915

American Journal of Botany 98(5): 915–922. 2011.

American Journal of Botany 98(5): 915–922, 2011; http://www.amjbot.org/ © 2011 Botanical Society of America

Over the last 10 years, ecologists have identifi ed specifi c plant functional traits that link ecological strategies with varia-tion in resource use and life history properties ( Curtis and Ackerly, 2008 ). Wood density, characteristics of the xylem conduction system ( Chave et al., 2009 ; Zanne and Falster, 2010 ; Zanne et al., 2010 ), and plant height ( Moles et al., 2004 ) have emerged as three of these key traits. Although the general trends in ecological xylem structure are well established ( Carlquist, 2001 ; Sperry et al., 2008 ), there are still open questions as to how plants adjust their xylem structure to meet transpiration and mechanical stability needs ( Zanne et al., 2010 ). Water conduction

traits, wood density, and height are interrelated, but the strength and even the direction of their relationship often vary among studies, most likely because studies vary in environmental and spatial scope of woody plants. Here we question whether inte-gration among these three traits varies as a function of growth form, i.e., shrubs vs. trees. Woody growth forms have been shown to differ in average wood traits ( Carlquist and Hoekman, 1985 ), but to date there are no comparisons between shrubs and trees with regard to the integration of xylem traits, wood den-sity, and height across a phylogenetically broad sampling of species from a range of habitats.

In woody angiosperms, vessel diameter has a large effect on water conduction effi ciency because conduction capacity scales as diameter to the fourth power ( Zimmerman, 1983 ). Conduc-tion effi ciency in turn infl uences transpiration and carbon fi xa-tion rates, and ultimately growth ( Tyree, 2003 ; Poorter et al., 2009 ). For a given stem cross-sectional area of a woody ang-iosperm, fewer wide vessels offer higher conductivity than do many small vessels ( Sperry et al., 2008 ). On the other hand, larger numbers of smaller vessels provide increased redundancy and resistance against hydraulic failure ( Carlquist, 1980 ; Ewers et al., 2007 ).

The density of wood, i.e., its dry mass per unit of fresh vol-ume, refl ects carbon investment in mechanical support as well as conduction effi ciency and consequently is related to plant

1 Manuscript received 3 September 2010; revision accepted 14 February 2011. The authors thank S. Espino for measurements of wood densities in

shrubs. They also thank B. Pratt, C. Schlichting, and two anonymous reviewers for comments on an earlier draft. Funding for this research was provided by a grant from the Andrew W. Mellon Foundation to H.J.S., from the National Science Foundation to H.J.S. (IOS-0641765) and C.S.J. (IOS-0641569). The scholarship provided by CONACyT to H.I.M.C. also is appreciated.

5 Author for correspondence (e-mail: hugo.martinez_cabrera@uconn.edu), phone: 1-860-486-4027, fax: 1-860-486-6364

doi:10.3732/ajb.1000335

BRIEF COMMUNICATION

INTEGRATION OF VESSEL TRAITS, WOOD DENSITY, AND HEIGHT IN ANGIOSPERM SHRUBS AND TREES 1

Hugo I. Mart í nez-Cabrera 2,5 , H. Jochen Schenk 3 , Sergio R. S. Cevallos-Ferriz 4 , and Cynthia S. Jones 2

2 Department of Ecology and Evolutionary Biology, Unit-3043, 75 N. Eagleville Road, University of Connecticut, Storrs, Connecticut 06269-3043 USA; 3 Department of Biological Science, California State University Fullerton, P.O. Box 6850, Fullerton, California 92834-6850 USA; and 4 Instituto de Geolog í a, Universidad Nacional Aut ó noma de M é xico, Ciudad

Universitaria, Circuito de la Investigaci ó n Cient í fi ca, Del. Coyoacan 04510 M é xico D.F., M é xico

• Premise of the study: Trees and shrubs tend to occupy different niches within and across ecosystems; therefore, traits related to their resource use and life history are expected to differ. Here we analyzed how growth form is related to variation in integration among vessel traits, wood density, and height. We also considered the ecological and evolutionary consequences of such differences.

• Method: In a sample of 200 woody plant species (65 shrubs and 135 trees) from Argentina, Mexico, and the United States, standardized major axis (SMA) regression, correlation analyses, and ANOVA were used to determine whether relationships among traits differed between growth forms. The infl uence of phylogenetic relationships was examined with a phylogenetic ANOVA and phylogenetically independent contrasts (PICs). A principal component analysis was conducted to determine whether trees and shrubs occupy different portions of multivariate trait space.

• Key results: Wood density did not differ between shrubs and trees, but there were signifi cant differences in vessel diameter, vessel density, theoretical conductivity, and as expected, height. In addition, relationships between vessel traits and wood den-sity differed between growth forms. Trees showed coordination among vessel traits, wood density, and height, but in shrubs, wood density and vessel traits were independent. These results hold when phylogenetic relationships were considered. In the multivariate analyses, these differences translated as signifi cantly different positions in multivariate trait space occupied by shrubs and trees.

• Conclusions: Differences in trait integration between growth forms suggest that evolution of growth form in some lineages might be associated with the degree of trait interrelation.

Key words: evolution; growth form; shrubs; trait integration; trees; wood anatomy; wood density.

916 American Journal of Botany [Vol. 98

on trees, suggesting that effi cient height growth necessitates coordination between conduction capacity and allocation to mechanical support. However, we previously found that in shrubs, wood density is not correlated with conduction effi -ciency traits ( Mart í nez-Cabrera et al., 2009 ), possibly indicat-ing that they are decoupled. Here we combine tree and shrub data to determine the extent to which vessel traits scale with wood density and height across growth forms.

MATERIALS AND METHODS

Sampling and traits — Shrubs and trees were distinguished by the number of basal stems and height of branching. Shrubs were defi ned as multistemmed with low branching points ( < 1 m above ground), trees as one-stemmed with higher branching points. One species, Prosopis glandulosa , can assume either of these growth forms, but we classifi ed it as a shrub, which is its only growth form at the collection site.

A database of vessel size, vessel density, and wood density was compiled for 200 species of woody angiosperms from 11 sites across America (Argentina, Mexico, and the United States) from our previous work (Appendix S1, see Supple-mental Data online at http://www.amjbot.org/cgi/content/full/ajb.1000335/DC1). Most of the 65 shrubs were collected from six very dry (100 mm/yr) to mesic sites (1490 mm/yr) in the United States and Argentina, and the trees came from three sites that ranged from dry (700 mm/yr) to tropical rain forest (ca. 4000 mm/yr) in Mexico. Most trees and shrubs were not from the same sites. Sapwood samples were taken at dbh for trees and at the most compact point of the stem base, 0 – 10 cm above ground level for shrubs, which is the closest equivalent to a tree trunk.

