gleason SM, Westoby M, Jansen S, et al. 2016. Weak tradeoff between xylem safety and xylem-specific hydraulic efficiency across the world’s woody plant species. New Phytologist 209: 123–136. [DOI] [PubMed] [Google Scholar]
  26. Gombin J, Vaidyanathan R, Agafonik V. 2017. concaveman: A Very Fast 2D Concave Hull Algorithm. R package version 1.0.0. https://CRAN.R-project.org/package=concaveman [Google Scholar]
  27. Groover A. 2016. Gravitropisms and reaction woods of forest trees - evolution, functions and mechanisms. New Phytologist 211: 790–802. [DOI] [PubMed] [Google Scholar]
  28. Hacke UG, Spicer R, Schreiber SG, Plavcová L. 2017. An ecophysiological and developmental perspective on variation in vessel diameter. Plant, Cell & Environment 40: 831–845. [DOI] [PubMed] [Google Scholar]
  29. Heinrich I, Gärtner H. 2008. Variations in tension wood of two broad‐leaved tree species in response to different mechanical treatments: implications for dendrochronology and mass movement studies. International Journal of Plant Sciences 169: 928–936. [Google Scholar]
  30. Hitz OM, Gärtner H, Heinrich I, Monbaron M. 2008. Application of ash (Fraxinus excelsior L.) roots to determine erosion rates in mountain torrents. Catena 72: 248–258. [Google Scholar]
  31. Holeksa J, Jaloviar P, Kucbel S, et al. 2017. Models of disturbance driven dynamics in the West Carpathian spruce forests. Forest Ecology and Management 388: 79–89. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Jacobsen AL, Pratt RB, Tobin MF, Hacke UG, Ewers FW. 2012. A global analysis of xylem vessel length in woody plants. American Journal of Botany 99: 1583–1591. [DOI] [PubMed] [Google Scholar]
  33. Matheny AM, Bohrer G, Vogel CS, et al. 2014. Species-specific transpiration responses to intermediate disturbance in a northern hardwood forest. Journal of Geophysical Research: Biogeosciences 119: 2292–2311. [Google Scholar]
  34. Matisons R, Elferts D, Brumelis G. 2012. Changes in climatic signals of English oak tree-ring width and cross-section area of earlywood vessels in Latvia during the period 1900–2009. Forest Ecology and Management 279: 34–44. [Google Scholar]
  35. Matisons R, Jansons J, Katrevičs J, Jansons Ā. 2015. Relation of tree-ring width and earlywood vessel size of alien Quercus rubra L. with climatic factors in Latvia. Silva Fennica 49: id 1391. [Google Scholar]
  36. Nakagawa S, Schielzeth H. 2013. A general and simple method for obtaining R2 from generalized linear mixed-effects models. Methods in Ecology and Evolution 4: 133–142. [Google Scholar]
  37. Olson ME, Anfodillo T, Rosell JA, et al. 2014. Universal hydraulics of the flowering plants: vessel diameter scales with stem length across angiosperm lineages, habits and climates. Ecology Letters 17: 988–997. [DOI] [PubMed] [Google Scholar]
  38. Panayotov M, Gogushev G, Tsavkov E, et al. 2017. Abiotic disturbances in Bulgarian mountain coniferous forests – An overview. Forest Ecology and Management 388: 13–28. [Google Scholar]
  39. R Core Team 2017. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. [Google Scholar]
  40. Régent Instruments 2011. WinCell 2011Pro For Wood Cell Analysis. Québec: Régent Instruments Inc. [Google Scholar]
  41. Rosell JA, Olson ME, Anfodillo T. 2017. Scaling of xylem vessel diameter with plant size: causes, predictions, and outstanding questions. Current Forestry Reports 3: 46–59. [Google Scholar]
  42. Schelhaas M-J, Nabuurs G-J, Schuck A. 2003. Natural disturbances in the European forests in the 19th and 20th centuries. Global Change Biology 9: 1620–1633. [Google Scholar]
  43. Schneider L, Gärtner H. 2013. The advantage of using a starch based non-Newtonian fluid to prepare micro sections. Dendrochronologia 31: 175–178. [Google Scholar]
  44. Scholz A, Klepsch M, Karimi Z, Jansen S. 2013. How to quantify conduits in wood? Frontiers in Plant Science 4: 1–11. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Schreiber SG, Hacke UG, Hamann A. 2015. Variation of xylem vessel diameters across a climate gradient: insight from a reciprocal transplant experiment with a widespread boreal tree. Functional Ecology 29: 1392–1401. [Google Scholar]
  46. Seidl R, Thom D, Kautz M, et al. 2017. Forest disturbances under climate change. Nature Climate Change 7: 395–402. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Shroder JF. 1980. Dendrogeomorphology: review and new techniques of tree-ring dating. Progress in Physical Geography 4: 161–188. [Google Scholar]
  48. Šilhán K, Stoffel M. 2015. Impacts of age-dependent tree sensitivity and dating approaches on dendrogeomorphic time series of landslides. Geomorphology 236: 34–43. [Google Scholar]
  49. Simard S, Giovannelli A, Treydte K, et al. 2013. Intra-annual dynamics of non-structural carbohydrates in the cambium of mature conifer trees reflects radial growth demands. Tree Physiology 33: 913–923. [DOI] [PubMed] [Google Scholar]
  50. Sperry JS, Meinzer FC, McCulloh KA. 2008. Safety and efficiency conflicts in hydraulic architecture: scaling from tissues to trees. Plant, Cell & Environment 31: 632–645. [DOI] [PubMed] [Google Scholar]
  51. Stoffel M, Corona C. 2014. Dendroecological dating of geomorphic disturbance in trees. Tree-Ring Research 70: 3–20. [Google Scholar]
  52. Svoboda M, Janda P, Bače R, et al. 2014. Landscape-level variability in historical disturbance in primary Picea abies mountain forests of the Eastern Carpathians, Romania. Journal of Vegetation Science 25: 386–401. [Google Scholar]
  53. Thom D, Seidl R. 2016. Natural disturbance impacts on ecosystem services and biodiversity in temperate and boreal forests. Biological Reviews 91: 760–781. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Tombesi S, Johnson RS, Day KR, DeJong TM. 2010. Relationships between xylem vessel characteristics, calculated axial hydraulic conductance and size-controlling capacity of peach rootstocks. Annals of Botany 105: 327–331. [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Tumajer J, Burda J, Treml V. 2015. Dating of rockfall events using vessel lumen area in Betula pendula. IAWA Journal 36: 286–299. [Google Scholar]
  56. Tyree MT, Zimmermann MH. 2002. Xylem structure and the ascent of sap. Berlin: Springer. [Google Scholar]
  57. Wigley TML, Briffa KR, Jones PD. 1984. On the average value of correlated time series, with applications in dendroclimatology and hydrometeorology. Journal of Climate and Applied Meteorology 23: 201–213. [Google Scholar]
  58. Zhao X. 2015. Effects of cambial age and flow path-length on vessel characteristics in birch. Journal of Forest Research 20: 175–185. [Google Scholar]
  59. Zielonka T, Holeksa J, Fleischer P, Kapusta P. 2010. A tree-ring reconstruction of wind disturbances in a forest of the Slovakian Tatra Mountains, Western Carpathians. Journal of Vegetation Science 21: 31–42. [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

mcy209_suppl_Supplementary_Material
Click here for additional data file. (1.3MB, docx)

Articles from Annals of Botany are provided here courtesy of Oxford University Press

RESOURCES