Journal of Experimental Botany, Vol. 48, No. 314, pp. 1703-1716, September 1997 Journal of Experimental Botany The function of buttress roots: a comparative study of the anchorage systems of buttressed (Aglaia and Nephelium ramboutan species) and non-buttressed (Mallotus wrayi) tropical trees M.J. Crook1, A.R. Ennos and J.R. Banks School of Biological Sciences, Stopford Building, University of Manchester, Oxford Road, Manchester M13 9PT, UK Received 18 December 1996; Accepted 15 March 1997 Abstract The anchorage mechanics of mature buttressed trees of Aglaia and Nephelium, and of non-buttressed Mallotus wrayi have been investigated by combining a study of the morphology of their root systems with a series of anchorage tests. Both types possessed tap roots, but only buttressed trees possessed sinker roots, which branched from the ends of the buttresses. The anchorage strength of the buttressed trees was almost double (10.6 kNm) that of the unbuttressed ones (4.9 kNm), and the maximum moment was generated at lower angles. In buttressed trees, the leeward buttresses were pushed into the soil before bending and eventually breaking towards their tip, whilst the windward buttresses pulled out of the soil or delaminated if they possessed sinker roots. The tap root rotated in the soil to windward. In contrast, during failure of unbuttressed trees the tap root both moved and bent towards the leeward, the windward roots were pulled out of the soil, and the leeward laterals simply buckled. Strains along buttresses were much higher than along the laterals of unbuttressed trees. These results suggest that buttresses act in both tension and compression and make a much larger contribution to anchorage than the thin laterals of nonbuttressed trees. The relative contribution of the buttresses was determined by carrying out a further series of anchorage tests in which both buttressed and unbuttressed trees were pulled over after all their 1 laterals had been cut away. These trees were therefore only anchored by their taproot. Failure of both types was similar to intact unbuttressed trees, and they had similar anchorage strengths to each other, 4 kNm, around 80% of the value for intact non-buttressed trees, but only 40% of the strength of intact buttressed trees. Buttresses therefore contribute around 60% of the anchorage of buttressed trees, producing around six times more anchorage than the thin laterals of unbuttressed trees. Key words: Anchorage, root architecture, sinker roots, tap roots, root bending strength, buttresses. Introduction Root buttresses, triangular flanges joining the roots and lower trunk, are a characteristic feature of many tree species of lowland tropical rainforest (Richards, 1952) which have intrigued biologists for generations. This is largely because they are very rare in temperate zones, though in tropical rainforests they are far from universal. There have been various theories proposed to the function of buttresses. For example, Black and Harper (1979) suggested that the buttresses serve to prevent vines climbing up them. This hypothesis, was later discredited by Boom and Mori (1982). The general consensus, however, is that buttresses, as their name suggests, have a mainly mechanical role, helping to provide stability and so preventing the tree from toppling over (Senn, 1923; Richards, 1952). The evidence researchers brought to To whom correspondence should be addressed at: Harper Adams Agricultural College, Newport, Shropshire TF10 8NB, UK. Fax: + 44 1951 825 340. E-mail:[email protected] Oxford University Press 1997 1704 Crook et al. support this theory is, however, mostly indirect; most workers have attempted to deduce their function by correlating the development of the tree with the environmental conditions in which they grow (Richards, 1952; Richter, 1984; Lewis, 1988). It has also been stated that trees with well developed tap roots rarely produce buttresses (Francis, 1924; Navez, 1930; Petch, 1930; Corner, 1988) and the development of buttresses is also correlated with tree height; buttressing is more common in emergent and canopy trees than in those growing in the understorey (Richards, 1952; Smith, 1972). The extent of buttressing is also correlated with soil texture and depth. Buttresses tend to be larger in trees growing in weak silty soils (Richards, 1952), in shallow waterlogged soils (Richards, 1952) or where there is a shallow humus layer overlying rock or subsoil (Richards, 1952; Navez, 1930; Petch, 1930). There is also a correlation with wind direction; though buttresses are randomly orientated around the trunk, the buttresses on the side of the trunk which faces the prevailing wind tend to extend further from the trunk (Warren et al., 1988; Lewis, 1988). Whilst these studies have shown that certain factors affect buttress root development, only a few investigations have attempted to explain how the buttresses actually anchor the tree. Henwood (1973) suggested that buttresses act as tensile elements which increase the anchorage provided by the lateral roots on the windward side of the trunk by lengthening the moment arm about which they act. His conclusions were based on the observations of Senn (1923), Navez (1930) and Baker (1973) who noted that buttresses on the side of the trunk facing the prevailing wind tend to extend further along the trunk. However, this model of anchorage is unconvincing, as the lateral roots are poorly orientated to transmit forces. More recently, Mattheck