Anatomical traits were measured in radial sectors of the two most recent growth rings. Ring porosity was present in a relatively small number of species (15%, 30 species); therefore, we decided to analyze diffuse-porous and ring-porous woods together. In a global study analyzing angiosperm wood density and vessel traits, no bias was detected when diffuse-porous and ring-porous woods were analyzed together ( Zanne et al., 2010 ). For the Argentinean and North American communities, vessel lumen diameters were calculated as the diameters of circles with the same area of individual vessel lumens ( Kolb and Sperry, 1999 ) and were based on ca. 200 vessels per sample ( Schenk et al., 2008 ; Mart í nez-Cabrera et al., 2009 ). For the Mexican sites, tangential vessel diameter was measured directly on the slides and was based on at least 25 ran-domly selected vessels ( IAWA Committee, 1989 ). In the species from Mexican sites, vessel density was determined in at least 10 fi elds of view; in the Argen-tinean and North American sites, it was determined in entire radial sectors. Theoretical conductivity ( K t ) per square millimeter of cross-sectional area of sapwood was calculated using a modifi ed version of the Hagen – Poiseuille equation ( Tyree and Ewers, 1991 ):

4

1

( /128 )t ii

nK dρ η

where ρ is the density of the fl uid in km · m − 3 ; η is the dynamic viscosity of the fl uid in MPa · s − 1 ; n is the number of vessels; and d is the diameter of the vessels. Vessel fraction was calculated as vessel area multiplied by vessel density ( Preston et al., 2006 ). Wood density data for the samples used for the anatomical analy-sis were available from previous studies ( Barajas-Morales, 1987 ; Schenk et al., 2008 ; Mart í nez-Cabrera et al., 2009 ) and was determined using the Archimedes principle: wood samples were immersed in a beaker of water that was placed on a high-precision balance, and the weight of water displacement was converted into volume. Afterward, samples were dried for 24 to 48 h. Because plant heights for the actual specimens were not available for the tree data, maximum plant height data for species were compiled for all species from the literature. For further descriptions of sites, sampling, and anatomical analyses procedures, see ( Mart í nez-Cabrera and Cevallos-Ferriz, 2008 ; Schenk et al., 2008 ; Mart í nez-Cabrera et al., 2009 ).

Phylogeny — Phylogenetic relationships among species were reconstructed using the program PHYLOMATIC ( Webb and Donoghue, 2004 ) and were refi ned using published phylogenies ( Olmstead and Palmer, 1992 ; Alverson et al., 1999 ; Robinson and Harris, 2000 ; Urbatsch et al., 2000 ; Lia et al., 2001 ; Floyd, 2002 ; Goertzen et al., 2003 ; Urbatsch et al., 2003 ; Baum et al., 2004 ; Wojciechowski et al., 2004 ; Martins and Barkmani, 2005 ) (Appendix S2, see

life history characteristics. Higher wood densities in trees are associated with slower growth rates, higher survival, longer life spans, and later successional status, whereas the converse is true for trees with lower wood densities ( ter Steege and Hammond, 2001 ; Muller-Landau, 2004 ; Chave et al., 2009 ; Poorter et al., 2009 ). It may be expected that the total amount of vessel lumen varies inversely with wood density ( Preston et al., 2006 ; Zanne et al., 2010 ), and this relationship has indeed been found in some studies ( Preston et al., 2006 ; Jacobsen et al., 2007 ), but not in others ( Mart í nez-Cabrera et al., 2009 ; Poorter et al., 2009 ). Higher wood densities are also associated with increas-ing resistance to embolism formation ( Hacke et al., 2001 ).

Height, on the other hand, determines the plant ’ s ability to compete for light and is correlated with traits related to carbon gain, such as leaf mass per area, leaf nitrogen per area, and can-opy area ( Falster and Westoby, 2003 ). Plant height is also posi-tively related with life history traits such as longevity and time to reproduction ( Moles et al., 2004 ; Petit and Hampe, 2006 ). In an analysis of 51 woody plants in central California, Preston et al. (2006) found height to be positively associated with vessel diameters and negatively with vessel densities. This association is thought to be due to increased hydraulic resistance imposed by increased gravitational forces and greater path lengths in taller plants ( Ryan and Yoder, 1997 ; Coomes et al., 2008 ). Ves-sel tapering is the anatomical response to these forces; vessel diameters are larger in the proximal part of the stems ( McCulloh et al., 2003 , 2004 ), which compensates for increased resistance due to height. Chave et al. (2009) also suggested that taller plants have wider conduits because wider conduits are more effi cient per unit area, leaving more space for fi bers for me-chanical support.

Interestingly, studies have found both positive and negative relationships between height and wood density ( Thomas, 1996 ; ter Steege and Hammond, 2001 ; Falster and Westoby, 2005a , b ; Preston et al., 2006 ; van Gelder et al., 2006 ; Aiba and Nakashizuka, 2009 ). These opposite patterns have been ex-plained by differences in ecological strategies. For example, light-demanding, short species that dominate early succession tend to grow fast, presumably temporarily outcompeting their neighbors, and are characterized by woods that are relatively less dense than those in taller pioneers that persist longer. Among shade-tolerant species, shorter species tend to be mechanically stronger, which helps them to survive damage in the understory from falling debris ( Gartner, 1989 ; Putz and Brokaw, 1989 ), while in larger species, lower wood density is correlated with higher growth rates ( Falster and Westoby, 2005a ; van Gelder et al., 2006 ).

Because growth form and height of self-supporting woody plants are strongly related to resource availability (e.g., King, 1990 ; Ryan and Yoder, 1997 ; Falster and Westoby, 2003 ; Petit and Hampe, 2006 ), shrubs are generally, but not always, found at the low-resource side of the ecological spectrum. They tend to occupy drier, colder, and more nutrient-poor habitats than trees, and are the prevailing growth form in arid regions. In more mesic areas, trees dominate the canopy layers, while shrubs thrive in the light-limited understory. Given the general relationship between low-resource environments and increas-ing wood density described, we hypothesized that shrubs will have higher wood density than trees.

