Root biomechanics in Rhizophora mangle: anatomy, morphology and ecology of mangrove’s flying buttresses

Abstract

Background and Aims Rhizophora species of mangroves have a conspicuous system of stilt-like roots (rhizophores) that grow from the main stem and resemble flying buttresses. As such, the development of rhizophores can be predicted to be important for the effective transmission of dynamic loads from the top of the tree to the ground, especially where the substrate is unstable, as is often the case in the habitats where Rhizophora species typically grow. This study tests the hypothesis that rhizophore architecture in R. mangle co-varies with their proximity to the main stem, and with stem size and crown position.

Methods The allometry and wood mechanical properties of R. mangle (red mangrove) trees growing in a mangrove basin forest within a coastal lagoon in Mexico were compared with those of coexisting, non-buttressed mangrove trees of Avicennia germinans. The anatomy of rhizophores was related to mechanical stress due to crown orientation (static load) and to prevailing winds (dynamic load) at the study site.

Key Results Rhizophores buttressed between 10 and 33 % of tree height. There were significant and direct scaling relationships between the number, height and length of rhizophores vs. basal area, tree height and crown area. Wood mechanical resistance was significantly higher in the buttressed R. mangle (modulus of elasticity, MOE = 18·1 ± 2 GPa) than in A. germinans (MOE = 12·1 ± 0·5 GPa). Slenderness ratios (total height/stem diameter) were higher in R. mangle, but there were no interspecies differences in critical buckling height. When in proximity to the main stem, rhizophores had a lower length/height ratio, higher eccentricity and higher xylem/bark and pith proportions. However, there were no directional trends with regard to prevailing winds or tree leaning.

Conclusions In comparison with A. germinans, a tree species with wide girth and flare at the base, R. mangle supports a thinner stem of higher mechanical resistance that is stabilized by rhizophores resembling flying buttresses. This provides a unique strategy to increase tree slenderness and height in the typically unstable substrate on which the trees grow, at a site that is subject to frequent storms.

INTRODUCTION

Buttresses of tropical trees are regarded as support structures in shallow substrates (Henwood, 1973; Herwitz, 1998; Tang et al., 2011) that counteract static and dynamic loads (Smith, 1972; Black and Harper, 1979; Richter, 1984; Warren et al., 1988). Buttresses are normally developed from stress-induced cambial activity that stimulates secondary growth to generate structures that transmit the dynamic loadings from the crown to the roots and, finally, to the ground (Ennos, 1993; Clair et al., 2003). The form and distribution of buttresses in tropical rainforest trees are critical in withstanding loads, and this may be the case with the complex stilt root morphology in 18 reported tropical species and six families (Fisher, 1982). For example, in the palms Iriartea deltoidea, Euterpe precatoria and Socratea exorriza, the size of the cone-shaped system of stilt roots is proportional to the biomass and height (Avalos et al., 2005; Avalos and Fernández Otárola, 2010).

In the Rhizophora spp., the aerial root system is composed of stilts that grow from the main stem, resembling flying buttresses (Gill and Tomlinson, 1969; Fisher, 1982). Such structures are derived from stems; thus, the term ‘rhizophore’ is more appropriate than stilt, prop or aerial root (De Menezes, 2006), and we will use it hereafter. Rhizophores develop from the cambium of the main stem or the main branches, frequently reaching the ground and dividing in a sympodial fashion (De Menezes, 2006). Concerning their morphology, they may stand as arches perpendicular to the main stem and as columns parallel to the main stem, with arches having more secondary thickening and less cortex than columns (Gill and Tomlinson, 1969, 1971, 1977). Above-ground rhizophores in Rhizophora apiculata represent 10–20 % of the total biomass, while below-ground roots represent 5 % (Ong et al., 2004). The relatively low proportion of biomass invested in below-ground roots and the fact that the anoxic environment due to frequent flooding restricts root development to a maximum depth of 50 cm (Gill and Tomlinson, 1977; Ong et al., 2004) suggest that the main function of the above-ground rhizophore system is structural. In addition, lateral branches of Rhizophora mangle respond to tensile stress by producing flexion wood to improve the mechanical stability of the crown (Fisher, 1982). Just as the Gothic cathedrals depend on flying buttresses to improve their slenderness, the development of rhizophores should be important for the effective transmission of dynamic loads from the top to the ground, more so in a typically unstable substrate.

