Trees form mechanics and hydraulics Tree stature

Individual trees of a large range of size are to be found in the tropical rain forest (see the profile diagram in Fig. 2.1). Each tree species also has a characteristic size at maturity and species are often referred to various stature classes, such as understorey trees, canopy trees and emergents, but, as Figure 2.2 shows, there is no discrete clustering of species in size classes. Maximum diameter for species on 50 ha in Pasoh Forest, Peninsular Malaysia (Fig. 2.2), was approximately a truncated log-normal distribution with a modal class in the 10-20 cm maximum dbh range. A breakdown of species into height classes (Table 2.1) shows about half of the tree species to have maximal heights of 20 m or less. The tallest trees at Pasoh probably reach to about 60 m in height, showing that the community of understorey specialists in the forest is as rich in tree species as that of the canopy, and is definitely richer on a unit depth basis. A similar pattern is seen in the forest at La Selva, Costa Rica (Hartshorn 1980), and on Barro Colorado Island, Panama (Hubbell & Foster 1992).

The vertical distribution of tree species diversity in the forest may reflect the relative illumination at different heights above the forest floor. At Pasoh, relative illumination increases logarithmically with height (Fig. 2.3). The understorey species in the bottom 20 m of the forest rarely receive more than 5% of the radiation arriving at the top of the canopy. However, the rate of change of relative illumination with height is greatest near the forest floor. In other words, for a given height increase the relative, rather than absolute, increase in illumination is greater nearest the forest floor. It may be that this steep gradient facilitates partitioning of the light resource among species, and hence more species per unit depth as specialists than higher in the forest.

There has been a growing tendency in the tropical forest literature to use the terms shrub and treelet to categorise species in terms of maximum height, with 'shrub' taken to represent the smallest trees and 'treelet' the next size class up, with an arbitrarily set limit between the two. More traditionally, shrubs are distinguished as small woody plants either with multiple stems, or with branching very close to the ground. It has been claimed that 'true' shrubs are rare in tropical rain forests. Givnish (1984) pointed out that only 6 out of 95 non-climbing woody species of less than 10 m in height at maturity native to Ghana were true shrubs whereas temperate forests frequently have many shrubby species in the understorey. The multi-stemmed crown construction is probably a more efficient way of supporting a low, broad canopy than a single stem with wide-reaching branches. As yet there is no strongly supported hypothesis as to why shrubs are infrequent members of the rain-forest community.

The treelet is a life form perhaps commoner in the tropical rain forest than any other vegetation type: a single-stemmed, unbranched or sparsely branched tree of diminutive stature at maturity. D'Arcy (1973) proposed two main evolutionary origins of the treelets. They arose either through precocious flowering in woody groups, or through the development of woody stems in otherwise herbaceous evolutionary lines. The latter is well exhibited in Hawaii, where species of Cyanea and Plantago occur among the forest treelets.

Peat Swamp Forest Whitmore 1984

Figure 2.1 Profile diagram of the forest at Ulu Dapoi, Tinjar, Borneo. A plot 60 m x 8 m (200 x 25 feet) is shown. Courtesy of Professor P.S. Ashton.

Figure 2.1 Profile diagram of the forest at Ulu Dapoi, Tinjar, Borneo. A plot 60 m x 8 m (200 x 25 feet) is shown. Courtesy of Professor P.S. Ashton.

Table 2.1. Species diversity and tree frequency by stature class for Pasoh Forest Reserve, Peninsular Malaysia (stems > 1 cm dbh for 50 ha of forest)

Stature

Definition

No. of species

Proportion of all species (%)

Trees per ha

Proportion of all stems (%)

'shrub'

to 2 m tall

54

6.6

303

4.6

'treelet'

2-10 m tall

111

13.5

1513

22.9

understorey

10-20 m tall

283

34.5

2114

32.1

canopy

20-30 m tall

317

38.7

1901

28.9

emergent

> 30 m tall

55

6.7

753

11.4

Data from Kochummen et al. (1990). 60-1-

Data from Kochummen et al. (1990). 60-1-

Observed maximum dbh (cm)

Figure 2.2 Distribution of the maximum trunk diameter for tree species with more than 300 individuals in a 50 ha plot at Pasoh, Malaysia. After Kohyama (1996).

I will use the terms shrub and treelet in their more exact morphological meaning in the text, and refer to pygmy trees when discussing the smallest of the arborescent taxa irrespective of architecture.

At Barro Colorado Island, the commonest species in the 50 ha plot was the treelet Hybanthusprunifolius, with nearly 40 000 individuals out of a total of 238 000 enumerated stems. One might argue that it is easier for pygmy trees to be common because they are smaller and hence more individuals can be

packed into an area. This ignores the fact that there are small juveniles of large trees, which makes it theoretically possible for the largest tree species to have the greatest population densities. However, small-statured species must have the greatest scope for high densities of reproductively mature individuals. Burger (1980) noted that pygmy trees of the rain forest in Costa Rica tend to be from large genera such as Piper and Psychotria, with notably more species per genus in the flora than for larger trees. On a pantropical scale, these two mammoth shrubby genera are joined by others such as Ardisia and Rinorea, although a more detailed analysis is required to confirm that rainforest pygmy trees generally have a greater species to genus ratio than larger life forms.

Wood

Wood is derived by secondary thickening of root and stem axes. Secondary thickening involves the periclinal division of cells. This occurs at a peripheral cambium in typical trees, but at scattered sites through the axis in many dendroid monocots. Functionally, wood contains three main elements: xylem conduits (tracheids and/or vessels) that conduct the xylem sap, xylem fibres that provide most of the mechanical support (in gymnosperm wood the conduction and support roles are less distinctly demarcated as tracheids are

Mean relative illuminance (%)

Figure 2.3 Vertical profile of mean relative illuminance (expressed as a percentage of light received above the canopy) at a rain-forest site at Pasoh, Malaysia. After Richards (1996).

Mean relative illuminance (%)

Figure 2.3 Vertical profile of mean relative illuminance (expressed as a percentage of light received above the canopy) at a rain-forest site at Pasoh, Malaysia. After Richards (1996).

solely responsible for both) and parenchyma cells that are probably important in the transverse conduction of water and storage, and possibly assist conduits to recover from cavitation. The first two cell types are very markedly elongated in the direction of the long axis of the tissue, the grain of the wood, with the conduits providing a pipework for the movement of water along the stem or root.

