Stem responses to damage: the evolutionary ecology of Quercus species in contrasting fire regimes


  • Claudia Romero,

    1. Department of Botany and Zoology, University of Florida, Bartram 227, PO Box 118525, Gainesville, FL 32611-8525, USA
    Search for more papers by this author
  • Benjamin M. Bolker,

    1. Department of Botany and Zoology, University of Florida, Bartram 227, PO Box 118525, Gainesville, FL 32611-8525, USA
    Search for more papers by this author
  • Christine E. Edwards

    1. Department of Botany and Zoology, University of Florida, Bartram 227, PO Box 118525, Gainesville, FL 32611-8525, USA
    2. Florida Museum of Natural History, PO Box 117800, University of Florida, Gainesville, FL 32611-7800, USA
    Search for more papers by this author

Author for correspondence:
Claudia Romero
Tel: +1 352 392 1468/1175
Fax:+1 352 392 3993


  • • The ability of tree stems to recover from damage is critical for tree survival and may explain species distributions across disturbance regimes. Two primary responses to stem damage, decay compartmentalization and wound closure, act in concert to limit decay and pathogen spread. A previous study demonstrated a tradeoff between wound responses that varied with anatomical traits, but its wide taxonomic range made it hard to analyze responses in an evolutionary context.
  • • Here, we tested the stem wound responses of 13 species of Quercus inhabiting three habitats across a gradient of fire intensity. We also quantified anatomical and structural traits and phylogenetic position, in order to assess the relative contributions of ecological adaptation and phylogenetic history in determining traits.
  • • Xylem anatomical traits were phylogenetically constrained, while phloem traits and damage responses varied with habitat. Across habitats, hammock and sandhill species closed bark wounds effectively, whereas scrub species limited the spread of xylem decay. There was a tradeoff between wound closure and decay compartmentalization within the white+live oaks.
  • • The fact that some wound response traits are phylogenetically constrained while others respond to ecological pressures suggests that damage responses integrate mechanisms operating at several levels within plants.


Given the biomechanical and physiological importance of tree stems and the frequency with which they are damaged by fire, herbivores, wind, and other agents, tree species have presumably acquired stem defense characteristics appropriate for the disturbance regimes in which they evolved. Some traits protect trees from damage while others allow damaged trees to close wounds and constrain pathogens and decay-producing organisms from spreading into adjacent tissues. Although many researchers have described the responses of tree stems to damage (Shigo, 1984; Pearce, 1990; White & Kile, 1993; Minore & Weatherly, 1994; Guariguata & Gilbert, 1996), the ecological forces and evolutionary constraints shaping these responses have received little attention. In this paper, we examine the relationship between two important responses to stem damage: wound closure and xylem decay compartmentalization. We evaluate how these responses depend on stem structural and anatomical traits in a group of related tree species that grow in habitats with contrasting disturbance regimes (i.e. differing fire frequencies and intensities).

Responses of tree stems to damage depend on the extent of damage and on which tissues are involved but also vary among species. Unless the damage is superficial and fails to penetrate beyond the last-formed periderm, the two main responses to stem damage are the closing of wounds by tissues produced by the vascular cambium and de-differentiated xylem parenchyma cells (i.e. callus formation) and the compartmentalization of decay in the xylem. Seven Bolivian tree species that inhabit the same forest type but that have contrasting stem characteristics (i.e. wood density, xylem and phloem anatomy, and bark thickness) varied in their responses to damage in ways that could be explained by anatomical and structural characteristics of stem tissues (Romero & Bolker, 2008). In particular, species with high wound closure rates but poor xylem decay compartmentalization had widely dilated phloem rays, low wood density, and wide xylem vessels. Contrary to the suggestion that species that close wounds rapidly should also suffer less extensive spread of xylem decay (Leben, 1985; Vasiliauskas & Stenlid, 1998), a tradeoff was evident: tree species that characteristically closed wounds rapidly failed to effectively compartmentalize xylem decay and vice versa (Romero & Bolker, 2008).

To explore further the relationship between anatomical/structural traits and responses to stem damage and to assess whether the tradeoff between wound closure and decay control is maintained across environments with contrasting disturbance regimes, we studied 13 species of oaks (Quercus) that characteristically inhabit one of three habitats in Florida (sandhill, scrub, and hammock). These habitats differ in soil moisture and fertility, but they differ most markedly in their fire regimes (Myers & Ewel, 1990). The 13 oak species represent three subclades (i.e. the red, white, and live oaks; Cavender-Bares et al., 2004a,b). Because species in each subclade occupy contrasting habitats, this group provides an excellent opportunity to explore how phylogenetic relatedness and ecological conditions are associated with particular stem traits and responses to stem damage.