Several studies have documented that vessel size decreases with greater wood density ( Rao and Aebischer, 1997 ; Rao et al., 2002 ; Thomas et al., 2004 , 2007 Naidoo et al., 2006 ; Moya Roque and Tomazelo-Filho, 200707 ). These all have focused

917May 2011] Mart í nez-Cabrera et al. — Trait integration in trees and shrubs

( F 1, 198 = 203.9, P < 0.001 for standard and P < 0.01 for phylo-genetic ANOVA) than shrubs (mean = 37.2 µ m, SE = 1.9), and as a consequence, they had signifi cantly higher K t than shrubs ( F 1, 198 = 161.7, P < 0.001 for standard and P < 0.01 for phylo-genetic ANOVA). In contrast, shrubs had signifi cantly higher vessel densities (mean = 89.3 vessels · mm − 2 , SE = 10.8) ( F 1, 198 = 93.1, P < 0.001 for standard and P < 0.05 for phylogenetic ANOVA) than trees (mean = 26.7 vessels · mm − 2 , SE = 2.5). As expected, trees were signifi cantly taller (mean = 17.9 m, SE = 0.7) ( F 1, 198 = 171.7, P < 0.001 for standard and P < 0.01 for phylogenetic ANOVA) than shrubs (mean = 3.4 m, SE = 0.4). Vessel fraction was different between growth forms only with standard ANOVA (trees: mean = 0.12, SE = 0.006; shrubs: mean = 0.07, SE = 0.005; F 1, 198 =20.7, P < 0.001) suggesting that phylogeny strongly affects the total fraction of wood cross-sectional area devoted to vessels. The signifi cance value of the phylogenetic ANOVA was always lower than the standard ANOVA analysis. Signifi cant differences persisted between growth forms when phylogeny was considered, except for ves-sel fraction.

Wood density and vessel traits — Vessel diameter increased as vessel density declined in both growth forms, but at a given vessel density, shrubs had smaller vessels ( Fig.1A ). Slopes did not differ between growth forms, but elevation did ( Table 1 , Fig. 1A ). Thus, the well-established tradeoff between vessel size and number previously reported (e.g. Carlquist, 2001 ) holds similarly across trees and shrubs. This tight vessel size – vessel density association drives the relationship of vessel traits with wood density. Across all data pooled, wood density was negatively correlated with vessel diameter and thus with K t and positively correlated with vessel density ( Table 2 ; Fig. 1B, C ). Across species, wood density was not correlated with vessel fraction ( Fig. 1D ; Table 2 ).

The correlations between wood density and vessel diameter and vessel density described were driven by the tree data. When shrubs were analyzed separately, no signifi cant associations be-tween vessel traits and wood density were observed ( Tables 1, 2 ). Because of this independence, the test of common slopes was not carried out, but interestingly, the SMA slopes of the rela-tionship between wood density and vessel traits had opposite signs in trees vs. shrubs ( Table 1 , Fig. 1B – D ).

Wood density and plant height — Wood density was negatively related to height using the Pearson correlation ( Table 2 ). However, this relationship was not signifi cant when PICs were used ( Table 2 ). SMA slopes showed that denser xylem is found in smaller plants regardless of growth form ( Fig. 1E ), but again this signifi -cant relation held only in the combined data and in trees.

Plant height and vessel traits — Height was positively asso-ciated with vessel diameter, K t , and vessel fraction, and nega-tively related with vessel density ( Table 2 ). Correlations were weaker or not signifi cant within shrubs when they were ana-lyzed separately ( Table 2 ). These trends were also observed using SMA regression ( Fig. 1E – G ). Trees and shrubs did not share common slopes for the height vs. vessel diameter and K t relationships. This test was not performed for vessel fraction and vessel density because their relationships with height were not signifi cant for shrubs or trees ( Table 1 ).

PCA analysis — Trees and shrubs tended to occupy different regions of the multivariate space represented by the fi rst and

online Supplemental Data). The remaining polytomies were treated as soft; therefore, phylogenetically independent contrasts (PICs) were calculated using the methods of Purvis and Garland (1993) and Garland and D í az-Uriarte (1999 ) as implemented in PDAP:PDTREE ( Midford et al., 2005 ), a module of the program Mesquite version 2.7.1 ( Maddison and Maddison, 2009 ). Characters that were not adequately standardized (i.e., the absolute values of the standard-ized PICs and their standard deviations showed a signifi cant association) ( Garland et al., 1991 , 1992 ), were log transformed. Branch lengths of the result-ing tree were calculated using the branch length adjusting algorithm (bladj) implemented in the program PHYLOCOM ( Webb et al., 2007 ). Angiosperm node ages were used according to Wikstrom et al. (2001) for calibrating the phylogenetic tree. The resulting tree was used for the PIC analysis. The existing polytomies of the base backbone tree were randomly resolved using the pro-gram APE ( Paradis et al., 2004 ).

Data analysis — The strengths of linear relationships among traits were de-termined using Pearson correlations and PICs ( Felsenstein, 1985 ) to account for nonindependence of data due to phylogenetic relationships. Phylogenetic inde-pendent contrasts were calculated using the module PDAP:PDTREE ( Midford et al., 2005 ) in Mesquite (version 2.73; Maddison and Maddison, 2009 ). We did not run independent PIC analyses for trees and shrubs because their actual evo-lutionary histories are interwoven and the growth forms do not generally fall into different clades. Splitting the phylogeny into two groups would not address the underlying evolutionary dynamics, so we analyzed the effect of phyloge-netic history in a combined phylogeny using phylogenetic ANOVA. Both standard and phylogenetic ANOVAs were conducted to determine whether anatomy, wood density, and height differed between trees and shrubs. The null hypothesis was that the anatomical traits would be similar in trees and shrubs once the phylogenetic relationship had been taken into account. We created the empirical null distribution of F values for hypothesis testing by simulating evolution of pairs of traits using Brownian motion, and then calcu-lated the F ratios of the resulting 1000 simulated data sets. The empirical value was compared to the simulated data sets to assess statistical signifi -cance. Phylogenetic ANOVA was performed using the program GEIGER version 1.2-14 ( Harmon et al., 2008 ) on the randomly resolved tree already described.