Wood is an excellent structural material (Jeronimidis 1980) because it combines strength and stiffness with a high resistance to the propagation of cracks in a relatively light composite that shows relatively little temperature dependence in its mechanical properties. It is, however, a highly anisotropic material. Its compressive strength is generally half the tensile strength, the strength across the grain is 10 to 100 times less than along it, and work of fracture (toughness) is only high across the grain.

The long columns of cells, particularly the fibres and tracheids, provide the strength of the timber along the grain, and the resistance of wood to crack propagation across it. The rays hold the axial fibres together, resisting tor-sional and shear stresses, and determine the transverse strength of the tree. Wood cells have cellulose microfibrils running helically along the long axis of the cell in the S2 wall (Fig. 2.4). The orientation of the microfibrils at about 25° to the perpendicular from the long axis of the cell optimises their contribution to both tensile and compressive strength of the timber. Lignin plays an important role in cross-linking and cementing the cellulose components of the cell walls thus improving the stiffness and strength of the timber. The S2 microfibrils are also important contributors to the resistance to cracking of timber across the grain. The avoidance of crack propagation is important because structural failure is usually brought about by crack formation at localised sites of high stress intensity rather than by forces exceeding the strength of the material. Lucas et al. (1997) proposed that the stress concentration at the tip of a crack propagating across the wood grain tends to cause the inward collapse of the S2-microfibril layer of the fibre cells. The microfibrils pull the walls into the lumen along almost the complete length of the cell. This collapse of the fibre layer represents considerable use of energy, and plastic deformation may be responsible for up to 90% of the work done during fracture of the timber.

The mechanical properties of wood are strongly positively correlated with its density. The actual density of the solid component of wood varies little among tree species, being around 1.5 g cm~3 for air-dry material (Williamson 1984; Detienne & Chanson 1996). It is the presence of spaces within and between cells that causes variation in wood density. Low-density timbers, such as balsa (Ochroma pyramidale), have thin-walled wood cells with large lumina and abundant spaces between the cells. The densest heartwood timbers approach the space-free density maximum. Strength properties are directly related to the amount of solid material present, but the cellular nature of wood does contribute to crack resistance (Lucas et al. 1997). Very dense timbers have little, if any, lumen to the fibres. Therefore the amount of plastic deformation possible during cracking is reduced and the work of fracture declines. Thus very dense timbers, such as lignum vitae (Guaiacum spp.), are very hard and strong, but apt to be brittle. Very low-density timbers have lower toughness than expected because their thin cell walls fracture before they can collapse.

Wood Cellular Wall Structure

Figure 2.4 Schematic representation of the ultrastructure of the wall components of wood fibres. ML, middle lamella, P, primary wall, S, secondary wall, which is present as three layers, S^ S2 and S3. The S2 wall contains helically wound cellulose microfibrils. W represents the warty inner luminal surface. In some species the surface is smooth.

Figure 2.4 Schematic representation of the ultrastructure of the wall components of wood fibres. ML, middle lamella, P, primary wall, S, secondary wall, which is present as three layers, S^ S2 and S3. The S2 wall contains helically wound cellulose microfibrils. W represents the warty inner luminal surface. In some species the surface is smooth.

Tropical timber

The distinctive feature of the wood of tropical trees is the presence of wide (large internal diameter) vessel elements, generally with simple perforation plates (Carlquist 1988). The major advantage of wide vessels is increased rates of water conduction because fluid flows through pipes at rates proportional to the fourth power of their diameter. However, not all the vessels in tropical wood are wide. There are usually many small-diameter vessels also. These probably provide safety through redundancy to the conductive system of the tree, and may be important in the radial conduction of water (Tyree et al. 1994). Very wide vessels, of more than 100 |im diameter, are rare in trees, being common only in lianas that require high sap-flow rates in stems of limited cross-sectional area. The possession of more wide vessels would allow tropical trees even greater rates of sap flow through trunks and branches but there generally comes a point when delivery rates of water to the leaves would not be improved by increased conductivity of the stem and main branches because of the higher resistance to flow in the outer branches and twigs. Wide vessels have often been considered as more susceptible to cavitation than narrow ones. Tyree & Ewers (1996) found a weak negative correlation between applied pressure required to cause a 50% reduction in hydraulic conductivity and mean vessel diameter for a large selection of tree species. However, the mechanism of greater susceptibility to cavitation in wider xylem conduits is only clear in the case of freezing-induced embolism. Large air bubbles take longer to dissolve than small ones, increasing the likelihood of serious cavitation after the freezing of xylem sap (Tyree et al. 1994). Freezing temperatures are not experienced in the lowland tropics, but they are in the temperate zone, where trees tend to have vessels of smaller diameter (van der Graaff & Baas 1974). This is probably why temperate timbers of the same density as tropical ones consistently have 20% lower cross-grain work of fracture (Lucas et al. 1997). The greater vessel number in the temperate woods reduces their mechanical performance. It is questionable, however, whether there has been selection for a certain vessel size in tropical trees to improve the mechanical properties of the wood. Roots tend to have wider vessels than stems (Ewers et al. 1997), probably because roots are less liable to cavitation and are better able to recover from it.

Tropical forests are remarkable for the wide range of timber densities present, frequently including species of both very low (less than 0.3 gcm~3) and very high (more than 0.7gcm~3) density (Williamson 1984). Wood densities are often reported as weight per volume at 12% moisture (Reyes et al. 1992; Detienne & Chanson 1996), though for ecological purposes the true density, oven-dry mass per green volume, is more useful. Reyes et al. (1992) provide a regression equation to make the conversion between the two types of density estimate. Multi-species samples from tropical forests show normal distributions of wood densities (Fig. 2.5) with means in the range 0.560.62 g cm(Reyes et al. 1992; Detienne & Chanson 1996). These studies provide support to the findings of Whitmore & Silva (1990) that South American, particularly Amazonian, timbers tend to be denser, or at least show a greater spread of densities with more very heavy woods present, than those of other tropical regions.

Wood density has some affinities with taxonomy. The Bombacaceae are notable for very light timber, particularly in genera such as Cavanillesia, Ceiba and Ochroma. Other low-density woods are found in genera such as Cecropia, Musanga, Ficus and Sterculia. Very heavy woods are often from the caesalpinoid subfamily of the legumes; Swartzia is a good example. The ebonies (Diospyros, Ebenaceae) are also renowned for their very heavy timber.

