• M. Mencuccini,

  • F. Magnani

Comment on ‘Hydraulic limitation of tree height: a critique’ by Becker, Meinzer & Wullschleger

Becker, Meinzer & Wullschleger (2000) recently presented a critique of the hypothesis that height growth in trees is limited by a hydraulic constraint determined by height itself ( Ryan & Yoder 1997). In the original paper, Ryan & Yoder (1997) discussed the various hypotheses proposed to explain the decline in above-ground productivity and height growth in trees and proposed that height may reduce the ability of tall trees to transport water to the leaves at the tops of tree crowns.

In other words, there would be a negative feedback between the early increments in tree height and the subsequent increases in the resistance to water transport, which would then act to reduce further growth.

Rightly so, Becker et al. (2000) warn against hasty conclusions of causality and illustrate with a few examples the complexity of the factors affecting tree ecological and physiological behaviour, and particularly its water relations during growth.

Here, we would like to draw attention to some aspects, which have relevance to the arguments presented but have not been considered by Becker et al. (2000) . For brevity, we will focus only on two aspects: (1) the proposed homeostasis of leaf-specific hydraulic conductance during a tree life cycle and (2) the role of leaf hydraulic resistance in the context of stomatal control.

Homeostasis in hydraulic supply: yes, no, maybe. but exactly how?

Becker et al. (2000) concluded that leaf-specific hydraulic conductance, i.e. the total plant hydraulic transport capacity relative to the leaf area present (units of kg MPa−1 m−2 s−1), is unlikely to be lower in tall trees. Rather, a homeostatic mechanism would continuously maintain a constant proportion between the transpiring leaf area and the transporting tissues throughout the tree life cycle. Becker et al. (2000) explicitly proposed one possibility, i.e. that reductions in leaf-sapwood area ratios occur in tall trees, and apparently also suggested a second one, i.e. that xylem water transport becomes more efficient with ontogeny.

We propose that the discussion start instead from an examination of absolute values of tree hydraulic conductance without considering, for the moment, the amount of leaf area present on each tree. Introducing the relationship between hydraulic transport capacity and leaf area from the beginning only confounds the matter, because growth in diameter serves the double purpose of supplying a larger crown and balancing the increases in path length.

We will adopt a scaling approach, i.e. we will try to interpret changes in the values of the allometric coefficients of log–log relationships between size and hydraulic properties (e.g. Niklas 1994). We will first examine the relationship between size and hydraulic properties at the scale of individual segments that compose a tree, and then we will move up to whole plants. Here, we define hydraulic conductivity (kh, kg s−1 MPa−1 m−1) as the water transport capacity of stem segments of unitary length, whereas we define hydraulic conductance (K, kg s−1 MPa−1) as the water transport capacity of whole plants of various lengths. The effect of height on the hydrostatic pressure gradient will not be considered for the time being.

In Table 1, we summarized the information for four species (two conifers and two broad-leaved), for which data are available on the allometric scaling of both stem segment kh (part A) and above-ground K (part B). Because data are available only for the above-ground portion of the pathway, K indicates the water transport capacity from ground level to the terminal branchlets. Scaling is calculated with respect to stem segment diameter in the first case, tree breast-height diameter in the second one.

Table 1.  Allometric scaling of hydraulic conductivity kh of stem segments (A) and of hydraulic conductance K of trees (B) with their respective diameters. Conductivity of stem segments and hydraulic conductance of whole trees are defined in the text: bLS, regression coefficient based on least square regression; bRMA, regression coefficient based on reduced major axis regression; SEb, standard error of the regression coefficient; R2, percentage of variance explained by least square regression. The regression for whole Pinus banksiana trees is based on resistances from ground level to the upper third of the crown only
 A Regression coefficients for khB Regression coefficients for K 
SpeciesbLSbRMA(SEb) R2bLSbRMA(SEb) R2Authors
Acer rubrum2·632·72(0·05)0·941·661·73(0·08)0·92Yang & Tyree (1994)
Acer saccharum2·642·67(0·05)0·981·621·64(0·14)0·97Tyree & Sperry (1988), Yang & Tyree (1994)
Pinus sylvestris2·552·63(0·08)0·940·971·22(0·14)0·63Mencuccini & Grace (1996)
Pinus banksiana2·422·52(0·24)0·921·141·34(0·31)0·72Pothier et al. (1989)

Both least square regression and reduced major axis regression show that the four regression coefficients (i.e. the slopes of the log–log regressions) of kh against segment diameter are significantly larger than 2·0 ( Table 1). If kh scaled isometrically with the cross-sectional area (which is proportional to diameter squared), then the regression exponent will equal 2·0. These results are typical of many other studies where kh has been scaled against segment diameter and illustrate that larger stems in fact have, as implied by Becker et al. (2000) , a more permeable woody matrix than smaller stems. This can be caused by an increase in the number of conducting elements (tracheids or vessels) and/or an increase in their size, but this last factor is normally prevalent in trees and results from the well-known processes of cambial maturation in older trees.

However, when above-ground K is scaled against tree diameter at 1·3 m height (part B of Table 1), regression coefficients significantly lower than two are obtained (at least P < 0·05), irrespective of the particular methods used for estimating above-ground conductance. If trees were perfectly capable of hydraulically compensating for their height solely by the production of a more efficient xylem, then the scaling coefficient of above-ground hydraulic conductance K with stem diameter should equal 2·0. In other words, above-ground hydraulic conductance K should increase in exact proportion to the square of diameter, irrespective of total height. However, despite cambial maturation, K scales allometrically with respect to the cross-sectional area.