Previous studies have shown that phylogenetic insights help to explain the distribution of ecological traits (Pausas & Verdu, 2005). Of particular relevance to the current study, Cavender-Bares et al. (2004a,b, 2006) examined foliar and xylem traits of many of the same Quercus species in some of the same sites we used in this study. Their research revealed functional traits (e.g. vulnerability to freezing, acorn maturation time, wood density, and average vessel diameters) that displayed either evolutionary convergence or conservatism. By interpreting patterns of species distribution across three different habitats, the authors concluded that closely related species did not co-occur in the same habitats, thereby displaying apparent phylogenetic repulsion (Cavender-Bares et al., 2004b, 2006). By contrast, in a study in Spain, Villar-Salvador et al. (1997) reported that three closely related species of Quercus coexist but display contrasting morphological and physiological strategies that presumably help them to cope with similar habitat constraints.

Given that closely related species are often anatomically and structurally similar (Carlquist, 1975; Baas, 1983), we expected them to show similar responses to stem damage (i.e. similar rates of bark wound closure and ability to control the advance of xylem decay). Supporting this expectation is the finding that similar stem structure and wound responses were observed in two co-occurring species of Bolivian Bombacaceae (Romero & Bolker, 2008). On the other hand, closely related species might differ in stem traits and capacities for xylem decay compartmentalization as a result of the influence of environmental conditions; in our case, especially to the inter-habitat differences between fire regimes. Many stem anatomical traits are phenotypically plastic (Baas et al., 1983), which may allow species to persist in a range of environments (Cavender-Bares & Holbrook, 2001). Thus, when possible, we sampled species across different sites to attempt to capture this variation. To infer patterns of correlated change in responses to stem damage, we constructed a phylogenetic hypothesis based on molecular data (Cavender-Bares et al., 2004a,b) and analyzed our data using comparative methods (Felsenstein, 1985).

This study tests the following hypotheses:

  • • there is a tradeoff between a species’ ability to compartmentalize xylem decay and to close wounds;
  • • rates of wound closure and xylem decay compartmentalization are correlated with anatomical and structural characteristics of the phloem and xylem, respectively;
  • • similarities in xylem/phloem traits in closely related species that occupy different habitats reflect phylogenetic trait conservatism.

Materials and Methods

The 13 oak species studied are distributed along a fire intensity gradient in three distinct habitats in north central Florida (Fig. 1; Myers & Ewel, 1990; Cavender-Bares et al., 2004a). The habitats are: hammock (six species); sandhill (four species); and scrub (three species). Species assignments to particular habitats were based on the literature (Kurz & Godfrey, 1962; Godfrey, 1988; Myers & Ewel, 1990; Abrams, 1996; Cavender-Bares et al., 2004a). Samples were obtained from several sites in Paynes Prairie State Preserve, Ordway Swisher Biological Station, San Felasco State Park, and Flamingo Hammock Landtrust, all of which are within 25 km of Gainesville. A hammock is a closed canopy forest with a high diversity of deciduous and evergreen broadleaved tree species (Platt & Schwartz, 1990). Hammocks are most characteristic of sites with relatively fertile and mesic soils. By contrast, a sandhill is a savanna or woodland habitat dominated by a canopy of Pinus palustris with a diverse array of grasses and forbs in the ground layer vegetation. Sandhills develop on a wide range of soil types but persist only where fires are frequent (return interval of 2–5 yr) and of low intensity (Myers, 1990). Scrub, an ecosystem nearly endemic to Florida, is dominated by xerophytic, woody shrubs, and develops on well-drained infertile sands; 40–60% of its plant species are endemic (Myers, 1990). Scrub is maintained by intensive stand-replacing fires that occur at 10–60 yr intervals (Menges & Root, 2004). Scrub fires top-kill all of the vegetation, but most of the woody species re-sprout from stumps, rhizomes, or roots. Among the ecological conditions that favor one vegetation type over another (e.g. soil characteristics: Kalisz & Stone, 1984; allelopathic effects: Hollis et al., 1982; and successional processes: Monk, 1968), fire regime and fire history are pivotal in determining habitat distributions (Menges & Hawkes, 1998).

Figure 1.

Maximum-likelihood phylogram showing subclade and habitat membership for 13 species of Florida Quercus. Bootstrap values > 50% are noted above branches. Habitats are represented as coded circles as follows: open, hammock; black, scrub; gray, sandhill. Branch lengths are scaled to number of substitutions per site.

The experimental damage treatment involved removing a small rectangular patch (8 cm2) of bark with a hammer and chisel from 10 young adult trees 5–9 cm DBH (stem diameter at 1.4 m above the ground) of each species. Bark removal and xylem exposure are the most common damage responses after fire occurrence. Two years later the damaged trees were destructively harvested to measure proportion of bark wound closure (WOCL) and to determine the horizontal (ALD) and vertical extents of xylem decay (UP and DOWN). To quantify the area of xylem decay, stem cross-sections were made with a chainsaw at intervals of 4–5 cm above and below the damage. On all sections with evidence of decay or discoloration, the affected area was traced on acetate sheets and later measured on a leaf area meter (LI-3100C Area Meter, Li-Cor Biosciences, Lincoln, NE, USA).

Anatomical/structural traits

The density (dry mass per fresh volume in g cm−3; n = 10 samples per species) of the wood and bark (inner and outer bark combined) were determined by measuring water displacement in a graduated cylinder and weighing the samples after drying for 72 h at 80°C. Bark thickness was evaluated using the contour method developed by Adams & Jackson (1995).