Once they reach the ground, rhizophores may stabilize the main stem, transmitting the compression–tension forces from the crown to the ground. If this is the case, we hypothesize that as the tree grows in height, the morphology of rhizophores will vary anatomically and allometrically according to their proximity to the main stem: those attached to the main stem (first-order rhizophores) will attain greater height than length compared with the smaller (second- and subsequent order rhizophores) that grow away from the main stem. Finally, the number and length of rhizophores will increase with progressing static (e.g. asymmetric crowns) or dynamic (e.g. prevailing winds) loads, both related to the production of buttresses in trees in lowland tropical forests (Richter, 1984; Warren et al., 1988). This study provides insights into the morphological, anatomical and ecological causes of the rhizophore system of red mangroves. Understanding the mechanical function of the rhizophore system is critical, since red mangroves may play a critical role in coastal protection against natural hazards, such as tropical storms and tsunamis (Mazda et al., 1997; Danielsen et al., 2005; Dahdouh-Guebas et al., 2005; Laso Bayas et al., 2011).

MATERIALS AND METHODS

Study site and plant selection

This study was carried out in a Rhizophora mangle (L.)-dominated forest within an interdistributary basin with water salinities from 11 ppt (wet season) to 28 ppt (dry season) in La Mancha Lagoon, Veracruz, Mexico (19°33′27″N, 96°23′04″W; Méndez-Alonzo et al., 2012). Height (H) and diameter above the uppermost rhizophore (D) were measured in 181 individuals within a 1200 m² plot. For comparison, H–D at breast height data from 111 trees of the black mangrove, Avicennia germinans, growing in a plot adjacent the study site in the same interdistributary basin, were included (Méndez-Alonzo et al., 2008).

Rhizophore geometry

Rhizophores were classified in orders according to the number of arches away from the main stem. In one of the largest trees within the study plot (24·4 cm stem diameter, 12·7 m high, 150 m² crown cover, 11·4 m² rhizophore projected area), all rhizophores were ordered according to their position with respect to the main stem and measured in length and height. Also, the following variables were measured in 32 undamaged trees having at least one rhizophore with more than two arches (Fig. 1): minor and major diameter of the main stem (A and B, Fig. 1); height and length of the first arch of the rhizophore (X and Y, Fig. 1); and total length of the rhizophore (i.e. sum of all arches). In large trees, the main stem did not touch the ground, so a plumb line below the main stem was used as a vertical reference for the central part of the tree (W, Fig. 1).

Fig. 1.

Morphological measurements of the Rhizophora mangle radicular system: A, maximum stem diameter; B, minimum stem diameter; Y, first-order rhizophore height; X, first-order rhizophore length. W is a plumb line that indicates the centre of mass of the tree, since usually there is no basal main stem.

Crown and rhizophore projected area

For each tree, the crown area was approximated to an ellipse, and the length and direction of all its rhizophores were projected in a polar plot using Sigma Plot 10 (Systat Software Inc., San Jose, CA, USA). The 32 polar plots were exported as JPG images and the rhizophore-projected areas were calculated using Arc View 3.2 (ESRI 1999, Redlands, CA, USA).

Crown leaning and rhizophore orientation

Vector analyses were used to test if the mean orientation of rhizophores was related to crown leaning (Richter, 1984; Warren et al., 1988). Two sets of co-ordinate pairs were developed for each tree: (1) crown orientation, a vector derived from the orientation of leaning (y, N = 0°), and the hypotenuse of leaning (x), calculated from the leaning angle (θ) and distance (m) from the centre of the main stem to the centre of the crown; and (2) rhizophore orientation, a vector derived from the angle of emergence from the main stem (θ, N = 0°) and the horizontal distance between the main stem and the insertion into the ground (m):