There is some evidence that the range of wood density in tropical forests narrows with elevation and widens with increased severity of drought (Williamson 1984). Tree species found in more exposed sites in the montane forests of Costa Rica mostly had higher wood densities than the species characteristic of more sheltered places (Lawton 1984). Presumably species with low-density, weak wood would be at a strong disadvantage under the greater wind loading of mountain ridges. The wood anatomy of species of the drought-deciduous forest at Chamela was compared with those from a wet lowland forest at Los Tuxtlas, also in Mexico (Barajas-Morales 1985). The evergreen-forest trees had softer, lighter-coloured and less dense wood with fewer crystalline and resinous inclusions than the deciduous-forest species. Average vessel diameter was lower in the trees from Chamela, as is typical for more seasonally dry forests (Tyree et al. 1994). However, in comparing 30 species that grew in both rain-forest and savannah sites in Ivory Coast, den Outer & van Veenendaal (1976) found a majority of species to have wider vessels in the drier savannah area. Overall the difference in average vessel diameters between the sites was not statistically significant.

Not all trees form heartwood: for instance palms, and a number of fast-growing soft-wooded dicot species e.g. Dyera costulata, do not. These sapwood trees can actively defend their wood through the agency of the living cells present throughout the trunk and are rarely, if ever, found hollow because any pest or disease that becomes established is likely to destroy the whole tree. The dead heartwood that constitutes the core of the trunk of most trees can only be protected from decay by the passive action of its constituents and by being covered by the overlying living sapwood.

Density may also be a measure of the degree to which woody organs are protected from attack by pests and pathogens. The mechanical properties of

Figure 2.5 Frequency distribution of tropical forest species by wood-density class for three tropical regions. After Reyes et al. (1992).

wood at a smaller scale may help protect it from decay. Animals will find very dense, hard wood difficult to comminute prior to ingestion, and fungal hyphae will not be able to penetrate and break down the woody tissue. However, heartrot does occur and hollow trees are not uncommon in tropical forests.

Wood density has often been used by foresters and ecologists as a measure of a species' maximal growth rate and of its relative shade tolerance. Fast-growing, shade-intolerant species have low wood densities, whereas the wood of slow-growing, shade-tolerant species is heavy. There is surprisingly little reliable quantitative evidence to substantiate this assertion. Wood density showed a significant, though weak, negative correlation with average diameter growth rate across 122 species from French Guiana (Favrichon 1994). In a rain forest in Borneo, trees with a stem specific gravity (including bark) of 0.2-0.49 exhibited significantly faster relative stem diameter growth than those with stem specific gravity of more than 0.5 (Suzuki 1999). Seedling persistence in deep shade exhibited a significant positive correlation with adult wood density among 18 wind-dispersed trees from Barro Colorado Island, Panama (Augspurger 1984). Wood density was the best (though still weak) predictor of light requirement index for regeneration among tree species from Queensland, Australia (Osunkoya 1996), but the index was a ranking of the impressions of ecologists, and not a quantifiable measure of shade-tolerance hypothesised a priori to be relevant.

Most of the cells in wood — fibres, tracheids and vessel elements—are dead, but the ray parenchyma and other sapwood tissues are living and hence respiring. The rates of respiration of woody tissues of two tree species at La Selva, Costa Rica, have been measured (Ryan et al. 1994). The fast-growing Simarouba amara (1.24 |jmolm~2 s"i) was found to respire at considerably faster rates than the slow-growing Minquartia guianensis (0.83 |imol m~2 s"i). When allowance was made for wood volume growth rates it was found that the two species had comparable maintenance respiration rates. These rates were twice those of temperate conifers. The high maintenance respiration of tropical trees is probably due to the high temperatures experienced. Rapid respiration is the main reason why, despite superior conditions for growth, wood production rates in tropical forests are not much larger than in temperate or even boreal forests (Jordan 1982).

The mechanical design of trees

One of the key constraints on the form of a tree is its necessary ability to support its own weight and the range of external forces, particularly those generated by the wind, it is likely to meet during its lifetime. Mattheck & Kubler (1995) provide an excellent overview of how tree design helps solve these mechanical problems. Failure usually involves localised regions of excessive stress. The mechanical design of trees can be summarised as growth and allocation of material to maintain a uniformity of stress within the tree and avoidance of critical stresses at particular sites. This is achieved by a variety of mechanisms including:

1. The passive bending of flexible parts. Not resisting the external forces minimises the loading, but large heavy stems would buckle under their own weight if they were very flexible. Therefore, the bending strategy is confined to small trees, the ultimate branches of large trees and notably to bamboos (Mosbrugger 1990). Rheophytic trees and shrubs that survive periods of inundation in fast-flowing rivers use their flexibility, and tenacious root systems, to survive. Tree crowns require some flexibility to reduce their drag against the flow of the moving air on windy days. Under heavy wind loadings trees may produce 'flexure' wood (Ennos 1997), which has more markedly spiral grain. It is denser than normal wood, with shorter, thicker-walled cells. The cellulose microfibrils are also wound at a larger angle than usual. Theoretically, this should result in wood of lower stiffness with the same breaking strain. As yet, only the former has been demonstrated experimentally (Ennos 1997). Branches may have denser wood than tree trunks so that they can be relatively thinner and thus bend and 're-configure' more easily. Putz et al. (1983) noted that is was low-density-timber trees that tended to lose limbs in windstorms in Panama.

2. The design of the wood as a cellular composite material. The various components of the wood are arranged to meet different mechanical challenges. While the axial arrangement of fibres and tracheids provides most of the strength of the stem, ray fibres resist critical shear and torsional stresses. Monocotyledonous wood, with its diffuse sclerenchymatous material, lacks efficient cross bracing and is not good at resisting forces oriented in directions other than parallel to the main axis. This is probably why the dendroid monocotyledons are often unbranched and have limited crown development (Mosbrugger 1990). Branch insertion and spreading crowns inevitably generate lateral stresses. Palms overcome this limitation to some extent by having enormous leaves, their equivalent of whole branches. These are supported by highly fibrous leaf bases that clasp the main trunk tightly.