For measurement of anatomical characteristics, samples of xylem and phloem (n = 3 sections per individual; three individuals per species) were preserved in the field in a mixture of glycerine, ethanol (95%) and water (1 : 1 : 1; Terrazas & Arias, 2002). Within 2 wk the samples were sectioned (20–30 µm) with a sliding microtome. Temporary mounts were stained with toluidine blue (pH 2.2; O’Brien & McCully, 1981), mounted in glycerine, and measured using a calibrated stage micrometer. For the xylem, we measured maximum vessel lumen diameter, vessel frequency, ray width, and distance between rays. For the phloem we measured ray width and inter-ray distances. Bark terminology follows Trockenbrodt (1990) and Junikka (1994).

Phylogenetic reconstruction

The 13 species studied belong to three subclades within the genus Quercus, as determined by phylogenetic analysis of gene sequences from the ITS region of nuclear ribosomal DNA (Manos et al., 1999; Manos & Stanford, 2001; Cavender-Bares et al., 2004b; Fig. 1). To reconstruct the phylogeny we downloaded the aligned data matrix of Cavender-Bares et al. (2004a) from TreeBASE (, removed four species not included in this study (Q. pumila Walter, Q. minima (Sarg.) Small, Q. stellata Wangenh., and Q. laurifolia Michx.), and visually realigned the matrix using Se-Al (Rambaut, 1996). Two species were not included because they grow as small stoloniferous shrubs with very thin stems (Q. pumila and Q. minima), and populations of the remaining two species were not found within the geographical range of the study. We reconstructed the phylogeny of three different data sets: (i) all 13 Quercus species; (ii) only white and live oaks (white oaks: Q. margaretta Ashe ex Small, Q. michauxii Nutt., Q. chapmanii Sarg., and Q. austrina Small; live oaks: Q. virginiana Mill. and Q. geminata Small); and (iii) only red oaks (Q. falcata Michx., Q. laevis Walter, Q. incana W. Bartram, Q. nigra L., Q. hemisphaerica Bartram, Q. shumardii Buckley and Q. myrtifolia Willd.). For phylogeny reconstruction, we carried out maximum-likelihood analyses using PAUP* 4.0b10 (Swofford, 2001), using the optimal model of evolution and parameters chosen for each data set by the Akaike information criterion (AIC) as implemented in Modeltest (Posada & Crandall, 1998). The model selected for all data sets was the GTR+I model.

We carried out branch and bound searches and saved the most likely tree (MLT) with branch lengths resulting from each analysis. Bootstrap analyses (1000 replicates; Felsenstein, 1985) were used to assess support for the proposed relationships. These trees were then used to carry out phylogenetically independent contrasts (PICs; see following section).

Data analyses

Separate principal components analyses (PCAs) were performed on xylem and phloem anatomical/structural data. The phloem PCA also included total bark thickness and % inner bark out of total bark. To establish how species, subclades, and species assemblages in the three habitats differ in xylem and phloem anatomy and structure, two-way ANOVAs with habitat and subclade as fixed effects were run on the PCA scores separately for xylem and phloem whenever those axes explained at least 30% of the total variation. We also determined whether habitat or subclade membership explained interspecific variation using univariate ANOVAs of anatomical/structural traits.

To determine response variable differences as a function of habitat and subclade membership, we utilized phylogenetically corrected ANOVAs using our most likely phylogenetic trees (see earlier) and the GEE method (generalized estimating equations; Paradis & Claude, 2002) within the APE package in R (Paradis et al., 2004; R Development Core Team, 2005). This method is appropriate for use with comparative data when species are nonindependent (Grafen, 1989).

Tests for trait correlations were examined with phylogenetically independent contrasts (Felsenstein, 1985) at the species and subclade levels. At the species level, we used a data set composed of all 13 taxa. For trait correlations at the subclade level, instead of designating three subclades (i.e. red, white and live oaks), we combined the closely related white and live oak subclades into a single subclade, hereafter referred to as white+live. This combination was necessary because the live oak subclade was only represented by two species in this study. Furthermore, the white oak subclade (Quercus sensu strictu) includes the live oaks (subsection Virentes), and the two may be sister taxa (Manos et al., 1999; Fig. 1). We thus carried out separate trait correlations using the data sets and phylogenies composed (i) only of the red oaks; and (ii) the white and live oaks combined.

To elucidate the relationships between anatomical/structural traits of xylem and phloem with the wounding response variables, we utilized the GEE method. These regressions were performed using all species, at the subclade and habitat levels. ANCOVAs were run to detect the effect of subclade or habitat on decay variable responses and xylem anatomical PCA scores.

Trait conservatism was measured as the similarity of the matrices of the scores of the PCAs (separately for both xylem and phloem trait values) to the matrix of phylogenetic distances, and assessed with Mantel tests with 1000 permutations. These tests were carried out using the APE package in R (Paradis et al., 2004; R Development Core Team, 2005).

Phylogenetically independent contrast analyses were performed on alternative phylogenetic trees that consider different modes of trait evolution: six most parsimonious trees, one of which was identical to the MLT we report, an ultrametric tree and a tree which reflected a constant rate of trait evolution. The ultrametric and three of the most parsimonious trees had consistent topology with the MLT, and overall, results of analyses were well supported. Thus we only present results of the MLT tree analyses.