Bivariate relationships in trees and shrubs were analyzed using standardized major axis regression (SMA), also called reduced major axis, or model II, re-gression. The SMA is more suitable than regression model I for examining the scaling relationship between two variables that are both measured with error ( Sokal and Rohlf, 1995 ). Besides, the SMA estimates the best line that de-scribes the bivariate scatter between two variables and is more appropriate than linear regression to describe how size variables are related (allometry; Warton et al., 2006 ). The SMA is most often applied on log-transformed variables. Tests for common slopes were conducted only when there was a signifi cant relationship ( P < 0.05) between variables in both trees and shrubs. If the com-mon slope was found (when the common slope test was not signifi cant), then the residual scores were compared to tests for differences in the elevation of the slopes between trees and shrubs. This test determines whether there is a change in the elevation, along the y -axis, in two groups with a common slope. SMA lines were fi tted on log-transformed variables using the R package SMATR ( Warton, 2007 ). Relationships among nontransformed variables are shown in online Appendix S3.

In addition, a PCA analysis was conducted to determine whether trees and shrubs occupy different regions of multivariate trait space. The traits used in the analysis were vessel diameter, density, fraction, K t , and maximum height. To determine whether growth form is well discriminated by trait distribution, a between-groups PCA (BPCA) analysis was performed ( Dol é dec and Chessel, 1987 ; Baty et al., 2006 ), and a Monte Carlo permutation test on the between-groups inertia percentage was conducted to assess statistical signifi cance ( Dol é dec and Chessel, 1987 ; Baty et al., 2006 ). Permutations were run 999 times and com-pared with the observed value. The test was performed using the R package ade4 ( Thioulouse et al., 1996 ; Chessel et al., 2004 ).

RESULTS

Mean differences between trees and shrubs — Wood density in trees (mean = 0.66, SE = 0.019) and shrubs (mean = 0.68, SE = 0.018) did not differ signifi cantly ( F 1, 198 =0.76, P > 0.05 for both standard and phylogenetic ANOVAs). Trees had a signifi -cantly larger mean vessel diameter (mean = 97.6 µ m, SE = 3.9)

918 American Journal of Botany [Vol. 98

Table 1. Standardized major axis regression relationships among wood density, vessel traits, and maximum height for trees and shrubs. Common slope tests were only calculated when the relationship for a given bivariate relationship was signifi cant ( P < 0.05) in both growth forms. When slopes were not heterogeneous ( P > 0.05), differences between elevations were tested. Data presented are slopes and 95% confi dence intervals (CI), r 2 , intercepts, and elevation tests.

Trees Shrubs Common slope test

Traits Slope (CI) r 2 Intercept Slope (CI) r 2 Intercept Likelihood ratio Elevation test

Vessel density Vessel diameter − 2.229 ( − 2.44, − 2.036) 0.72*** 5.55 − 2.284 ( − 2.706, − 1.928) 0.54*** 5.35 0.063 ns 6.414**Wood density Vessel diameter − 0.298 ( − 0.349, − 0.256) 0.18*** 0.79 0.265 (0.206, 0.339) 0.002 ns − 0.18 Vessel density 0.134 (0.115, 0.156) 0.19*** 0.05 − 0.116 ( − 0.149, − 0.09) 0.002 ns 0.43 Vessel fraction 2.852 (2.405, 3.382) 0.001 ns 0.09 − 2.414 ( − 3.095, − 1.883) 0.001ns 0.29 K t − 0.074 ( − 0.086, − 0.063) 0.17*** 0.46 0.063 (0.049, 0.081) 0.004ns 0.11 Maximum height − 0.292 ( − 0.345, − 0.247) 0.04* 0.58 0.108 (0.084, 0.139) 0.006 ns 0.17Maximum height Vessel diameter 1.023 (0.877, 1.194) 0.19*** − 0.76 2.441 (1.929, 3.089) 0.11** − 3.3 34.84*** Vessel density − 0.459 ( − 0.535, − 0.393) 0.18*** 1.79 − 1.069 ( − 1.365, − 0.837) 0.04ns 2.41 Vessel fraction 9.771 (8.239, 11.588) 0.002 ns 0.79 22.274 (17.412, 28.493) 0.024ns − 0.22 K t 0.2525 (0.216, 0.295) 0.18*** 0.40 0.583 (0.459, 0.739) 0.094* − 0.61 31.99***

Notes: *** P < 0.001; ** P < 0.01; * P < 0.05; ns = not signifi cant

second components in the principal component axes ( Fig. 2 ). Trees had larger vessels, higher K t , lower vessel density, and (not surprisingly) were taller than shrubs. The Monte Carlo per-mutation test of the ratio of between-class and total inertia in the BPCA analysis confi rmed that growth forms were well dis-criminated by these traits (ratio = 0.30, P < 0.001; Fig. 2 ).

The fi rst three principal components explained more than 87% of the variation. Vessel diameter, vessel density, height, and K t had highest loadings on the fi rst principal component; vessel fraction and wood density had higher loadings in the second and third principal components, respectively. These trends paralleled those found with the SMA regression ( Fig. 1 ) and correlation analyses ( Table 2 ) in that vessel traits, wood density, and height were significantly correlated in trees, but in shrubs wood density was independent of vessel traits ( Tables 1, 2 ).

DISCUSSION

We hypothesized that because shrubs tend to live in colder, dryer, or darker environments, they would have denser wood than trees. We saw no a priori reason to expect that scaling be-tween vessel dimensions and wood density would differ be-tween the two growth forms. Confl icting results regarding this relationship have been reported, but growth form has not been previously considered as a contributing factor. Neither of our predictions held. Standard and phylogenetic ANOVAs showed that there is not a signifi cant difference in mean wood densities between shrubs and trees, but surprisingly, xylem traits scaled with wood density differently in each growth form.

Trees and shrubs often differ in their ecological requirements along several dimensions of resource availability (water, light, and soil nutrients), and we suggest these ecological differences are linked to the occupation of different regions of multivariate xylem trait space by trees and shrubs. Interestingly, for most traits, shrubs showed smaller exploration of trait space than trees, but integration among traits in shrubs (correlation) was lower. The results from the standard and phylogenetic ANOVAs presented here indicate that trees had signifi cantly higher vessel diameter, and thus higher conduction capacity and vessel fraction

(only with standard ANOVA) and lower vessel density. Some authors have shown ( Carlquist and Hoekman, 1985 ) that trees have larger vessels and lower vessel density than do shrubs; this larger vessel size allows trees, all else being equal, to have a higher water conduction capacity that correlates positively with height.

Preston et al. (2006) observed low levels of correlated evolu-tion between vessel traits and wood density. They suggested that wood density could be kept constant while vessel density and ves-sel diameters may change in opposite directions (due to its very tight negative correlation); therefore, vessel traits could poten-tially respond to ecological pressures independently of wood den-sity. We observed such decoupling in shrubs, but not in trees.