3. The outer shape of the tree and the internal quality of the wood are optimised to meet the external forces. Tree trunks are generally roughly circular in cross section. Angled stems would be disadvantageous because they would lead to high stress intensities at the angle. Buttresses are another example of using growth to overcome mechanical challenges (see below). Significant positive gradients in wood density from pith to bark were found in

16 out of 20 species of tree tested in Costa Rica (Wiemann & Williamson 1989). Palms also tend to have denser wood around the periphery of the base of the trunk (Rich 1987) (Fig. 2.6). Wiemann & Williamson (1989) found that these gradients were most marked in low-density timbers such as Ceiba pentandra (Fig. 2.7). Studies of such trees have shown that the wood density gradient is brought about by changes in the density of the wood produced as the tree ages, rather than by localised growth of denser wood (Rueda & Williamson 1992; de Castro et al. 1993). The sapling wood of the fast-growing Laetia procera had half the density of the adult tree, but the sapling and adult woods of the shade-tolerant Dipteryx panamensis were of the same density (King 1996). It is possible that the intense competition for height growth in

Fracture Height Symbol

Figure 2.6 Schematic density distribution in the stem of the mature palm tree (Cocos nucifera). After Killmann (1983).

Radius (cm)

Figure 2.6 Schematic density distribution in the stem of the mature palm tree (Cocos nucifera). After Killmann (1983).

Figure 2.7 Specific gravity as a function of distance from the pith for two trees (different symbols) of Ceiba pentandra. After Wiemann & Williamson (1989).

juveniles of light-wooded tropical trees favours the 'low cost-high risk' use of very soft wood for stems and branches. The major stresses in tree trunks are produced through bending and occur in the outermost portions of the stem cross section. The central portion mainly supports the weight of material above. Removing the central portion, i.e. making a hollow tree, alters the mechanical properties of the trunk relatively little as long as the ratio of the thickness of the outer solid wall (t) to the diameter of the hollow (D) remains fairly large. The weakening of the stem by removing the central cylinder is made up for by the mass lost by its removal (Mosbrugger 1990). Hollow stems make mechanical sense and are found in some fast-growing species including members of the genera Cecropia and Musanga, where t/D usually falls in the range 1/3 to 1/5. However, the engineering difficulty with hollow stems is branch insertion. A hollow branch meeting a hollow stem tends to lead to high stresses at the point of insertion. Cecropia generally has short, stiff branches to keep these stresses low, and may rely on pith turgor to reduce the risk of buckling in smaller branches. Large, old trees often have hollow trunks. They appear to be mechanically effective until t/D approaches a critical value of 0.3 (Mattheck & Kubler 1995).

4. Production of reaction wood at points of high stress intensity. Reaction wood differs between conifers and dicotyledonous trees. Conifers produce compression wood that is rich in lignin and has a high microfibril angle in the S2 walls of the wood cells. The conifer reaction wood has a high strength in compression and is produced at sites of high compression such as the underside of main limbs. Dicotyledons produce tension wood that has little lignin and a low microfibrillar angle. It is grown in sites of high tension stresses such as the upper sides of large branches.

5. Internal pre-stressing of wood to counterbalance critical loads. There is evidence that tree trunks are pre-stressed, with the surface of the trunk under tension and its core under compression. This improves the tree's performance against its most likely mode of general failure: the compressive stresses generated through bending on the side of the tree in the direction of the bend (the leeward side if it is the wind that is causing bending). If the wood in this region starts under tension it will take a greater force to reach critical compressive loads than for a trunk without pre-stressing. The pre-stressing will mean higher tension stresses on the windward side of the trunk, but wood has greater tensile than compressive strength.

Buttresses

A prominent feature of the lowland tropical forest is the relatively high frequency of buttress roots growing from the base of large trees. In a survey of trees in a forest in the Malay Peninsula, 41% of individuals of 45 cm dbh or greater had buttresses to 1.3 m or higher (Setten 1953). A figure of 37% was found for the same tree-size and buttress-height groups in Venezuela (Rollet 1969). Eighteen out of 78 species (23%) of more than 10 cm dbh had buttresses at Kibale in Uganda (Chapman et al. 1998), with larger size classes tending to have a higher frequency of buttressing. Buttresses are variable in size and shape among species, being very tall in some. For instance, Rollet (1969) reported Sloanea guianensis and Ceibapentandra to have them extending to more than 7 m up the tree base. Some tropical tree species are nearly always found with buttresses (88% of Intsia palembanica trees surveyed (Setten 1953)), some rarely if ever have them (0.8% of Calophyllum), and for the rest buttressing varies among sites and genotypes. The common use of buttress form as an aid to species identification in manuals of forest botany indicates a strong genetic control on buttress design.

Engineering arguments have frequently been put forward to explain why trees need buttresses. Their name, derived from the architectural devices used to keep buildings standing, implies a mechanical function. However, only recently have the principles of physics been applied to trees in order to develop a functional explanation of buttressing in trees (Mattheck 1993). The basic premise is that buttresses are part of the tree's mechanism for the efficient transmission to the ground of the stresses caused by loading due to wind (Ennos 1993a). Permanent anchorage of the tree necessitates the transmission of the stresses. If not the tree will fall. A tree which has a deep and thick tap-root immediately beneath the trunk can transmit forces directly into the ground. However, many trees have their largest roots radiating from the base of the trunk in the upper layers of the soil, creating a shallow plate of woody roots. Such root plates are easy to observe on wind-thrown trees in any tropical rain forest. Sinker roots, descending from the plate deeper into the soil, transmit the stresses generated in the aboveground parts of the tree into the ground. Mattheck proposed that very large forces would be generated at the junction between the base of the trunk and the lateral roots when the tree is subjected to loading by the wind. Such forces could easily snap or split the roots at the junction with the trunk. Responding to these stress concentrations, the juvenile tree develops outgrowths between the trunk and the main lateral roots that eventually become buttresses. These brace the lateral to the trunk and allow more efficient transmission of stress with a reduced stress concentration at the junction between trunk base and root.

Buttresses can probably function as both tension and compression elements depending on the direction of the wind (Fig. 2.8). On the windward side of the trunk the bending of the tree tends to pull the roots out of the ground. The buttresses transmit this tension smoothly to the roots, acting like guy ropes. The root system is pushed into the ground by the compressive forces generated on the leeward side of the tree. Here the buttresses act as props spreading out the stress. Ennos (1995) used electronic strain gauges to demonstrate that the qualitative patterns of strain within buttressed trees were as Mattheck had predicted, with the highest strains along the tops of the developing buttresses.