Anatomical/structural traits

Xylem traits varied substantially among the 13 species. In particular, the red oaks had relatively low-density wood and large vessels, the white oaks displayed intermediate traits, and the live oaks had the densest wood and smallest vessels. These results are in contrast with those obtained in a previous study in which live oak vessel diameters were found to be similar in size to those in the red oak subclade (Cavender-Bares & Holbrook, 2001; Tables 1, 2).

Table 1.  Xylem and phloem anatomical traits of Quercus spp. in Florida
Vessel diameter (µm)Vessel density (number cm−2)Ray width (µm)Inter-ray distance (µm)Ray width (µm)Inter-ray distance (µm)
  1. Means based on measures of three sections/individual from three individuals/species. Species organized by subclade membership (red oaks (r), seven species; white oaks (w), four species; live oaks (l) two species). Standard deviations noted in parentheses.

Q. laevis (r)469 (30.9)15 (2.6)365 (41.2)149 (9.3)110 (25.4)118 (10.2)
Q. falcata (r)492 (38.2)16 (1.6)494 (42.1)122 (10.9)204 (7.1)112 (22.3)
Q. nigra (r)517 (132.2)13 (0.7)427 (23.7)174 (14.2)224 (76.6)206 (18.2)
Q. shumardii (r)569 (42.7)10 (4.8)488 (54.7)128 (14.3)266 (216.3)117 (15.4)
Q. myrtifolia (r)455 (62.2)17 (3.7)546 (13.5)134 (9.7)155 (24.6)113 (61.3)
Q. incana (r)470 (81.8)20 (5.5)375 (52.0)109 (14.8)218 (73.7)193 (92.3)
Q. hemisphaerica (r)474 (12.8)14 (0.3)454\ (37.7)193 (23.5)268 (63.8)325 (58.2)
Q. austrina (w)340 (21.5)12 (1.0)404 (44.1)140 (0)167 (22.5)106 (18.5)
Q. margaretta (w)345 (26.3) 9 (1.1)444 (74.0)113 (4.0)300 (44.4)203 (35.3)
Q. chapmanni (w)333 (42.0)20 (4.9)594 (21.5)140 (0)274 (116.3)175 (26.2)
Q. michauxii (w)332 (19.8)13 (1.3)368 (16.4)144 (4.7)145 (5.4)111 (15.9)
Q. geminata (l)273 (7) 9 (0)471 (77.1)106 (14.0)229 (219.9)141 (33.9)
Q. virginiana (l)286 (63.4)11 (3.0)323 (134.6) 95 (5.0)172 (34.1)160 (7.0)
Table 2.  Structural traits of xylem and phloem of Quercus spp. in Florida
SpeciesBark thickness (cm)Inner bark (%)Xylem density (g cm−3)Phloem density (g cm−3)
  1. Means of 10 individuals 5–9 cm DBH for bark thickness and proportion of inner bark. Species organized by subclade membership (red oaks (r), seven species; white oaks (w), four species; live oaks (l), two species). Standard deviations noted in parentheses.

Q. laevis (r)1.07 (0.2)63.1 (6.4)0.59 (0.06)0.79 (0.14)
Q. falcata (r)0.84 (0.3)44.0 (5.7)0.55 (0.03)0.55 (0.01)
Q. nigra (r)0.65 (0.1)64.3 (9.0)0.68 (0.05)0.64 (0.10)
Q. shumardii (r)0.32 (0.1)73.8 (6.0)0.60 (0.01)0.63 (0.02)
Q. myrtifolia (r)0.38 (0.1)71.0 (4.9)0.76 (0.02)0.58 (0.16)
Q. incana (r)1.06 (0.1)46.1 (5.3)0.63 (0.03)0.63 (0.01)
Q. hemisphaerica (r)0.30 (0.1)71.2 (4.2)0.67 (0.07)0.71 (0.02)
Q. austrina (w)0.47 (0.1)57.6 (5.3)0.72 (0.02)0.38 (0.01)
Q. margaretta (w)0.99 (0.1)44.7 (7.6)0.70 (0.01)0.55 (0.07)
Q. chapmanni (w)0.75 (0.2)46.7 (8.1)0.73 (0.02)0.57 (0.04)
Q. michauxii (w)0.43 (0.1)55.1 (5.5)0.68 (0.02)0.36 (0.04)
Q. geminata (l)0.88 (0.1)48.9 (11.0)0.79 (0.05)0.52 (0.06)
Q. virginiana (l)0.73 (0.1)44.4 (7.0)0.90 (0.06)0.65 (0.02)

Maximum vessel diameters ranged from 273 µm in Q. geminata to 569 µm in Q. shumardii. Q. geminata also had the lowest vessel density (9 vessels mm−2) whereas the highest vessel density (20 vessels mm−2) was found in Q. incana. Ray width ranged from 323 µm in Q. virginiana to 594 µm in Q. chapmanii. Xylem tissue density ranged from 0.52 g cm−3 in Q. laevis to 0.91 g cm−3 in Q. virginiana. Q. virginiana also had the most closely spaced xylem rays (94 µm), whereas the most widely separated rays (193 µm) were found in Q. hemisphaerica.