$$x=\text{sin}\theta ·m;y=\text{cos}\theta ·m;$$

For each tree, the mean rhizophore orientation was calculated as:

$$\begin{array}{c}\overline{\theta }=\text{arctan}\left(\overline{x}/\overline{y}\right), \text{\hspace{0.17em}if} \overline{y}>0, \text{or}\\ \overline{\theta }=180+\arctan \left(\overline{x}/\overline{y}\right), \text{if} \overline{y}<0.\end{array}$$

and the mean rhizophore magnitude (M) was expressed as:

$$M={\left({\overline{x}}^2+{\overline{y}}^2\right)}^{\mathrm{0·5}}$$

Angular dispersion for all rhizophores was calculated as r = [Σ(sin θi/N)²+ Σ(cos θi/N)²]^0·5, where r = 0 indicates a uniform distribution of angles and r = 1 indicates perfect concentration of directions (Zar, 2010).

Wood biomechanics and functional anatomical traits

To perform the mechanical tests, eight stems of R. mangle and four of A. germinans were collected to obtain 12 samples measuring 25 × 25 × 410 mm for flexion analyses and 36 samples measuring 25 × 25 × 100 mm for compression analyses. The samples were tested for bending and compression in a Universal Testing Machine (model 5543 Instron Inc., Norwood, MA, USA) to obtain the modulus of elasticity (MOE), the modulus of rupture (MOR) and the compressive strength (Fc_max) (Gere and Goodno, 2009). The relative density (ρ) was obtained from 1 cm³ segments from these samples using the dry weight/fresh volume immersion method (Pérez-Harguindeguy et al., 2013).

The changes in cross-sectional geometry in relation to length were studied in three 1 m long, first-order rhizophores (i.e. those attached to the main stem) by cutting them in 10 cm long segments. Also, 10 cm long segments were cut from the basal insertion of three sets of first- to sixth-order rhizophores. All segments were scanned in a flatbed scanner (Epson V700 perfection, Epson Corp. Nagano, Japan) and their eccentricity (ε, the deviation from a perfect circle) was measured as ε = [1 − (a²/b²)]^0·5, where a and b are the major and minor radii (Rigg and Harrar, 1931).

The bark, xylem and pith area were measured in all segments, and ε was calculated for each tissue using Image J (http://imagej.nih.gov/ij/).

Statistical analyses

Scaling relationships were employed to determine the associations between log-transformed H and D for A. germinans and R. mangle: (1) ordinary least squares (OLS) and (2) reduced major axis regressions (Niklas, 1994, 2006), and (3) asymptotic non-linear regression (here, H and D were not log transformed):

$$H=H_{\text{max}}\left(\text{1}-{\text{e}}^{aD}\right)$$

where H_max indicates the upper limit of tree height, D the diameter, and a is a regression constant (Thomas, 1996). Data were fitted with Sigma Plot ver. 10.0 (Systat Software Inc., IL, USA) and the slope and intercept for OLS on log-transformed data was compared between species by means of analysis of covariance (ANCOVA) using JMP 8.0 (SAS Institute Inc., Cary, NC, USA). OLS were also used to test for correlations between the proportion of tissues and eccentricity with the length of the first-order rhizophore and between rhizophore height and height/length ratios across the base of three sets of rhizophores, including at least five sequential arches.

The slenderness ratios (S_r) were calculated as H/D; the critical buckling height, H_crit, was calculated by the Greenhill equation (Niklas, 1994):

$$H_{\text{crit}}=\mathrm{0·}\text{792}{\left(\text{MOE}/\rho \right)}^{\text{1}/\text{3}}{D}^{\text{2}/\text{3}}$$

and the tree stability safety factors (SSF) were calculated as H_crit/H. Mann–Whitney Rank sum tests were used to compare SSF and S_r between species.

The Rayleigh test was used to detect significant departures from random (N = 0°) of the mean direction of leaning crowns. A Hotelling’s T² test (Richter, 1984; Warren et al., 1988) was used to test for significant differences in the mean rhizophore vector orientation, and a canonical correlation between the sine and cosine of the two sets of angular data was performed to determine if the rhizophore vectors and crown leaning orientation were correlated (Richter, 1984).