Tighter linear associations among height, wood density, and vessel traits suggest that evolution of trees is accompanied by a high degree of trait integration. Taller stature places a greater premium on effi cient conduction, in part because higher car-bon fi xation necessary to optimize growth is generally associ-ated with high transpiration rates. Concomitantly, a lower investment in cell wall material (i.e., wood density) is ex-pected as conduction capacity increases (vessel diameter). The relationship among vessel traits and wood density is not via vessel fraction. In shrubs, it has been found that more space occupied by vessels does not necessarily lower wood density because the latter is mainly driven by fi ber size, which deter-mines fi ber wall thickness and wall-to-lumen ratio ( Mart í nez-Cabrera et al., 2009 ). Our previous work showed mean and maximum vessel diameters, hydraulic diameter, and vessel density to be independent from density in shrubs ( Mart í nez-Cabrera et al., 2009 ). In trees, wood density and vessel traits are coordinated because low investment in fi ber cell walls, the main determinant of wood density, should be coupled with high water transport capacity via higher vessel diameter. This is in agreement with the fi ndings of Zanne et al. (2010) who decomposed the relationship between vessel diameter and ves-sel density into a vessel lumen fraction (the product of vessel diameter and vessel density) and a vessel composition index (vessel size divided by vessel number) to fi nd that the latter metric is more closely related to potential conductivity than ves-sel fraction because it captures the disproportionate contribution of larger vessels to conduction ( Zanne et al., 2010 ; Zanne and

919May 2011] Mart í nez-Cabrera et al. — Trait integration in trees and shrubs

Fig. 1. Standardized major axis (SMA) relationships between (A) vessel density and vessel diameter; between wood density and (B) vessel diameter, (C) vessel density, (D) vessel fraction, and (E) maximum height; and between maximum height and (F) vessel diameter, (G) vessel density, and (H) vessel fraction. All values were log-transformed. Solid lines represent the SMA slopes of the combined data set (not included in Fig. 1A ). Dashed and dotted lines indicate SMA slopes of shrubs and trees, respectively. Solid circles are shrubs and open circles trees. Slopes and intercept values and test for signifi cance differences for these parameters in trees and shrubs are given in Table 2 . Correlation coeffi cients are given in Table 1 . Variables are log transformed.

Falster, 2010 ). Neither of these metrics correlated with wood density, but they did not distinguish between growth forms ( Zanne et al., 2010 ).

Evolution of trees as a growth form clearly refl ects a strat-egy of competition for available light when other resources are not limiting. It has been shown that expenditure of carbon

920 American Journal of Botany [Vol. 98

Fig. 2. (A) Plot of the between-class principal component analysis (PCA) showing the ellipses and gravity center of trees and shrubs. Ellipsoids rep-resent the distribution around the trees and shrubs per class centers of gravity. Insert: histogram of simulated values for the between-groups PCA (BPCA) and the observed value. Sim = ratio of between class and total inertia.

Table 2. Pearson and phylogenetically independent contrast (PIC) correlation coeffi cients for the relationship between wood density and maximum height and vessel traits.

Traits All ( r ) Trees ( r ) Shrubs ( r ) PIC ( r )

Wood density Vessel diameter − 0.28*** − 0.42*** 0.04 ns − 0.34*** Vessel density 0.32*** 0.43*** − 0.04 ns 0.40*** Vessel fraction − 0.02 ns 0.01 ns 0.08 ns 0.03 ns K t − 0.28*** − 0.41*** 0.06 ns − 0.34*** Maximum height − 0.21** − 0.23** − 0.08 ns − 0.13 nsMaximum height Vessel diameter 0.72*** 0.43*** 0.33** 0.59*** Vessel density − 0.63*** − 0.43*** − 0.20 ns − 0.52*** Vessel fraction 0.24** 0.04 ns 0.15 ns 0.21** K t 0.69*** 0.41*** 0.31** 0.57***

Notes: *** P < 0.001; ** P < 0.01; * P < 0.05; ns = not signifi cant.

in stems (height growth) is favored by competition for light ( Iwasa et al., 1985 ; Loehle, 2000 ; Falster and Westoby, 2003 ). High wood density is associated with higher survival and shade tolerance in low light ( Falster and Westoby, 2005a ). At one extreme are small trees with high wood density and rela-tively lower conduction capacity; at the opposite end, tall trees tend to have lower wood density and higher conduction capacity. In a tropical wet forest, slow-growing species are for the most part shorter ( Condit et al., 1993 ). Species with low wood densities also have faster growth rates for height and diameter ( Arets et al., 2003 ; Muller-Landau, 2004 ) be-cause less biomass is required to build the same amount of tissue volume ( King et al., 2005 ). These strategies are also related to life history traits: the fi rst strategy involves slow

growth and high survival in low light environments; the sec-ond involves fast growth, low survival, and high light require-ments ( King, 1991 , 1996 ; Thomas and Bazzaz, 1999 ; Poorter et al., 2003 ).

According to our results, such coordination appears to be lacking in shrubs because these traits have different degrees of independence among them. This opens up the question what shrubs might gain from increasing independence between me-chanical and hydraulic traits. Evolutionary trait dissociation has been used to explain functional and morphological diversi-fi cation in certain groups such as fi shes ( Liem, 1973 ; Hulsey et al., 2006 ), mammals ( Stock, 2001 ), and rotifers ( Vermeij, 1973 ). Reduction in trait integration is thought to promote trait variation and more liberty to evolve, potentially leading to the occupation of new morphospaces ( Liem, 1973 ). It is therefore possible that the broad range of ecological habitats occupied by shrubs could be the result of this fl exibility of trait integra-tion that may have facilitated the occupancy of low resource niches via a higher variety of trait combinations. In addition, it seems likely that shrubs explore a wider range of integration among other wood traits, such as patterns of vessel grouping and spatial arrangements of living cells and fi bers. For exam-ple, Waisel et al. (1972) suggested that shrubs may be defi ned as having a high degree of sectoriality in their hydraulic systems.

This work is the fi rst broad-scale study comparing vessel traits, wood density, and height in trees and shrubs. A potential shortcoming of our analysis is that the data for trees and shrubs came from different localities, albeit from sites with overlap-ping environmental conditions. We present our analysis as a fi rst step toward a better understanding of functional wood traits in trees and shrubs.