The best experimental test of Mattheck's hypothesis of the functional significance of buttresses was conducted on young trees in the lowland rain forest in the Danum Valley, Sabah, Malaysia (Crook et al. 1997). Two groups of trees were studied. Mallotus wrayi represented a typically unbuttressed species. A pinnate-leaved, typically buttressed, morphospecies was found to include individuals of several species of Aglaia and Nephelium ramboutan-ake. The buttressed trees did not always have sinker roots, and they did sometimes have tap-roots. The trees were subjected to imitations of severe

Buttress Root Simulation

Figure 2.8 Mattheck's model for the function and development of buttresses. If a tree is pushed over by the wind (a) the bending force is transmitted smoothly to lateral sinker roots by the buttresses. Windward sinkers resist upward forces and the buttress is put in tension while leeward sinkers resist downward forces and the buttresses are put in compression. Parts (b), (c) and (d) show successive stages in Mattheck's simulation of buttress development. When the trunk is pulled over stress is concentrated at the top of the junction between the lateral root and the trunk (stippling). Growth in heavily stressed regions (b) results in the formation of buttresses (c, d) and a great reduction in stress concentrations. After Ennos (1995).

Figure 2.8 Mattheck's model for the function and development of buttresses. If a tree is pushed over by the wind (a) the bending force is transmitted smoothly to lateral sinker roots by the buttresses. Windward sinkers resist upward forces and the buttress is put in tension while leeward sinkers resist downward forces and the buttresses are put in compression. Parts (b), (c) and (d) show successive stages in Mattheck's simulation of buttress development. When the trunk is pulled over stress is concentrated at the top of the junction between the lateral root and the trunk (stippling). Growth in heavily stressed regions (b) results in the formation of buttresses (c, d) and a great reduction in stress concentrations. After Ennos (1995).

wind loadings by means of a winch. In buttressed trees without sinkers, failure was by breakage of leeward laterals and pulling out of windward ones (Fig. 2.9). With sinkers, windward laterals suffered delamination. The lateral roots were tall rectangles in cross-section, up to eight times tall as wide. This helps the roots resist bending up or down. Laterals were relatively ineffective at anchoring in non-buttressed trees. The laterals pulled out, or bent in compression, easily. The buttressed individuals were about twice as stable as the unbuttressed trees of the same size. The study confirmed Mattheck's predictions that buttresses could be effective both as tension and compression elements, stopping roots either being pulled out of the ground or pushed into it. The presence of sinkers on the buttresses improves their effectiveness in tension considerably, but even without sinkers they work well in compression. Possibly, as the trees grow larger more sinker roots develop. Reaction wood is not commonly found in the buttresses of tropical dicotyledonous trees (Fisher 1982). The increased tensile strength of reaction wood would possibly carry a disadvantage of reduced compressive strength.

Kaufman (1988) has argued that buttress formation is largely the result of a developmental crisis in the tree's life due to factors such as crown asymmetry or sudden exposure to high winds after gap formation or growth above the main canopy. The buttress may persist on the tree despite the reduction of the factor that led to its initial formation. In support of this theory, Chapman et al. (1998) found that understorey species at Kibale, Uganda, showed significantly less buttressing than canopy and emergent species, even after a correction for individual size had been made.

Crown asymmetry may be quite common in trees of the tropical rain forest. The average degree of crown asymmetry of 127 trees (greater than 20 cm dbh) observed at Barro Colorado Island was such that three quarters of the crown area of a typical tree was on the heavy side (Young & Hubbell 1991). The trees tended to develop crowns into gaps or away from larger neighbours. Emergents were typically more symmetrical than individuals of smaller stature. Over a period of 6.7 years the more asymmetric individuals had a higher chance of falling than those with more symmetrical crowns, and they tended to fall towards the heavy side (Young & Perkocha 1994). Buttress formation in the trees was more pronounced on the side of the trunk opposite to the heavy side of the crown. On large trees, buttresses often grew in size more quickly than total tree size (Young & Perkocha 1994). This would decrease the effective bole length of the trees concerned and might help reduce the risk of trunk buckling. The buttresses make the base of the trees stiffer and possibly allow greater heights and wider crowns to be achieved relatively safely.

Another possible advantage of buttresses is a reduced requirement for

Arborescent Prop

Figure 2.9 Trunk and root movements during anchorage failure of buttressed and non-buttressed trees. (1) Buttressed tree without sinker roots (Aglaia affinis), (2) buttressed tree with sinker roots (Nephelium ramboutan-ake). Note: sinker roots may be present or absent on both Aglaia and Nephelium species. (1a) Buttressed tree without sinker roots. The tree is anchored into the ground by the thick buttressed lateral roots and the tap-root. (1b) As the tree is pulled over, the trunk rotates about a point just on the leeward side. Initially, the roots firmly anchor the tree in the ground, the leeward lateral resisting bending being pushed into the ground and the tap-root resisting uprooting. The windward buttress, held in the ground by fine roots only, uproots easily. (1c) As the test proceeds the leeward buttress finally fails, breaking towards end of the buttressing. The centre of rotation (C.O.R.) changes so that the tree rotates about this leeward hinge and the tap-root is levered out of the ground or breaks. (2a) Buttressed tree with sinker roots. The tree is anchored into the ground by the thick buttressed laterals by their sinker roots and by the tap-root. As the tree is pulled over material in the base of the trunk. In some species, the trunk has a narrower diameter beneath the top of the buttresses than above them. Presumably the transmission of stresses through the buttresses allows the tree to remain mechanically viable with less material in the lower butt. This can be seen at its most extreme in species with highly developed stilt roots. These, we must presume, act like flying buttresses, transmitting stresses widely to the extensive lateral root system. In some stilt-rooted species the primary stem is extremely thin beneath the stilt roots. It is much too narrow to support the weight of the tree above. The development of stilt roots allows palms to gain more height for a given stem diameter than those without stilt roots (Schatz et al. 1985). Fisher (1982) found reaction wood in the stilt roots of Cecropia species, possibly indicating the requirement for a greater tensile strength in buttresses of more limited cross-sectional area. Stilt roots are not a common feature of trees of well-drained lowland rain forest. For instance, only 5 out of 246 species surveyed in Gabon, West Africa, had stilt roots (Reitsma 1988). Species of swamp forest are much more frequently stilt-rooted.