What is the cause for the reduced scaling of the above-ground water transport capacity despite production of a more efficient xylem? We propose that tree height was a major factor in all four cases of Table 1. Height can be involved directly in more than one way but we present only one argument here: in tall trees the proportion of total hydraulic resistance accounted for by the central stem increases, while the importance of small branchlets declines as trees grow in height ( Yang & Tyree 1994; Mencuccini & Grace 1996). Height is also involved indirectly, because taller trees will also tend to have longer branches.


Clearly, from the perspective of leaf gas exchange, what is relevant is not the absolute value of plant hydraulic conductance but the relationship between this property and the leaf area supported. This underscores the importance of considering how leaf area changes with functional sapwood as trees grow taller. From what has been said above, it is obvious that reductions in leaf-sapwood (AL/ASW) area ratios are a necessity, if one postulates that a homeostasis in the above-ground leaf-specific hydraulic conductance is maintained during growth.

The available data support the hypothesis that AL/ASW frequently decline with age. However, in three of the four studies of Table 1, despite changes in AL/ASW, the above-ground leaf-specific hydraulic conductance declined during growth. Even for Acer saccharum ( Yang & Tyree 1994) a re-analysis of the published data set indicates that the reduction with size was highly significant (y = 5·308 + 0·084/x, adjusted R2 = 0·38, F = 20·4, P < 0·0001). For the second species, Acer rubrum, the sample size was too limited to draw any significant conclusion. The published evidence does not unequivocally suggest that above-ground leaf-specific hydraulic conductance is independent of height. In fact, it is quite the contrary.


Also, the question must be raised whether the decreases in AL/ASW are mediated by increases in ASW or, vice versa, decreases in AL. In either case, the carbon costs and gains to the plant are affected and, one would think, growth would be negatively affected as well. Is increased allocation to sapwood production obtained at the expense of further leaf area growth? If so, from an evolutionary perspective, would it be selectively adventageous for a tree to maintain more leaves even though the hydraulic transport capacity is less than optimal?

An alternative

In an alternative treatment of the concept of hydraulic limitations to growth, Magnani et al. (2000) proposed that allocation to fine roots compensate for increased hydraulic resistance above-ground, so that a constant level of hydraulic supply to the leaves is maintained at the whole-plant (not the shoot) scale during tree growth. The carbon cost of maintaining and replacing fine roots was very large in mature trees, and enough to reduce the forest net primary productivity (see Magnani et al. 2000 ). The suggestion here is that the hydraulic limitation hypothesis should be reformulated considering partitioning of dry matter above and below ground and the consequent carbon gains and costs in addition to considering structural changes and leaf area-sapwood area changes with increasing tree height.

Leaf hydraulic resistance. Are we interpreting the data correctly

One argument that Becker et al. (2000) put forward to discount the role of path length was that axial resistance only accounts for a minor fraction of the total soil-to-leaf resistance, the major components of which are located elsewhere.

While it is difficult to disagree with the statement that it is the total soil-to-leaf resistance that it is of relevance in this context, it is also necessary to clarify some confusion in relation to the magnitude and role played by leaf hydraulic resistance.

It has been shown a number of times that leaf hydraulic resistance can play a significant role in plant water relations. However, we urge caution in the interpretation of recent measurements (e.g. Yang & Tyree 1994; Cochard et al. 1997 ), for two main reasons. The method used (perfusion of individual leaves and subsequent measurement of liquid water flow under a known pressure gradient, ΔP) does not give comparable results with alternative approaches (water potential Ψ of bagged vs non-bagged leaves; kynetics of water loss during pressure-bomb experiments – see discussion in Yang & Tyree 1994). The discrepancy among methods is large: the estimate of ΔΨ within a leaf varied from 0·15 to 0·25 MPa for one method to 0·4–1·0 MPa for another. Until such discrepancies are solved, conclusions about the role of leaf resistance are not well rested (see discussion in Yang & Tyree 1994).

More important, the method yields a measurement of hydraulic resistance, and consequently of water potential Ψ, referred to the ultimate sites of evaporation within the leaf (cf. Yang & Tyree 1994). In the context of stomatal responses to changes in axial resistances, however, the relevant parameter is the hydraulic resistance, and the consequent Ψ, of the putative controlling sites where the hydraulic/metabolic signals for stomatal responses are produced.

It is not presently clear whether these putative sites are physically within the bulk of the transpiration stream and how close they are to the final sites of bulk evaporation. It may also be that they are located fairly closely to the central leaf bundle sheaths and xylem vessels, so that the relevant hydraulic resistance would only be slightly larger than the resistance from the petiole to the veins.

Because it is very unlikely that these signals are produced right at the sites of evaporation, it can be concluded that the values reported for leaf resistance, although correct, may not bear much relevance to the problem in question.

Concluding Remarks

Path length is a fundamental component of the plant hydraulic system ( Table 1). Ontogenetic changes in leaf-sapwood area ratios cannot be considered separately from the simultaneous changes in height and branch length that occur during growth. Increased allocation to sapwood volume (above and below ground) must occur at the expense of further leaf area production, with likely costs for plant growth.

There is a contradiction in arguing that homeostasis exists in leaf-specific hydraulic conductance and that old trees have a larger capacitance volume. If that were the case, large trees would have higher per-unit-leaf gas-exchange rates or less negative water potentials, and homeostasis would not really be present. It is instead more likely that increased reliance on capacitance water is accompanied by a reduced carbon investment in transport tissues.

The hypothesis of hydraulic limitations to tree height growth is likely to be a simplistic approximation of a much more complex phenomenon, but it may be useful in highlighting the areas in which further investigations and experiments are necessary.

We propose here that future studies of tree hydraulics should concentrate on determining the ontogenetic changes in (1) root hydraulic resistance and (2) whole-plant hydraulic capacitance.