921May 2011] Mart í nez-Cabrera et al. — Trait integration in trees and shrubs

Garland , T. Jr. , P. H. Harvey , and A. R. Ives . 1992 . Procedures for the analysis of comparative data using phylogenetic independent con-trast. Systematic Biology 41 : 18 – 32 .

Garland , T. Jr. , R. B. Huey , and A. F. Bennett . 1991 . Phylogeny and thermal physiology in lizards: A reanalysis. Evolution 45 : 1969 – 1975 .

Gartner , B. L. 1989 . Breakage and regrowth of Piper species in rain forest understorey. Biotropica 21 : 303 – 307 .

Goertzen , L. R. , J. J. Cannone , R. R. Gutell , and R. K. Jansen . 2003 . ITS secondary structure derived from comparative analysis: Implications for sequence alignment and phylogeny of the Asteraceae. Molecular Phylogenetics and Evolution 29 : 216 – 234 .

Hacke , U. G. , J. S. Sperry , W. T. Pockman , S. D. Davis , and K. A. McCulloh . 2001 . Trends in wood density and structure are linked to prevention of xylem implosion by negative pressure. Oecologia 126 : 457 – 461 .

Harmon , L. J. , J. T. Weir , C. D. Brock , R. E. Glor , and W. Challenger . 2008 . GEIGER: Investigating evolutionary radiations. Bioinformatics 24 : 129 – 131 .

Hulsey , C. D. , F. J. Garc í a de Le ó n , and R. Rodiles-Hern á ndez . 2006 . Micro and macroevolutionary decoupling of cichlid jaws: A test of Liem ’ s key innovation hypothesis. Evolution 60 : 2096 – 2109 .

IAWA Committee . 1989 . List of microscopic features for hardwood identifi cation. International Association of Wood Anatomists Bulletin 10 : 219 – 329 .

Iwasa , Y. , D. Cohen , and J. A. Leon . 1985 . Tree height and crown shape, as results of competitive games. Journal of Theoretical Biology 112 : 279 – 298 .

Jacobsen, A. L., L. Agenbag, K. J. Esler, R. B. Pratt, F. W. Ewers, and S. D. Davis . 2007 . Xylem density, biomechanics and anatomi-cal traits correlate with water stress in 17 evergreen shrub species of the Mediterranean-type climate region of South Africa. Journal of Ecology 95: 171 – 183.

King , D. A. 1990 . The adaptive signifi cance of tree height. American Naturalist 135 : 809 – 828 .

King , D. A. 1991 . Tree allometry leaf size and adult tree size in old-growth forests of western Oregon. Tree Physiology 9 : 369 – 382 .

King , D. A. 1996 . Allometry and life history of tropical trees. Journal of Tropical Ecology 12 : 25 – 44 .

King , D. A. , S. J. Davies , M. N. Nur Supardi , and S. Tan . 2005 . Treee growth is related to light interception and wood density in two mixed dipterocarp forest of Malaysia. Functional Ecology 19 : 445 – 453 .

Kolb , K. J. , and J. S. Sperry . 1999 . Differences in drought adaptation between subspecies of sagebrush ( Artemisia tridentata ). Ecology 80 : 2373 – 2384 .

Lia , V. V. , V. A. Confalonieri , C. I. Comas , and J. H. Hunziker . 2001 . Molecular phylogeny of Larrea and its allies (Zygophyllaceae): Reticulate evolution and the probable time of creosote bush arrival to North America. Molecular Phylogenetics and Evolution 21 : 309 – 320 .

Liem , K. F. 1973 . Evolutionary strategies and morphological innovations: Cichlid pharyngeal jaws. Systematic Zoology 22 : 425 – 441 .

Loehle , C. 2000 . Strategy space and the disturbance spectrum: A life-history model for tree species coexistence. American Naturalist 156 : 14 – 33 .

Maddison , W. P. , and D. R. Maddison . 2009 . Mesquite: A modular system for evolutionary analysis, version 2.73 [computer program]. Website http://mesquiteproject.org/mesquite.html.

Mart í nez-Cabrera , H. I. , and S. R. S. Cevallos-Ferriz . 2008 . Palaeoecology of the Miocene El Cien Formation (Mexico) as deter-mined from wood anatomical characters. Review of Palaeobotany and Palynology 150 : 154 – 167 .

Mart í nez-Cabrera , H. I. , C. S. Jones , S. Espino , and H. J. Schenk . 2009 . Wood anatomy and wood density in shrubs: Responses to varying aridity along transcontinental transects. American Journal of Botany 96 : 1388 – 1398 .

Martins , T. R. , and T. J. Barkmani . 2005 . Reconstruction of Solanaceae phylogeny using the nuclear gene SAMT. Systematic Botany 30 : 435 – 447 .

LITERATURE CITED Aiba , M. , and T. Nakashizuka . 2009 . Architectural differences associ-

ated with adult stature and wood density in 30 temperate tree species. Functional Ecology 23 : 265 – 273 .

Alverson , W. S. , B. A. Whitlock , R. Nyffeler , C. Bayer , and D. A. Baum . 1999 . Phylogeny of the core Malvales: Evidence from ndhF sequence data. American Journal of Botany 86 : 1474 – 1486 .

Arets , E. , P. van der Hout , and R. Zagt . 2003 . Responses of tree populations and forest composition to selective logging in Guyana. In H. ter Steege [ed.], Long-term changes in tropical tree diversity: Studies from the Guiana Shield, Africa, Borneo, and Melanesia, 95 – 116. Tropenbos International, Wageningen, Netherlands.

Barajas-Morales , J. 1987 . Wood specifi c gravity in species from two tropical forests in Mexico. International Association of Wood Anatomists Bulletin 8 : 143 – 148 .

Baty , F. , M. Facompr é , J. Wiegand , J. Schwager , and M. H. Brutsche . 2006 . Analysis with respect to instrumental variables for the explo-ration of microarray data structures. BMC Bioinformatics 7 : 422 .

Baum , D. A. , S. D. Smith , S. Yen , W. S. Alverson , R. Nyffeler , B. A. Whitlock , and R. L. Oldham . 2004 . Phylogenetic relationships of Malvatheca (Bombacoideae and Malvoideae: Malvaceae sensu lato) as inferred from plastid DNA sequences. American Journal of Botany 91 : 1863 – 1871 .

Carlquist , S. 1980 . Further concepts in ecological wood anatomy, with comments on recent work in wood anatomy and evolution. Aliso 9 : 499 – 553 .