RESULTS

Whole-plant allometry

The height–diameter (H–D) slopes were steeper in Rhizophora mangle than in Avicennia germinans in the two linear allometric models (Table 1). Furthermore, the ANCOVA on the log-transformed H–D values indicated that the slopes of the two regression lines were significantly different (F_1,289 = 72; P < 0·001), confirming the steeper slope for R. mangle compared with A. germinans. Regarding the non-linear regression, the predicted asymptotic height was lower in R. mangle at 16·8 m, with a maximum diameter of around 30 cm compared with A. germinans (18·8 m with a maximum diameter of 83 cm; Table 1, Fig. 2).

Table 1.

Parameters and determination coefficients for height–diameter ordinary least squares (OLS), reduced major axis (RMA) and the non-linear asymptotic (NLA) regression for Rhizophora mangle and Avicennia germinans in an interdistributary basin mangrove forest at La Mancha Lagoon, Veracruz, Mexico

Species	n	OLS (95 % CI)			RMA (95 % CI)			NLA (s.e.)
		a	b	R2	α	β	R2	a	Hmax	R2
R. mangle	181	1·18 (0·06)	−0·15 (0·03)	0·86	1·27 (0·17)	−0 ·20 (0·12)	0·86	0·09 (0·009)	16·43 (1·06)	0·85
A. germinans	111	0·74 (0·06)	0·09 (0·06)	0·82	0·81 (0·23)	0 ·007 (0·28)	0·82	0·05 (0·004)	18·87 (0·87)	0·83

In OLS, b is the origin and a is the slope; in RMA, β is the allometric constant and α is the scaling exponent, and in NLA, a is the origin and H_max is the asymptotic height.

Confidence intervals (±95 % CI) or standard errors (±s.e.) are in parentheses.

Fig. 2.

Tree height–diameter allometry in 181 Rhizophora mangle and 111 Avicennia germinans trees (as indicated in the key) growing at La Mancha Lagoon, Veracruz, Mexico. In R. mangle, height is larger at a given diameter above the upper rhizophore than at an equivalent diameter at breast height in A. germinans.

Rhizophores buttressed from between 10 and 33 % of total main stem height (H) of R. mangle (Table 2). Furthermore, H was significantly correlated with maximum rhizophore height (Fig. 3A), maximum rhizophore length (Fig. 3B) and number of rhizophores per tree (Fig. 3C). Finally, there was a significant and positive correlation between the crown and the projected area of the rhizophores (Fig. 4).

Table 2.

Descriptive statistics for 32 Rhizophora mangle trees and their rhizophores growing in an interdistributary basin mangrove forest at La Mancha Lagoon, Veracruz, Mexico

	Minimum	Maximum	Mean	CV
Trees (n = 32)
No. of rhizophores	1	14	5 ·03	0 ·74
Tree height (m)	3	17	7 ·5	0 ·45
Main stem diameter (cm)	1 ·9	24 ·4	11 ·3	0 ·46
% of tree buttressed with rhizophores	10	33	19 ·7	0 ·36
Rhizophores (n = 169)
Length (m)	0 ·08	5 ·60	1 ·13	1 ·15
Height (m)	0 ·30	5 ·30	1 ·27	0 ·55

CV, coefficient of variation.

Fig. 3.

Allometric relationships between tree height and rhizophore morphology for 32 Rhizophora mangle mangrove trees growing at La Mancha Lagoon, Veracruz, Mexico. There were significant correlations between tree height and rhizophores per tree (A), maximum rhizophore length (B) and maximum rhizophore height (C).

Fig. 4.

Relationship between the basal and crown area of 32 Rhizophora mangle trees growing at La Mancha Lagoon, Veracruz, Mexico.