Using an approach based on allometry, Ennos (1993b) has shown that the mechanical efficiency of a root-plate system increases with plant size because the anchorage produced by the root-soil plate (largely provided by its weight) increases with size faster than stem strength. To be effective in a large tree, a tap-root needs to be of similar dimensions to the trunk, or even larger. In

(2b), the tree rotates about a point just on the leeward side. Initially the roots firmly anchor the tree in the ground, the leeward lateral resisting being pushed into the ground and the tap-root resisting uprooting. The windward buttress, held in the ground securely by the sinker roots, also withstands uprooting and instead begins to delaminate. As the test proceeds (2c) the leeward buttress finally fails, breaking towards the end of the buttressing. The centre of rotation changes so that the tree rotates about this leeward hinge and the windward root continues to delaminate. (3a) Non-buttressed tree (Mallotus wrayi). The tree is anchored into the ground by the tap-root and to a lesser extent the lateral roots. (3b) As the tree is pulled over, the tree rotates about a point just on the leeward side of the tap-root at a depth of ca. 0.5 m. The leeward laterals are pushed only slightly into the soil and then buckle whilst the laterals on the windward side resist being pulled up, acting in tension. The tap-root pushes into the soil on the leeward side both bending slightly and rotating above the centre of rotation and below this bends and moves slightly windward. A crevice is formed on the windward side as the tap-root rotates. As the test proceeds (3c) these root movements continue, the leeward laterals buckling, the windward laterals uprooting and the tap-root pushing into the soil, increasing the size of the crevice. After Crook etal. (1997).

tropical soils, it may be difficult for a tree to produce as large a tap-root to sufficient depth as is required for adequate anchorage, particularly in species with heavy timber. Crook & Ennos (1998) analysed the mechanical failure of tap-rooted Mallotus wrayi trees at Danum Valley in Sabah. In smaller trees the trunks snapped when heavily loaded by means of a winch, but in larger individuals failure was through the tap-roots being displaced. This indicates an inability by large trees to produce big enough tap-roots to anchor themselves effectively.

In conclusion, there is now evidence that buttresses are effective structures that play an important role in supporting and anchoring large trees. However, many questions remain to be answered. Do the many unbuttressed big trees in the forest all have large tap-roots? Are buttresses and root-plate systems cost-effective in terms of material required to build them? Do buttresses perform other roles such as occupying space on the forest floor to reduce the likelihood of establishment by competing trees?

Leaning trees

Most trees stand upright, even on sloping ground. This is because it is mechanically far more efficient to place the support directly under the crown and keep the main stresses within the line of the trunk and trunk base. In addition, leaning trees gain height at a greater construction cost per unit height, which will result in a slower rate of height growth as well. Loehle (1986) pointed out that most trees that do lean are small, or are fast-growing species with early reproduction, or lean into conditions of permanent high light such as out over rivers. The costs of leaning in small trees are less than in large ones because they can utilise the elastic properties of wood to support the lean with little extra investment in support. Large trees need to brace the tree with thicker trunks and more massive tree bases if they are to lean. These represent major investments that will only be worthwhile if the rewards of leaning are high. Fast-growing species may risk death through shading if they do not lean into nearby gaps. It is also possible that their low-density wood makes leaning less problematic. Ishii & Higashi (1997) came to similar conclusions in comparing small and large trees growing on slopes where the gradient of relative illumination is perpendicular to the slope. The quantitative model they derived has been criticised as unrealistic (Loehle 1997), even though its general predictions were found to be correct for a comparison of two species growing on slopes in warm temperate rain forest in Japan. None of these hypotheses have been explicitly tested for tropical rain forest trees. Riverbank trees have been noted as often leaning. For instance, Dipterocar-pus oblongifolius is characteristic of the banks of some rivers in the Malay Peninsula and Borneo, and habitually leans, often out over the river (Whit-more 1975).

Tree fall

Despite the ability of trees to grow in ways that reduce the likelihood of localised high stress intensities, mechanical failure is common in tropical rain forests. On Barro Colorado Island in Panama, a survey of 310 fallen trees (Putz et al. 1983) found that 70% of the trees had snapped trunks, 25% had uprooted and 5% had broken off at ground level. Uprooted trees tended to be larger than snapped ones, with trunks of greater diameter for a given height, and to have denser, stiffer and stronger wood. The degree of buttressing did not appear to be related to tendency to snap or uproot. Susceptibility to hurricane damage was found to be negatively correlated with wood density in tree species on Puerto Rico (J.K. Zimmerman et al. 1994). However, for five species from Kauai, Hawaii, stem elastic modulus for bending was a better correlate of the likelihood of snapping during a hurricane than wood density (Asner & Goldstein 1997). The stiff stems with low elastic modulus were more likely to survive the hurricane without snapping. The mechanical properties of the wood of a species influence its ability successfully to resist wind loading and other external stresses, but more research is needed to pinpoint which properties are the most influential.

It must not be assumed that fallen trees always die. Of 165 trees of 10 cm dbh and above that snapped in the period 1976-1980 on Barro Colorado Island, 88 (53%) re-sprouted from the broken base, of which 26 (16%) were still alive in 1987 (Putz & Brokaw 1989). In the Atlantic forests of Brazil, 82 out of 100 uprooted trees had new sprouts, mostly from the base of the trunk (Negrelle 1995). In both studies it was noted that pioneer species were poor at re-sprouting. Putz & Brokaw (1989) found that smaller trees were more likely to re-sprout successfully in the long-term. Negrelle (1995) noted that Tapirira guianensis was particularly good at re-growing after treefall, and was able to re-sprout from the whole length of its trunk.

Sap ascent

One of the marvels of nature is the apparent ease with which big trees lift large volumes of water to great heights while they transpire. Large quantities of water are needed by a tree crown, not only to replace the water inevitably lost during the uptake of carbon dioxide from the atmosphere during photosynthesis, but also because the evaporative loss of water is a highly effective way of cooling leaves exposed to the intense tropical sun. The xylem sap ascends the tree through the tiny-diameter pipes made by columns of tracheids or vessel elements. The driving force for the ascent against gravity is the difference in free energy between water in the gaseous state in the atmosphere around the leaves and their transpiring surfaces and the liquid water in the roots. The widely accepted cohesion-tension theory of xylem sap ascent proposes that the evaporation of water from cell walls in the leaves creates microscopic menisci that produce very strong tensions in the continuous columns of water that stretch from these sites to the roots through the xylem conduits. The cohesive properties of water allow the columns under tension to be pulled up rather than break and hence provide the transpiration stream. The cohesion-tension theory predicts that high tensions should be found in the xylem. At least 1 bar (0.1 MPa) is needed to overcome gravity for each 10 m increment in height, and an equivalent amount is probably required to counter the internal resistance to flow of the conduits.