has proposed a much more plausible explanation for the function of buttresses (Mattheck, 1991, 1993). Though most of the roots of tropical trees are superficial (Stark and Jordan, 1978; Jordan, 1982) they also possess sinker roots which branch from the laterals along the trunk and penetrate down into the subsoil (Jenik, 1978; Baillie and Mamit, 1983). Mattheck suggested that it is these roots which anchor the tree, those on the leeward side resisting downward and those on the windward side upward forces. Being vertically orientated, these roots are ideally placed to resist these forces. The buttresses, meanwhile, prevent the roots from splitting along their length or snapping by bracing the roots to the trunk, smoothly transmitting tension to the windward sinkers and compression to the leeward sinkers (Fig. 1). Whilst Mattheck's model of anchorage is mechanically feasible, to date, his theory has not been tested. In this paper, therefore, how the roots of mature subcanopy buttressed tress (Aglaia and Nephelium species) anchor Wind Force Tapered trunk Buttress To sinker From sinker To sinker Fig. 1. Mattheck's model for the function of buttresses If a tree is pushed over by the wind the bending force is transmitted smoothly to lateral sinker roots by the buttresses. Windward sinkers resist upward forces, the buttresses being put into tension, while the leeward sinkers resist downward forces, the buttresses being put into compression. the tree has been investigated by carrying out simulated windthrow tests, using techniques developed in earlier studies (Coutts, 1983a, 1986; Crook and Ennos, 1996) and examining their root morphology. To assess the contribution in anchorage strength that buttresses make to tree stability and to increase our understanding of why some trees produce buttresses whilst others do not, trees of the non-buttressed Mallotus wrayi were also winched over and the results compared. Materials and methods Field site All trees investigated in this study were growing in an area of primary dipterocarp rainforests just outside the conservation area at Danum Valley, Lahad Datu, Sabah, Malaysia. The original intention was to compare two species of tree, one buttressed and one non-buttressed. Aglaia affinis (family Meliaceae) was chosen as the buttressed species because it develops large buttresses whilst still small enough to be pulled over with a hand winch and seemed to have characteristic fluted trunks and red bark. However, it was impossible to identify the species accurately in the field and therefore leaf samples were taken from each tree and stored until they could be identified later by an experienced rainforest taxonomist. This meant that species were not identified properly until after all the tests were performed. As a result, therefore, as well as nine mature A. affirms examined, other tree species were also included in the tests. These were additional Aglaia species {A. affinis Meur, A. luzonensis (vid) Merr et Rolfe, A. elliptica Bl and A. crassinervia Kurz ex Hiern) and Nephelium ramboutan-ake (Labill). However, their morphological similarity resulted in sufficient similarity in their mechanical behaviour to describe these trees together as a single population of buttressed trees. All these Anchorage mechanics of buttressed trees species are of similar size; Aglaia species typically grow up to 25 m and N. ramboutan up to 35 m (Ng, 1989). The non-buttressed M. wrayi (family Euphorbiaceae) was used as the non-buttressed tree species for this study, not only because of its ubiquity at Danum valley and its ease of accurate identification in the field, but because large M. wrayi grow to comparable sizes as the Aglaia and Nephelium species, growing up to a height of 24 m (Whitmore, 1972). The trees which were examined varied in girth at breast height (gbh) from 30-50 cm for the buttressed species and 25-45 cm for the non-buttressed M. wrayi. Although these size classes are slightly different this was unavoidable due to restraints in finding enough trees suitable for testing. Preliminary tests: root movements during uprooting Before any quantitative experiments were carried out the movements of the root system of buttressed and non-buttressed trees were examined qualitatively during simulated anchorage failure. Two buttress rooted trees, one Aglaia affinis and one Nephelium ramboutan, of girth at breast height (gbh) 35 cm and 42 cm, respectively, and two M. wrayi of gbh 33 cm and 34 cm were studied. Soil was dug away from one side of each tree, parallel to the direction in which it would be pulled over, making a trench 60 cm deep, 60 cm wide and extending 1 m either side of the trunk. Any lateral root growing out from the trunk and interfering with the digging of the trench was cut away with an axe. The location and orientation of the main lateral roots and any sinker roots were also noted. The tree was then winched over at about 15° min"' whilst paying particular attention to the movements of the roots, the soil and the centre of rotation of the tree. Sounds of roots breaking or being uprooted were also noted. Once the tree had toppled the root system was examined closely to see whether the roots had broken or for other signs that they had failed mechanically. Anchorage mechanics (a) Preparation and soil shear strength: Twelve buttressed (three Nephelium ramboutan-ake (Labill) Leech and nine Aglaia trees: six A. affinis Meur, one A. luzonensis (vid) Merr et Rolfe, one A. elliptica Bl, and one