Wood properties

The wood of R. mangle was stronger and stiffer than that of A. germinans, as indicated by its higher ρ, MOE, MOR and Fc_max, parallel or perpendicular to the grain for the compressive or flexural test, respectively (Table 3). The S_r and the SSF were significantly higher in R. mangle than in A. germinans (S_r, 85·2 ± 2·8 vs. 64·7 ± 2·0; U = 14211, P < 0·001; SSF = 7·41 ± 0·38 vs. 6·32 ± 0·22, U = 11695, P = 0·02, respectively).

Table 3.

Mean (± s.e.) of wood properties of Avicennia germinans and Rhizophora mangle trees in an interdistributary basin forest at La Mancha Lagoon, Veracruz, Mexico

Species	MOE-C (GPa)	Fcmax (MPa)	MOE-F (GPa)	MOR-F (MPa)	RWC (%)	ρ (g cm−3)
A. germinans	16·2 (0·75)	33·4 (3·46)	12·1 (0·54)	72 ·2 (3·08)	60·0 (1·16)	0·705 (0·011)
R. mangle	23·2 (0·64)	46·3 (1·09)	18·1 (1·98)	118 ·6 (2·12)	56·0 (1·17)	0·798 (0·004)

Tests using flexural (F) and compressive (C) strength were performed in green samples.

MOE, modulus of elasticity; MOR, modulus of rupture; Fc_max, compressive strength parallel to grain; RWC, relative water content; ρ, wood density.

Stem and rhizophore functional anatomy

There was a monotonic decrease in rhizophore height (R = 0·98, P = 0·004; Fig. 5A) and an increase in the length/height proportion in the rhizophores as a function of the rhizophore order (R = 0·86, P = 0·06; Fig. 5B). Within first-order rhizophores, the xylem proportion in the cross-section was larger when closer to the main stem, and decreased progressively as the rhizophore approached the ground, while increasing the proportion of bark and pith (Fig. 6A). This pattern was also observed in higher order rhizophores (Fig. 6B). Bark eccentricity was significantly correlated with rhizophore length, but xylem and pith eccentricity were not (Fig. 7). Across rhizophore orders, bark and xylem eccentricity were nearly elliptical at the main stem junction and almost circular close to the ground; however, the pith was always circular in transversal shape (Fig. 7).

Fig. 5.

Change in height and in length/height proportion in five sequential orders of 40 rhizophores from a 13 m height Rhizophora mangle tree growing at La Mancha Lagoon, Veracruz, Mexico.

Fig. 6.

Relative proportions of bark (including aerenchyma), xylem and pith along the length of individual first-order rhizophores (A), and at the base of rhizophores of sequential orders (B) of three Rhizophora mangle mangrove trees growing at La Mancha Lagoon, Veracruz, Mexico. Individual rhizophores were approx. 1 m long and were cut every 10 cm.

Fig. 7.

Changes in the eccentricity of bark, xylem and pith along the length of three first-order rhizophores (left), and at the base of rhizophores of sequential order of three Rhizophora mangle mangrove trees (right) growing at La Mancha Lagoon, Veracruz, Mexico. The eccentricity of a perfect circle is zero and that of an extreme ellipse is one.

Buttress and crown orientation

Rhizophore mean angle was 113° (i.e. south-east), angular dispersion (r) was 0·04 and mean length was 78 cm. Mean crown leaning was 80 ± 1·4° (straight trees = 90°), with a mean distance from the trunk to the centre of the crown of 1·2 ± 0·2 m. Both rhizophores and crowns were randomly disposed around the azimuth (Hotelling test for rhizophores, F_2,30 = 0·23, P = 0·21; Rayleigh test for crowns, z₃₀ = 1·36, P = 0·91), and there was no association between mean rhizophore orientation and crown leaning per tree (R = 0·22, P = 0·3). As no directional trend was found in rhizophores or crowns, no association tests with prevailing winds or direction of wave or storm surges were performed.