The first axis of the xylem anatomical/structural trait PCA explained 43% of the variation in these traits. Species differences were associated with vessel diameter and wood density; species with small vessels and low wood density were on the positive side of this axis. The second axis explained an additional 19% of the variation and was positively associated with high vessel density (Fig. 2).

Figure 2.

Results of a principal components analysis on xylem anatomical and structural traits of 13 species of Florida Quercus from north central Florida (n = 3 samples/individual; three individuals/species, except for wood density, see below). Characteristics included in this analysis are: mean maximum vessel diameter (VEDI) and frequency (VEDE); ray width (RAWI); distance between small rays (SMARADIST); and wood density (DENS; n = 10/species). Species abbreviations are located at the center of the ellipse formed by the individual observations. Species abbreviations are as follows: lae, Q. laevis; fal, Q. falcata; nig, Q. nigra; shu, Q. shumardii; myr, Q. myrtifolia; inc, Q. incana; hem, Q. hemisphaerica; aus, Q. austrina; mar, Q. margaretta; chap, Q. chapmanii; mic, Q. michauxii; gem, Q. geminata; vir, Q. virginiana. Subclade membership indicated as: L, live; R, red; W, white.

Phloem traits also differed markedly among species (Tables 1, 2). The thinnest rays were found in Q. laevis and the thickest in Q. margaretta (110 and 300 µm, respectively). Inter-ray distances ranged from 106 µm in Q. austrina to 325 µm in Q. hemisphaerica. Total bark thickness of the 5–9 cm DBH trees varied from 0.30 cm in Q. hemisphaerica to 1.08 cm in Q. laevis. The proportion of inner bark ranged from 44% in Q. falcata to 74% in Q. shumardii, which was a thin-barked species (Table 2). Finally, bark tissue density varied from 0.36 g cm−3 in Q. michauxii to 0.79 g cm−3 in Q. laevis.

The first axis of the phloem anatomical/structural trait PCA, which explained 33% of the variation, separated species with thick bark with a high proportion of inner bark from species with the opposite traits. The second axis, which explained 31% of the variation, was positively associated with distance between rays (Fig. 3).

Figure 3.

Results of a principal components analysis on phloem anatomical traits of 13 species of Florida Quercus (n = 3 samples/individual; three individuals/species except for phloem density, see below). Characteristics included in this analysis are: ray width (RAWI), inter-ray distance (RADIST), proportion of bark that is inner bark (INNERPROP), total bark thickness (TOTAL), and density (DENS; n = 10/species). Species abbreviations are located at the center of the ellipse formed by the individual observations. Species abbreviations are as follows: lae, Q. laevis; fal, Q. falcata; nig, Q. nigra; shu, Q. shumardii; myr, Q. myrtifolia; inc, Q. incana; hem, Q. hemisphaerica; aus, Q. austrina; mar, Q. margaretta; chap, Q. chapmanii; mic, Q. michauxii; gem, Q. geminata; vir, Q. virginiana. Subclade membership is indicated as: L, live; R, red; W, white.

Differences on the first PCA axis for xylem traits were significant among species (ANOVA: F12,38 = 10.27; P < 0.001), among subclades (two-way ANOVA: F2,38 = 14.12; P < 0.007), and within habitats (ANOVA: P < 0.005 for sandhill: df = 3,11; and hammock: df = 5,17). Scrub was omitted from the previous analysis because there is only one species from each subclade in this habitat. Two-way ANOVAs of the first PCA for xylem, with subclade and habitat as fixed levels, revealed no differences among habitats (F2,38 = 0.89; P = 0.45). All seven species in the red oak subclade were similar and had low wood density and large vessels (F6,38 = 0.66; P = 0.68), whereas white+live subclade species differed from each other (F5,38 = 8.67; P < 0.001).

Comparisons of the two PCA axes generated from the phloem anatomical/structural traits revealed significant differences among species on the first axis within subclades and among habitats (MANOVAsubclade F2,38 = 17.1; P < 0.004; MANOVAhabitat F2,38 = 6.2; P < 0.02), but not among subclades or within habitats. Species were located along a gradient from those with overall thin bark with a high proportion of inner bark (Q. hemisphaerica, Q. shumardii, Q. nigra, and Q. myrtifolia) to species with the opposite set of traits (Q. incana, Q. margaretta, Q. geminata, and Q. falcata). Barks were thickest on sandhill species, intermediate on scrub species, and thinnest on hammock species. Hammock and sandhill species also had contrasting mean proportions of inner bark (62 and 49%, respectively).

Responses to stem damage

Rates of wound closure and xylem decay spread varied widely among the 13 species (Table 3; Fig. 4). Comparisons among habitats showed that hammock species suffered significantly higher rates of radial and vertical spread of decay than scrub and sandhill species. By contrast, wound closure rates were similar in hammock and sandhill species and significantly higher than in scrub species (GEE; FALL VARIABLES = 32.08; P < 0.03; Table 4).