Measurements made with the Scholander pressure bomb were seen as a vindication of the cohesion-tension theory. The gas pressure applied in the bomb to express sap from the test shoot was assumed to equal the xylem tension, and the values measured generally fell within the range predicted by the cohesion-tension theory. However, the pressure bomb does not measure xylem tension directly. It is possible to construct a probe with a minute pressure sensor that can be inserted right into vessels to measure the internal pressure. Such devices have consistently failed to record large tensions in plant shoots (Zimmermann et al. 1994), including high up in a tropical rain-forest tree. Other arguments against the cohesion-tension theory include theoretical and experimental studies of water columns that make it seem unlikely that the predicted tensions could exist in xylem (Smith 1994). The proponents of the cohesion-tension theory have launched a series of counter-arguments and experimental demonstrations to back continued support of the theory. The pressure-probe method is susceptible to technical scepticism. The probe tip is little smaller than the bore of the conduits being measured, and the technique requires that the hole in the vessel wall created by the entry of the probe seals perfectly so that neither the pressure in the conduit alters, nor air enters and causes embolism. The fluid in the probe is liable to cavitation because of gas seeds on the internal surfaces (Wei et al. 1999). Milburn (1996) has also pointed out that the probe only accesses the surface vessels. These are liable to have lower tensions than average, either because they are still-living proto-xylem or because water recycling from phloem transport, so-called Munch water, rehydrates the outermost vessels. Others have used centrifugation techniques to show that significantly negative water pressures exist and can be maintained in xylem and that they correlate well with pressure-bomb measurements (Holbrook et al. 1995; Pockman et al. 1995). They also argue that glass capillaries used in experimental studies of water columns and their cohesiveness may not be realistic analogues of xylem conduits.

Recently, Canny (1997a,b) has used a rapid freezing technique to visualise the activity of xylem conduits in sunflower petioles. The surprising result of this study was evidence of frequent cavitation of vessels at relatively low tensions and their rapid re-filling by the movement of water from parenchyma cells around the conduits. The proportion of vessels cavitated on a day course did not coincide with the change in xylem tension as measured by a pressure bomb: there were fewer vessels cavitated when the bomb balance pressure was greatest. This is at variance with classic cohesion-tension theory. Canny supports a compensating-pressure theory of xylem function (Canny 1995). This proposes that xylem operates at much lower tensions than cohesion-tension theory predicts because a compensatory positive pressure from surrounding cells in, for example, the stele, ray parenchyma or phloem, allows this to be possible. In addition, the marked ability to re-fill cavitated conduits means that water supplies can be maintained by the xylem when tensions exceed the normally low operating values. The compensating-pressure theory has in turn been met with scepticism. Its formulation does not comply with generally accepted models of biophysics (Comstock 1999). Killed stems do not show differences in vulnerability to embolism that would be expected if compensating pressure from nearby living tissues is needed to maintain the flow within the xylem (Stiller & Sperry 1999).

In conclusion, the recent doubt has spurred researchers to produce stronger validation for the cohesion-tension theory.

Tree hydraulics

Techniques for accurately measuring the rate of sap flow in trees and the internal resistance of the plant to the flow are now available. Using the compensation heat-pulse velocity technique, Becker (1996) found that the daytime sap-flow rate of individual trees was strongly, linearly correlated to crown area, with different species, and different sites of different forest type, fitting the same regression (Fig. 2.10). The regression slope changed between wet and dry seasons. Similarly, Andrade et al. (1998) found that tree size was the main determinant of whole-tree water use in Panama, with a 35 m individual of Anacardium excelsum estimated to use 379 litres per day. Sap-flow rates correlate with incident solar radiation, which is the driving force for transpiration (Fig. 2.11).

Stems provide a water store that can be tapped for evapotranspiration when water demand is high. The size of the water store rises exponentially with tree size and the volume of water used from the stem store can represent 9-15% of the total daily water use (Goldstein et al. 1998). The stem is recharged with water during periods of low transpirational water demand.

Tropical tree species show a wide range in the values for the different measures of hydraulic conductivity that can be estimated (Tyree & Ewers 1996). In general, tropical trees have average to high values for stem conductivity in comparison with temperate trees and conifers. Fast-growing, soft-wooded species typically have the highest leaf-specific conductivity values, that is, stem conductivity per unit leaf area distal to the measured segment. Therefore they should be able to deliver large volumes of water quickly to each leaf and maintain high rates of evapotranspiration at relatively low water potential gradients. This is achieved not by having a high conductivity per unit sapwood cross section, but by having a high proportion of sapwood in the branch cross section.

The branch-specific leaf area to sapwood cross-sectional area ratio (LA/ SA) appears to be an important determinant of branch transpiration rate. Individuals of various sizes of four different species growing in Panama followed essentially the same relationship between conductance and evaporative demand when scaled by LA/SA (Meinzer et al. 1997). In other words, transpiration rate per unit leaf area was identical when LA/SA was allowed for, and thus LA/SA was more influential on transpiration than stomatal conductance in many instances because of decoupling through low boundary

0 100 200 300

Figure 2.10 Relation between mean daily daytime sap flow and projected crown area for dipterocarp-forest trees (triangles) and heath-forest trees (circles) during wet (solid) and dry (open) periods in Brunei. The reduced major axis regression slopes of the wet and dry periods were significantly different. After Becker (1996).

0 600 1200 1800 2400

Local time

Figure 2.11 Daily course of short-wave radiation and sap flow in a Neonauclea tree for a sunny day during the dry season on Rakata (Krakatau), Indonesia. After Bruijnzeel & Proctor (1995).

0 600 1200 1800 2400

Local time

Figure 2.11 Daily course of short-wave radiation and sap flow in a Neonauclea tree for a sunny day during the dry season on Rakata (Krakatau), Indonesia. After Bruijnzeel & Proctor (1995).

layer conductance, particularly in still-air conditions.