Carlquist , S. 2001 . Comparative wood anatomy: Systematic, ecologi-cal, and evolutionary aspects of dicotyledon wood. Springer-Verlag, Berlin, Germany.

Carlquist , S. , and D. A. Hoekman . 1985 . Ecological wood anatomy of the woody southern Californian fl ora. International Association of Wood Anatomists Bulletin 6 : 319 – 347 .

Chave , J. , D. Coomes , S. Jansen , S. L. Lewis , N. G. Swenson , and A. E. Zanne . 2009 . Towards a worldwide wood economics spectrum. Ecology Letters 12 : 351 – 366 .

Chessel , D. , A. B. Dufour , and J. Thioulouse . 2004 . The ade4 package-I: One-table methods. R News 4 : 5 – 10 .

Condit , R. , S. P. Hubbell , and R. B. Foster . 1993 . Mortality and growth of a commercial hardwood ‘ el cativo ’ , Prioria copaifera , in Panama. Forest Ecology and Management 62 : 107 – 122 .

Coomes , D. A. , S. Heathcote , E. R. Godfrey , J. J. Shepherd , and L. Sack . 2008 . Scaling of xylem vessels and veins within the leaves of oak species. Biology Letters 4 : 302 – 306 .

Curtis , P. S. , and D. D. Ackerly . 2008 . Introduction to a Virtual Special Issue on plant ecological strategy axes in leaf and wood traits. New Phytologist 179 : 901 – 903 .

Dol é dec , S. , and D. Chessel . 1987 . Rythmes saisonniers et compo-santes stationnelles en milieu aquatique I: Description d ’ un plan d ’ observations complet par projection de variables. Acta Oecologica . Oecologia Generalis 8 : 403 – 426 .

Ewers , F. W. , J. M. Ewers , A. L. Jacobsen , and J. L ó pez-Portillo . 2007 . Vessel redundancy: Modeling safety in numbers. International Association of Wood Anatomists Journal 28 : 373 – 388.

Falster , D. S. , and M. Westoby . 2003 . Plant height and evolutionary games. Trends in Ecology & Evolution 18 : 337 – 343 .

Falster , D. S. , and M. Westoby . 2005a . Alternative height strat-egies among 45 dicot rain forest species from tropical Queensland . Australian Journal of Ecology 93 : 521 – 535.

Falster , D. S. , and M. Westoby . 2005b . Tradeoffs between height growth rate, stem persistence and maximum height among plant spe-cies in a post-fi re succession. Oikos 111 : 57 – 66 .

Felsenstein , J. 1985 . Phylogenies and the comparative method. American Naturalist 125 : 1 – 15 .

Floyd , J. W. 2002 . Phylogenetic and biogeographic patterns in Gaylussacia (Ericaceae) based on morphological, nuclear DNA, and chloroplast DNA variation. Systematic Botany 27 : 99 – 115 .

Garland , T. Jr. , and R. D í az-Uriarte . 1999 . Polytomies and phyloge-netically independent contrasts: An examination of the bounded de-grees of freedom approach. Systematic Biology 48 : 547 – 558 .

922 American Journal of Botany

Stock , D. W. 2001 . The genetic basis of modularity in the development and evolution of the vertebrate dentition. Philosophical Transactions of the Royal Society, B, Biological Sciences 356 : 1633 – 1653 .

ter Steege , H. , and D. S. Hammond . 2001 . Character convergence, di-versity, and disturbance in tropical rain forest in Guyana. Ecology 82 : 3197 – 3212 .

Thioulouse , J. , D. Chessel , S. Dol é dec , and J. M. Olivier . 1996 . ADE-4: A multivariate analysis and graphical display software. Statistics and Computing 7 : 75 – 83 .

Thomas , D. S. , K. D. Montagu , and J. P. Conroy . 2004 . Changes in wood density of Eucalyptus camaldulensis due to temperature — The physiological link between water viscosity and wood anatomy. Forest Ecology and Management 193 : 157 – 165 .

Thomas , D. S. , K. D. Montagu , and J. P. Conroy . 2007 . Temperature effects on wood anatomy, wood density, photosynthesis and biomass par-titioning of Eucalyptus grandis seedlings. Tree Physiology 27 : 251 – 260 .

Thomas , S. C. 1996 . Asymptotic height as a predictor of growth and al-lometric characteristics in Malaysian rain forest trees. American Journal of Botany 83 : 556 – 566 .

Thomas , S. C. , and F. A. Bazzaz . 1999 . Asymptotic height as a predictor of photosynthetic characteristics in Malaysian rain forest trees. Ecology 80 : 1607 – 1622 .

Tyree , M. T. 2003 . Hydraulic limits on tree performance: Transpiration, carbon gain and growth of trees. Trees 17 : 95 – 100 .

Tyree , M. T. , and F. W. Ewers . 1991 . The hydraulic architecture of trees and other woody plants. New Phytologist 119 : 345 – 360 .

Urbatsch , L. E. , B. G. Baldwin , and M. J. Donoghue . 2000 . Phylogeny of the conefl owers and relatives (Heliantheae: Asteraceae) based on nu-clear rDNA internal transcribed spacer (ITS) sequences and chloroplast DNA restriction site data. Systematic Botany 25 : 539 – 565 .

Urbatsch , L. E. , R. P. Roberts , and V. Karaman . 2003 . Phylogenetic evaluation of Xylothamia , Gundlachia , and related genera (Asteraceae, Astereae) based on ETS and ITS nrDNA sequence data. American Journal of Botany 90 : 634 – 649 .

van Gelder , H. A. , L. Poorter , and F. J. Sterck . 2006 . Wood mechanics, allometry, and life-history variation in a tropical rain forest tree commu-nity. New Phytologist 171 : 367 – 378 .

Vermeij , G. J. 1973 . Adaptation, versatility, and evolution. Systematic Zoology 22 : 466 – 477 .

Waisel , Y. , N. Liphschitz , and Z. Kuller . 1972 . Patterns of water move-ment in trees and shrubs. Ecology 53 : 520 – 523 .

Warton , D. 2007 . Smatr: (Standardised) major axis estimation and test-ing routines [computer program]. Website http://cran.r-project.org/web/packages/smatr/.

Warton , D. , I. J. Wright , D. S. Falster , and M. Westoby . 2006 . Bivariate linefi tting methods for allometry. Biological Reviews of the Cambridge Philosophical Society 81 : 259 – 291 .