Table 3.  Wound closure and xylem decay spread 2 yr after experimental stem damage of 5–9 cm DBH Quercus trees
SpeciesWound closure (%)Xylem area lost to decay (cm2)Xylem decay spread up (cm)Xylem decay spread down (cm)
  1. Species organized by subclade membership (red oaks (r), seven species; white oaks (w), four species; live oak (l), two species). Means of 10 individuals per species. Standard deviations noted in parentheses.

Q. laevis (r)36.30 (31.2)2.01 (0.8)12.8 (8.7)9.9 (5.7)
Q. falcata (r)68.06 (27.8)2.18 (0.9)7.2 (4.1)12.0 (7.8)
Q. nigra (r)14.71 (23.6)2.76 (2.49)20.0 (10.4)19.1 (10.6)
Q. shumardii (r)40.70 (27.5)2.06 (0.9)13.2 (10.2)9.7 (8.5)
Q. myrtifolia (r)3.89 (1.5)2.86 (1.4)10.3 (7.9)10.0 (7.3)
Q. incana (r)73.10 (37.6)2.15 (0.86)11.4 (6.7)13.0 (7.0)
Q. hemisphaerica (r)53.96 (35.7)2.54 (1.0)25.5 (7.0)28.2 (12.5)
Q. austrina (w)35.29 (32.8)3.21 (1.5)20.5 (11.3)21.2 (6.6)
Q. margaretta (w)10.21 (13.9)2.30 (0.6)14.6 (8.2)16.0 (12.2)
Q. chapmanni (w)7.82 (5.8)2.59 (1.1)11.3 (2.8)11.2 (4.0)
Q. michauxii (w)86.58 (19.0)4.27 (2.5)23.1 (7.3)16.4 (6.0)
Q. geminata (l)10.78 (9.5)1.08 (0.5)11.5 (4.3)8.9 (3.3)
Q. virginiana (l)3.98 (7.1)2.52 (1.1)13.6 (4.9)14.9 (6.0)
Figure 4.

Responses to experimental stem damage of 13 species of Florida oaks (n = 10 individuals/species) in the three habitats studied. Decay variables are: area lost, basal area lost to decay (cm2); spread up, vertical spread of decay above the wound (cm); spread down, vertical spread of decay below the wound (cm); and, wound closure, proportion of wound closed within 2 yr. Subclade membership indicated by symbols: open triangles, white oaks; plus signs, red oaks; open circles, live oaks.

Table 4.  Phylogenetic ANOVA of variables related to responses to stem damage and habitat (hammock, sandhill, and scrub) and subclade membership (red, white, and live oaks)
Response variableLevelFP
  • *

    Significance at P < 0.05. Phylogenetic df = 4.95.

Wound closureHabitat18.970.05*
Xylem area lost to decayHabitat27.350.03*
Xylem decay spread upHabitat56.630.01*
Xylem decay spread downHabitat28.280.03*

At the subclade level, mean xylem cross-sectional area lost to decay (ALD) was highest in the white oaks and lowest in the live oaks. Xylem decay spread above the wound ranged from 7.2 cm in Q. falcata to 25.5 cm in Q. hemisphaerica, both red oaks. Xylem decay below the wound ranged from 8.9 cm in Q. geminata to 28.2 cm in Q. hemisphaerica. Subclades differed in radial but not vertical spread of decay, with live oaks having the lowest mean loss to radial decay (1.08 cm2 in Q. geminata) and white oaks having the highest value for this variable (4.27 cm2 in Q. michauxii).

Trait correlation

Phylogenetically independent contrasts across species showed a significant correlation between wound closure and upward spread of xylem decay (r = 0.59; P < 0.04), and a strong correlation between the two directions of vertical spread of decay (r = 0.87; P < 0.0001; Table 5). This pattern was not consistent when trait correlations were examined at the subclade level. Within the white+live subclade all decay variables were strongly correlated, as was the case for wound closure and vertical decay spread upwards. By contrast, the only significant correlation in the red oak subclade was between measures of vertical spread of decay (Table 5).

Table 5.  Results of a phylogenetic independent contrast of trait correlations using variables associated with responses to stem damage in 13 species of Quercus
SubcladeAll speciesWhite+liveRed
  1. df, degrees of freedom. A significant positive correlation indicates the existence of a tradeoff. P-value of the correlation coefficient noted in parentheses.*, Significance at P < 0.05.

Area lost to decay – wound closure0.41 (0.18)0.63 (0.17)−0.51 (0.30)
Area lost to decay – xylem decay spread up0.48 (0.10)0.85 (0.02)*0.03 (0.9)
Area lost to decay – xylem decay spread down0.43 (0.16)0.81 (0.04)*0.15 (0.76)
Wound closure – xylem decay spread up0.59 (0.04)*0.84 (0.03)*0.39 (0.43)
Wound closure – xylem decay spread down0.48 (0.10)0.31 (0.56)0.57 (0.23)
Xylem decay spread up – xylem decay spread down0.87 (0.0001)*0.75 (0.08)0.92 (0.008)*

Anatomical/structural traits and wounding responses

Phylogenetic regression analyses of anatomical/structural scores of the xylem PCA on all variables describing the spread of xylem decay were not significant. Likewise, there were no apparent effects of phloem anatomical/structural scores on rates of wound closure. ANCOVA revealed no effects of subclade or habitat membership on the relationship between wounding response variables and anatomical/structural traits.