A. crassinervia Kurz ex Hiern) and fourteen non-buttressed Mallotus wrayi King trees were prepared for testing by cutting down the trunk at a height of 3 m above the ground. This is essential because the nearby trees would otherwise hinder the tests, preventing them from being winched over. Tests on the buttressed trees took place from November 1995 to January 1996 and on the non-buttressed trees during February and March 1996. Prior to tests, however, the shear strength of the soil around each test tree was measured using a shearvane, three readings being taken for each tree just prior to testing at a soil depth of 10 cm. These measurements were necessary because the tests were carried out over an extended period during which changes in moisture content of the soil might have affected soil strength and hence the anchorage strength of the trees (Spoor and Godwin, 1979; Ennos, 1990; Crook and Ennos, 1993). If the shear strength of the soil was higher than 100 kPa the tests were postponed until it had rained sufficiently for the shear strength to fall. A sling was looped around the trunk of the test tree at a height of 2.2 m from the ground and connected to the winch cable, and the winch (Haloagger RQ417/4), capable of producing a maximum force of 0.9 kN was secured to the base of another tree, typically 8 m away using a further sling. A 1705 force transducer capable of measuring forces up to 20 kN was shackled in between the sling and the winch cable and its output was recorded on an ADC 11 data logger (PicoLog, Hardwick, Cambridgeshire). Any slack in the winch cable was then taken up. (b) Anchorage tests: strain measurements along the trunk and roots-preparation: To gain further insight about how the roots anchor the tree, strain gauges were attached along the trunk and lateral roots, so 'mapping' the force flow along the trunk and into the ground when a tree is pulled over. All of the buttressed trees and five of the non-buttressed M. wrayi trees were tested. After examining the overall root system morphology only the straightest and most windward or leeward roots (depending upon the direction in which the tree was to be pulled over) were considered suitable for affixing strain gauges. This was to ensure that the force flow along the trunk would run straight along the length of the lateral root. The trunk and lateral root were then prepared for the attachment of strain gauges: first, a clean, smooth, woody surface 20 mm long and 15 mm wide just beneath the surface of the bark was made using a chisel at each of the locations at which gauges were to be positioned. The 5 mm length gauges (Kyowa KFG series, 5 mm gauge length, 120 ohm resistance, supplied by Graham and White Instruments, St Albans, Hertfordshire) were then glued in position with general purpose cyanoacrylate (superglue) and allowed to set. Up to eight gauges were glued along the trunk and the lateral root at the following points: (1) 1.0 m up the trunk on the leeward face. (2) 0.1 m up the trunk on the leeward face ('base'). (3) At the 'join' of the leeward lateral with the trunk, along the top edge. (4) 0.5 m along from the 'join' between leeward lateral and trunk, along the top edge. (5) 1.0 m up the trunk on the windward face (directly opposite gauge 1). (6) 0.1m up the trunk on the windward face ('base'). (7) At the 'join' of the windward lateral with the trunk, along the top edge. (8) 0.5 m along from the 'join', away from the trunk, between windward lateral and trunk, along the top edge. Gauges were connected to a Kyowa strain meter (model SM-60D) via a switching and balancing box (Kyowa SS-12R, Graham and White Instruments) along with a common 'dummy' gauge and zeroed. The dummy gauge reduces drift caused by temperature variations. Because the buttressed trees were also a part of a different experiment (presented in another paper, Crook and Ennos, in preparation) it was not possible to attach gauges to both the leeward and windward sides. Instead, six of the trees had gauges positioned on the leeward side only (gauges positions 1-4) and six trees had gauges positioned on the windward side only (gauges 5-8). All of the non-buttressed M. wrayi trees had gauges positioned on both the lee and windward sides (gauge positions 1-8). (c) Anchorage tests-testing: The winch handle was then cranked at a rate of one cycle every 5 s. After five complete cycles the winching was stopped so that any root movement or damage could be examined in greater detail. The process was repeated, cranking the winch a further five cycles and recording the loads. This procedure was repeated at 15, 20, 25, 30, 35, 40, 45, and 50 winches, approximating to a trunk lean of between 5° and 48° from the vertical. The strains along the trunk and root were also measured at each incremental cycle. 