Tropical rain-forest species appear to be highly vulnerable to xylem embolism, particularly in comparison with species from seasonally droughted forests (Tyree et al. 1998a). A comparison of species from mixed dipterocarp forest and heath forest from Brunei failed to show any significant difference in vulnerability to embolism between the sites (Tyree et al. 1998a). The freely draining sandy soils of the heath forest might have been expected to select for species with less vulnerable xylem than the dipterocarp forest.

The hydraulic conductance of leaves of tropical tree species has been found to span a similar range to that shown by temperate species (Tyree et al. 1999). The measured values are possibly low enough to indicate that, in some species, the leaf provides the largest resistance to water movement in the soil-plant continuum. Specific leaf area showed no obvious relationship to hydraulic conductance.

Using estimates of xylem sap-flow rates and nutrient concentrations, Barker & Becker (1995) were able to calculate the delivery rate of nutrients to the crowns of Dryobalanops aromatica trees in Brunei. The xylem sap concentrations of important elements were found to be such that N > K > Ca > P. The nutrient concentrations were not related to tree size, and often showed a positive correlation with sap-flow rate. Delivery rates per tree peaked at about 40mmol h"i for N, 30mmol h"i for K, 5mmol h"i for Ca and 3mmolh_1 for P.

Tree architecture

The form of construction of the above-ground parts of a tree-the general pattern in space of stems and branches-is governed by two factors: the genetic bauplan of shoot growth and branching and the fate of those shoots through their interaction with the environment. Halle & Oldeman (1970, 1975) proposed a series of 23 architectural models for tree form, each named after a botanist who has contributed to the study of plant morphology. This system is a typology of the genetically controlled component of aerial axis organisation based on character states for number of main axes, branching, presence of resting periods in development, relative apical dominance and positions of inflorescences. Recently, Robinson (1996) has used symbolic logic to represent and analyse the Halle & Oldeman system. He inferred the possible existence of several, as yet, unreported models.

Many trees in the forest do not appear to conform to one of the 23 basic models, even when allowance is made for loss of branches due to damage or disease. This failure to conform is often due to a process that Halle & Oldeman called reiteration (Halle et al. 1978). One or more meristems, instead of carrying on with the original model in the appropriate place, begins again, sometimes with a completely different model altogether. Reiteration appears to be an important process in the development of tree form. It allows the tree greater architectural flexibility.

Some trees never reiterate but conform to a single model throughout their lives. For large trees, Halle (1986) referred to this as gigantism and pointed out three groups where this occurs quite commonly. These are conifers (e.g. Araucaria spp.), angiosperm families generally regarded as relatively primitive (e.g. Myristicaceae) and fast-growing, soft-wooded tropical trees such as Cecropia species. All the groups use simple, sparsely branched crowns to fill space with foliage. In the conifers, long-lived leaves persisting far back on the branches make them efficient light-harvesting units. In the fast-growing species, the huge leaves at the ends of the branches produce a crown with an outer covering of a layer or two of leaves that is supported with relatively low investment in wood. These tropical parasol trees can be considered neotenic: reproducing in their 'juvenile' architectural phase.

The architectural models are a useful tool in the description of tropical trees and have a clear phylogenetic component (Keller 1994), but as yet relatively little progress has been made in relating particular models with particular ecological roles (Bongers & Sterck 1998). This may be because the characters used to define the models are not precise enough to mirror any ecologically meaningful relationships of crown form. Two species of the same architectural model can look very different in the forest, and alternatively two species of different models can appear very similar.

This does not mean that tree morphology is not relevant to ecology. Branching patterns affect leaf display, efficiency of mechanical support and supply of water to foliage. All these factors are likely to act as selective pressures in the evolution of crown form. Comparison of two species of treelet in Peruvian lowland rain forest indicated the importance of crown form (Terborgh & Mathews 1999). Neea chlorantha was found in sites with more direct overhead illumination on average than Rinorea viridifolia, generally small gaps in the canopy. Neea was more highly branched with a shell of drooping leaves on the outside of the crown. Rinorea had whorled branches in tiers with planar foliage arrays. The latter design appeared more efficient at intercepting the larger amounts of lateral light available in forest understorey sites.

Allometry

The change in relative dimensions of an organism as it grows, or across a range of organisms of different size, can reflect a range of constraints involved with growth. These can be physical where scale-dependent forces necessitate proportionally less, or more, investment in certain organs if the organism is to maintain a uniform likelihood of damage or destruction from the force concerned. Alternatively, the size-related investment in different parts may reflect scale-dependent evolutionary pressures to optimise performance at each growth stage. Allometric studies are also useful as empirical tools to gain insight into the confidence with which certain measures can be used as proxies for others. For example, many individual and stand characteristics are often estimated from tree bole diameter measurements.

The most frequently studied allometric relationship of trees is that between stem diameter (D) and height (H). A major goal has been to assess the magnitude of the allometric constant a where D oc Ha, and to ascertain its nearness to predicted values based on engineering or other principles. A linear relation (a =1) between two allometric variables is referred to as geometric similarity. A free-standing column of uniform taper and material composition requires its diameter to increase faster than the height in order to maintain a constant mechanical stability. Engineering predicts a = 1.5 for this elastic similarity condition. Studies of the height-diameter relation for tropical trees over a range of heights, either for individual species or for multi-species samples, show that elastic similarity is probably a reasonable approximation for the height-diameter relation (Rich et al. 1986; King 1991a; O'Brien et al. 1995), as shown in Figure 2.12. This implies that young trees have relatively more slender trunks than large old ones. (This is a fact of which most people are consciously unaware yet use subconsciously in determining the relative height of trees when no scale object is available to view.)

Figure 2.12 Allometry of stem diameter (dbh) and height for mixed dicotyledonous trees, the canopy dominant tree Pentaclethra macroloba, the gap-dependent tree Pourouma aspera and the arborescent palm Socratea durissima in tropical rain forest of Costa Rica. P. macroloba comprises 20% of the individuals. The solid line is a linear regression. The upper dashed line is the allometric curve for record-size North American trees and the lower dashed line a theoretical buckling limit for an 'average' tree, beyond which the tree will buckle under its own weioht After Rich pi nl H Qftfi^

Height (m)

Figure 2.12 Allometry of stem diameter (dbh) and height for mixed dicotyledonous trees, the canopy dominant tree Pentaclethra macroloba

0 0

Post a comment