Trait conservatism

Results from the Mantel test for scores of xylem traits show that traits were conserved through the phylogenetic tree (z = 7.87; P < 0.001). The results for the Mantel test on phloem scores were inconclusive for determining trait conservatism or convergence (z = 5.95; P = 0.4).


Responses to stem damage varied substantially among the 13 oak species studied in Florida and were influenced by habitat and, to a lesser extent, by phylogeny (i.e. only for xylem area lost to decay). The data suggest a tradeoff in the white+live oak subclade between rates of wound closure and xylem decay compartmentalization. Contrary to our expectations, xylem and phloem anatomical/structural traits were not correlated with responses to stem damage. Xylem traits were phylogenetically conserved, as previously reported for a partially overlapping set of traits (Cavender-Bares & Holbrook, 2001), but no conclusion regarding convergence or conservatism could be reached for phloem traits.

Anatomical/structural trait variation among habitats

Xylem characteristics are conserved according to subclade membership in the 13 oak species we studied. The discrepancy between our results and values reported in other studies might relate to our working with young adult stems instead of branches (Cavender-Bares & Holbrook, 2001). Furthermore, we measured mean maximum vessel diameter as opposed to mean vessel diameter because of the critical importance of maximum vessel diameter in conductive efficiency (Baas et al., 1983) and potentially in facilitating decay spread.

We detected no differences in mean xylem traits among habitats. Instead, we found substantial variation in xylem traits among species in a habitat, reflecting variation among species from differing subclades. Cavender-Bares et al. (2004b) similarly found that xylem traits were conserved within each Quercus subclade. Subclade membership reportedly influenced water relation-related traits (e.g. leaf habit: white oaks are deciduous, live oaks are evergreen and red oaks have species with both leaf habits; Cavender-Bares & Holbrook, 2001).

The ability of closely related species to live in habitats with a gradient in resource availability, despite the lack of variation in wood density and vessel diameter, may reflect plasticity in nonanatomical traits (e.g. leaf size and sapwood area; Cavender-Bares & Holbrook, 2001). It may also be attributable to small xylem trait differences within subclade traits that we did not measure. For example, species growing in drier habitats (e.g. scrub) may have shorter vessels than closely related species that inhabit more mesic environments (Zimmermann & Jeje, 1981). Short vessels are efficient at preventing embolism propagation (Baas et al., 1983), and might also explain the slower rate of decay spread observed in scrub species.

In contrast to the apparent lack of habitat influences on xylem traits, substantial differences in phloem anatomical/structural traits were found both within subclades and among habitats in the oaks studied, indicating that phloem traits are under strong selection in the three different habitats. Species were separated along an axis of thick bark with a high proportion of outer bark in sandhill habitats, where fires are frequent but not hot, to thin bark and a low proportion of outer bark in hammocks, where fires are mostly absent. The scrub species were intermediate in these traits because two species (Q. chapmanii and Q. geminata) have thick bark and high proportion of inner bark, whereas one (Q. myrtifolia) has the opposite set of traits. Although scrub is a fire-maintained habitat, variation in phloem traits in scrub species may result from the fact that scrub fires are generally of such high intensity that all trees are top-killed and thus phloem traits (e.g. thick bark) that confer resistance to damage from low-intensity fires may be of little value. Scrub species do not avoid fire damage to their stems but instead re-sprout from roots and rhizomes (Guerin, 1993; Greenberg & Simons, 1999).

One unusual species, Quercus laevis, has the typical thick bark of other sandhill species but with an uncharacteristically low proportion of outer bark (36% compared with a mean of 56% for the other three sandhill species; all-species range, 27–64%). While this difference is surprising, similar bark characteristics in Pinus ponderosa (i.e. thick inner bark with a high moisture content) were shown to be better fire insulators than thick dead outer bark (van Mantgem & Schwartz, 2003). Q. laevis, which has both thick inner bark and thick bark overall may thus benefit from both characteristics when exposed to the frequent fires characteristics of sandhills.

Responses to stem damage

Vertical spread of decay values were similar to those obtained in a study in the north American midwest with two species of the same genus (in 6–23 cm DBH individuals of Q. prinus L. (Q. montana Willd.) and Q. velutina Lam.; Smith & Sutherland, 1999). Although species in a subclade typically shared many anatomical and structural traits, interspecific differences in responses to stem wounding in the 13 oak species did not follow any apparent phylogenetic pattern. The only exception is radial extent of xylem decay, which was high in white oaks, low in live oaks and intermediate in red oaks. White and red subclade species also displayed rates of wound closure that were marginally higher than in live oaks (meanwhite = 35%; meanred = 41%; meanlive = 7.4; P = 0.09). Based on the larger xylem vessel diameter in red oaks and their reported lack of tyloses (Harlow, 1970; Core et al., 1979), we expected, but did not consistently find, that red oaks suffer higher rates of xylem decay than white and live oaks. While the red oaks we studied had the widest vessels of all subclades, at least one of them (Q. laevis) also develops tyloses (C. Romero, unpublished). Red oaks may also limit decay spread more than might be expected from their xylem vessel dimensions with phenolic compounds, both constitutive and induced (Smith, 1997). Radial rates of decay spread may also be influenced by differences in sectoriality (i.e. spatial heterogeneity within plants as a function of species-specific characteristics of the vascular system; Ellmore et al., 2006) among subclades, which might explain inter-subclade differences in rates of decay advance.