DISCUSSION

The results provided in this study support the hypothesis that the production and morphology of rhizophores in Rhizophora mangle vary anatomically and allometrically according to tree size. The combination of this unique geometry with a wood of high mechanical resistance allows R. mangle to reach the canopy with a much more slender main stem than other mangrove species in the New World, such as its immediate competitor, Avicennia germinans. The supportive function is enhanced in the first-order rhizophores, with lower length/height proportion, higher cross-sectional eccentricity and higher proportion of xylem compared with bark and pith. In conifers, flexure wood is produced as a thigmomorphogenetic response to wind, increasing the eccentricity of stems in the direction of flexure (Telewski and Joffe, 1986; Telewski, 1989). We found that first-order rhizophores were more eccentric when in proximity to the main stem and that eccentricity was positively associated with rhizophore size. Further experiments are required to test if the production of rhizophores is an adaptive response to tree growth or if rhizophores actively acclimate to chronic stress by developing more or stronger rhizophores to respond to new crown positions (e.g. after tilts caused by storms or surges, especially in unstable substrates). Since first-order rhizophores are taller than longer, the dynamic loadings are transmitted to the ground in an analogous manner to the flying buttresses of European Gothic cathedrals (Mark, 1993). Alternatively, second- and further order rhizophores (we measured up to six orders) may contribute to the stabilization of the first-order rhizophores by increasing the basal area of the tree, but their mechanical importance may be reduced in comparison with other tasks. Having higher proportions of bark and pith, they may be increasingly important in water conduction, nutrient uptake, storage and gas exchange via lenticels (Tomlinson, 1986), which would explain the high amount of aerenchyma in the cortex of higher order rhizophores.

During the ontogeny of the trees, rhizophores are continually produced to act as guy lines, allowing the upper branches to search actively for a place in the canopy. We found neither directional trends in the rhizophore or crown mean orientation, nor an association between tree lean and rhizophore production. The efficiency of this structure is remarkable, as in our study site there is a yearly recurrence of north trade winds that can reach up to 100 km h⁻¹, but R. mangle trees were practically straight from base to crown (mean tree lean = 80°). We did not detect any influence of waves or storm surges, an important mechanical stressor in this environment (Mazda et al., 1997, 2006), on the production of rhizophores. Our study plot was set on the inland area of a coastal lagoon that is probably protected from tropical storms and surges by its geographic location. This would change in exposed mangrove forests, where the drag induced by waves or storms is much higher and where trees would produce more rhizophores at equivalent trunk size than trees in more protected areas. Additionally, R. mangle trees on the fringes of lagoons, where the substrate is more unstable, seem to be shorter, and may also produce more rhizophores, but we are not aware of any study in this respect.

The network of rhizophores might aid in withstanding the tension and compression forces produced as the crown moves during storms, resisting uprooting. This structural solution can be compared with the function of buttresses in tropical rainforest trees. Buttresses increase the area covered by roots, optimizing anchorage and allowing a few distant and deep roots to transmit dynamic loadings to the ground by directing compression forces to the leeward roots and tensional forces to the windward roots (Mattheck, 1991; Ennos, 1993; Dupuy et al., 2005), thus reducing the risk of structural failure and increasing the resistance to lateral strains (Young and Perkocha, 1994; Crook et al., 1997; Clair et al., 2003). The ‘flying buttresses’ of R. mangle, instead of transmitting forces to a few critical roots, form a profuse net of tensional supports that attaches the main stem to the ground as the tree grows in height. Rhizophores may act as a structural solution that substantially increases the contact area of the trees with the ground while allowing atmospheric air interchange through lenticels and aerenchyma, which is critical given the shallow depth of below-ground roots in a fundamentally anaerobic substrate.

A remarkable environmental service provided by the rhizophore system of R. mangle is to act as a natural barrier against natural catastrophes such as hurricanes and tsunamis, which has been proposed in before–after evaluations (Das and Vicent, 2009; Temmerman et al., 2013) and in modelling studies (Mazda et al., 1997, 2006; Ohira et al., 2013), although its relative contribution depends on the intensity of the storm (Cochard, 2013). Further experiments should explore the potential of individual species of mangroves to mitigate natural hazards (Dahdouh-Guebas et al., 2005). Given the suite of biomechanical adaptations in the rhizophore system, it can be expected that the effectiveness of Rhizophora as a protective barrier against storm surges is higher than that of other mangrove species.