Webb , C. O. , D. D. Ackerly , and S. W. Kembel . 2007 . Phylocom: Software for the analysis of community phylogenetic structure and trait evolution, version 3.41. Bioinformatics 24: 2098 – 2100. [program downloadable at website http:// www.phylodiversity.net/phylocom/]

Webb , C. O. , and M. J. Donoghue . 2004 . Phylomatic: Tree retrieval for applied phlogenetics. Molecular Ecology Notes 5 : 181 – 183 .

Wikstrom , N. , V. Savolainen , and M. W. Chase . 2001 . Evolution of angiosperms: Calibrating the family tree. Proceedings of the Royal Society of London, B, Biological Sciences 268 : 2211 – 2220 .

Wojciechowski , M. F. , M. Lavin , and J. S. Sanderson . 2004 . A phy-logeny of legumes (Leguminosae) based on analysis of the plastid matK gene resolves many well-supported subclades within the family. American Journal of Botany 91 : 1846 – 1862 .

Zanne , A. E. , and D. S. Falster . 2010 . Plant functional traits — Link-ages among stem anatomy, plant performance and life history. New Phytologist 185 : 348 – 351 .

Zanne , A. E. , M. Westoby , D. S. Falster , D. D. Ackerly , S. R. Loarie , E. J. S. Arnold , and D. Coomes . 2010 . Angiosperm wood structure: Global patterns in vessel anatomy and their relation to wood density and potential conductivity. American Journal of Botany 97 : 207 – 215 .

Zimmerman , M. H. 1983 . Xylem structure and ascent of the sap. Springer-Verlag, Berlin, Germany.

McCulloh , K. A. , J. S. Sperry , and F. R. Adler . 2003 . Water trans-port in plants obeys Murray ’ s law. Nature 421 : 939 – 942 .

McCulloh , K. A. , J. S. Sperry , and F. R. Adler . 2004 . Murray ’ s law and the hydraulic vs. mechanic functioning of wood. Functional Ecology 18 : 931 – 938 .

Midford , P. E. , T. J. Garland Jr. , and W. P. Maddison . 2005 . PDAP package of Mesquite, version 1.06 [computer program]. Website http://mesquiteproject.org/pdap_mesquite /index.html.

Moles , A. , D. S. Falster , M. Leishman , and M. Westoby . 2004 . Small-seeded plants produce more seeds per square metre of canopy per year, but not per individual per lifetime. Journal of Ecology 92 : 384 – 396.

Moya Roque , R. , and M. Tomazelo-Filho . 2007 . Relationships be-tween anatomical features and intra-ring wood density profi les in Gmelina arborea applying X-ray densitometry. Cerne 13 : 384 – 392 .

Muller-Landau , H. C. 2004 . Interspecifi c and intersite variation in wood specifi c gravity of tropical trees. Biotropica 36 : 20 – 32 .

Naidoo , S. , A. Zbo n á k , and F. Ahmed . 2006 . The effect of moisture availability on wood density and vessel characteristics of Eucalyptus grandis in the warm temperate region of South Africa. In S. Kurjatko, J. K ú dela, and R. Lagana [eds.], Proceedings of the 5th International Symposium on Wood Structure and Properties, 117 – 122, 2006, Sliak, Slovakia. Arbora Publishers, Zvolen, Slovakia.

Olmstead , R. G. , and J. D. Palmer . 1992 . A chloroplast DNA phylog-eny of the Solanaceae: Subfamilial relationships and character evolu-tion. Annals of the Missouri Botanical Garden 79 : 346 – 360 .

Paradis , E. , J. Claude , and K. Strimmer . 2004 . APE: Analyses of phylogenetics and evolution in R language. Bioinformatics 20 : 289 – 290 .

Petit , R. J. , and A. Hampe . 2006 . Some evolutionary consequences of being a tree. Annual Review of Ecology, Evolution and Systematics 37 : 187 – 214 .

Poorter , L. , F. Bongers , F. J. Sterck , and H. W ö ll . 2003 . Architecture of 53 rain forest tree species of a tropical moist forest. Ecology 84 : 602 – 608 .

Poorter , L. , I. McDonald , A. Alarc ó n , E. Fichtler , J. C. Licona , M. Pe ñ a-Claros , F. Sterck , Z. Villegas , and U. Sass-Klaassen . 2009 . The importance of wood traits and hydraulic conductance for the performance and life history strategies of 42 rainforest tree spe-cies. New Phytologist 185 : 481 – 492 .

Preston , K. A. , W. K. Cornwell , and J. L. DeNoyer . 2006 . Wood den-sity and vessel traits as distinct correlates of ecological strategy in 51 California coast range angiosperms. New Phytologist 170 : 807 – 818 .

Purvis , A. P. , and T. Garland Jr . 1993 . Polytomies in comparative analyses of continuous characters. Systematic Biology 42 : 569 – 575 .

Putz , F. E. , and N. V. L. Brokaw . 1989 . Sprouting of broken trees on Barro Colorado Island, Panama. Ecology 70 : 508 – 512 .

Rao , R. V. , and D. P. Aebischer . 1997 . Latewood density in relation to wood fi bre diameter, wall thickness, and fi bre and vessel percentages in Quercus robur L. International Association of Wood Anatomists Journal 18 : 127 – 138 .

Rao , R. V. , S. Shashikala , P. Sreevani , V. Kothiyal , C. R. Sarma , and P. Lal . 2002 . Within tree variation in anatomical properties of some clones of Eucalyptus tereticornis Sm. Wood Science and Technology 36 : 271 – 285 .

Robinson , J. , and S. A. Harris . 2000 . A plastid DNA phylogeny of the genus Acacia miller (Acacieae, Leguminoseae). Botanical Journal of the Linnean Society 132 : 195 – 222 .

Ryan , G. M. , and B. J. Yoder . 1997 . Hydraulic limits to tree height and tree growth. BioScience 47 : 235 – 242 .

Schenk , H. J. , S. Espino , C. M. Goedhart , M. Nordenstahl , H. I. Martinez Cabrera , and C. S. Jones . 2008 . Hydraulic integration and shrub growth form linked across continental aridity gradients. Proceed-ings of the National Academy of Sciences, USA 105 : 11248 – 11253 .

Sokal , R. R. , and F. Rohlf . 1995 . Biometry: The principle and practice of statistics in biological research. W. H. Freeman, New York, New York, USA.

Sperry , J. S. , F. C. Meinzer , and K. A. McCulloh . 2008 . Safety and ef-fi ciency confl icts in hydraulic architecture: Scaling from tissues to trees. Plant, Cell & Environment 31 : 632 – 645 .