Instead of phylogenetic conservation in all decay response variables, we found significant differences among habitats. The ability of tree species to limit decay spread is not favored in hammock environments, although some species in this environment closed wounds rapidly. At the other extreme, scrub species suffered the lowest losses to decay of all habitats, but also showed the lowest wound closure rates. Overall, species in fire-prone habitats had low radial losses to decay but contrasting abilities to close wounds: high in sandhill and low in scrub (47 and 7.5%, respectively). The fact that xylem decay response variables varied according to habitat, even though xylem anatomical/structural traits were phylogenetically conserved and did not display variation among habitats, suggests that a species’ ability to respond to stem damage is decoupled from the anatomical and structural traits we measured (see later).

Response variable correlations

Some clear patterns emerged from the analyses of intertrait correlations among oak species. The hypothesized tradeoff between wound closure and compartmentalization of xylem decay was apparent in the white+live oak subclade (at least for decay spread upwards), but not in the red oaks. Within each subclade, species also varied substantially in these responses, particularly within the red oaks. In fact, some red oaks had at least one response to stem damage that could enhance stem survival (e.g. high WOCL: Q. incana, Q. hemispherica, Q. falcata; or low ALD: Q. laevis, Q. shumardii) while others demonstrated neither (e.g. Q. nigra, Q. myrtifolia). Moreover, the fact that species in the red oak subclade occupy all the habitats studied suggests that the ability of these species to adapt to stem-damaging disturbances, and thus their ability to persist in landscapes with particular disturbance regimes, might not involve any of the traits measured.

Anatomical/structural traits and wounding responses

The hypothesized relationship between xylem anatomical/structural characteristics and decay compartmentalization was not apparent in any of the subclades, although it was marginally significant in the white+live oak subclade (P = 0.08). Additionally, the relationship between phloem anatomical/structural traits and wound closure rates was not significant in any of the subclades. By contrast, in a previous study of Bolivian tree species, xylem and phloem traits were clearly related to bark wound closure rates and xylem decay compartmentalization ability (Romero & Bolker, 2008). Because this previous study included a wide taxonomic range of species growing in the same habitat, one explanation for the difference in results with this study is that interspecific differences within a genus might typically be smaller than across genera. In particular, the Quercus species studied vary relatively little among themselves and all lack the dilating rays that were so closely associated with rapid wound closure in the Bolivian species. Another explanation for the lack of association is that by focusing on fire, we have downplayed other habitat characteristics and other sorts of disturbances to which these trees are subjected (e.g. diseases) that may influence wound closure. Regardless, the present study on Florida oaks highlights the need for phylogenetic control in studies designed to reveal correlated evolution in functional traits related to species persistence, such as responses to stem damage.

The results of this study demonstrate that oak species vary in the mechanisms involved in responses to stem damage. While members of the white+live oak subclade rely to some extent on anatomical/structural features to deal with stem damage, the red oaks studied display an array of mechanisms depending on the habitat they occupy. A phylogenetically controlled study, similar to this, that focuses on constitutive and inducible defenses might help to elucidate constraints that species face to efficiently avoid the spread of xylem decay.

In conclusion, the observed influence of habitat on stem damage responses variables in closely related species suggests that these responses are of functional significance for species survival. Nevertheless, patterns of responses to stem damage in the Quercus species studied might be better understood if life-history traits are considered. In particular, re-sprouting is a well-studied characteristic of many species that occupy fire-prone habitats. While all oaks reportedly have the ability to re-sprout (Abrams, 1996), less is known about the origins of these sprouts (i.e. from roots, rhizomes, or stumps), their longevities and growth rates, and their role in population maintenance. Especially for stump sprouts, traits that serve to contain decay spread seem important to consider. The observed low rate of radial losses to decay in scrub species is consistent with this consideration.

The results of this study reveal the importance of building explanations about complex issues, such as responses to stem damage, by considering the relative roles of the several factors that affect these responses in particular (e.g. anatomy and structure; physiology; secondary chemistry; phylogeny; life history), and plant performance in general.


The authors thank the Florida Division of Recreation and Parks and Ordway Swisher Biological Station for research permits. Useful comments were provided by J. Ash, D. A. Jones, T. A. Martin, S. Mulkey, J. Pausas, and F. E. Putz. This paper also benefited from valuable suggestions made by members of the PEERS group in the Department of Botany at the University of Florida, and the Ecology Group at Wageningen University, the Netherlands. Finally, the insights of two anonymous reviewers improved the manuscript.