1706 Crook et al. Each cycle of the winch drew in 3.1 cm of cable, so it was also possible, by geometry, to calculate the angle through which the trunk had moved. The mechanism by which the tree failed was also recorded. Although measurements were made over an extended period (about 1 h) the soil will creep only slightly (Crook, 1994), and is therefore unlikely to influence the results. (d) Root system morphology: In order to allow the roots to be examined more easily after the tests, the trees were excavated and manhandled clear of the soil. The morphology and orientation of structural roots (defined as any root having a width greater than 1 cm at a distance of 10 cm out from the trunk) of each tree was then investigated. The number of laterals (buttressed or non buttressed) was recorded along with their height and width 10 cm from their join with the trunk. The distance along the laterals from which the sinker roots branched was measured. The tap roots had both their maximum and minimum width measured at the join with the trunk together with the length to a taper of 35 mm (in height) and their maximum depth. A record of the overall form of the root system was also made, noting the degree to which laterals branched and the extent of branching of finer roots. Measuring the morphology of all the laterals, tap roots and sinkers was sometimes difficult because on occasion the roots had snapped off and remained in the ground. These roots were dug up and 'pieced' back into position so that the measurements could be made. The orientation of individual roots was recorded in one of three categories: (1) windward root: any root emerging from the trunk on the counter-winchward side of the tree, (2) leeward root: any root emerging from the trunk on the winchward side of the tree and (3) side root: any root that emerged from the trunk on neither the windward or leeward sides was classified in this category. Root morphology was considered evenly distributed if the tree had both leeward and windward roots and asymmetrically distributed if either leeward or windward roots were absent. Because models of anchorage mechanics are based only on the overall form of root systems, this simple method of assessing root orientation was sufficient for this study. Hence, a detailed analysis of root architecture (Mardia, 1972; Henderson et al., 1983) was not made. (e) Damage to the trunk and root system: Damage to the trunk and root system was recorded, noting the extent of any delamination of the wood, together with the position along the lateral of any break and the height and the width of the root at the break. Components of anchorage To determine the extent to which the tap roots contribute to the anchorage strength of the tree two further A. affinis, one N. ramboutan and five M. wrayi were pulled over after all their lateral roots had been cut away using a hand axe. The trees were then winched over in the same way as the trees that had their root systems left intact. To ensure a valid comparison, the trees that were chosen were of the same gbh as trees that had been pulled over with their lateral roots intact. Four of the buttressed trees from the initial anchorage mechanics tests proved too well anchored to be pulled over with the winch. These trees also had their buttresses cut away and retested, winching them over as before. Despite inaccuracies caused by cutting the roots away and thereby altering the anchorage mechanics of the system (Ennos et al., 1993) this method is still useful because, although altered, the centre of rotation of the system remained on the leeward side and in a similar position to that of trees with an entire root system. Results Root system morphology All the non-buttressed M. wrayi and 11 of the 15 buttressed trees had symmetrical root systems with laterals emerging from all around the trunk. Of the remaining 4 trees with asymmetrically distributed roots 2 had no roots on the leeside (counter-winchward) and 2 had none on the windward (winchward) side. Details of the root system morphology of buttressed and non-buttressed trees are given in Table 1 which includes only trees with gbh greater than 30 cm and less than 40 cm. This was to ensure that the two tree populations were directly comparable. This meant that four of the buttressed trees (over 40 cm gbh) and five of the nonbuttressed trees (under 30 cm gbh) were excluded from the analysis. However, in order to increase sample size wherever possible, all trees were included in the analysis of the number of roots of a tree. This is valid because the number of lateral roots is determined whilst the tree is still small (Fayle, 1975; Coutts, 1983*). Both buttressed and non-buttressed trees produced a single tap root directly under the bole, but though the two tree types were similar in this respect their lateral root morphology was obviously very different: M. wrayi produced more laterals than the buttressed trees and each of these was circular in cross-section in contrast to the lateral roots of the buttressed trees which had a rectangular cross-section. These differences are quantified in the ratio of height of the root to its width, buttressed roots are almost eight times as high as they were wide, while in contrast, non-buttressed trees produced circular roots with a ratio of height to width of around 1 (Table 1). Only the buttressed trees produced sinker roots along their laterals. Surprisingly, however, this was only true in 10 of the 15 buttressed trees, and only 28% of the total number of roots examined. There was also a general trend that the proportion of buttresses with sinker roots increased with tree size. However, possibly because the sample size was small, and root variation large, this correlation was not significant (P = 0.096). The lateral roots of both species branched little, with the main roots remaining straight for over 2 m out from the trunk. The sinker roots of buttressed trees tended to branch at their tips into a mass of finer roots. Many of the tap roots of M. wrayi also produced laterals about half-way along their length. These roots, like the surface laterals, remained growing horizontally rather than down into the soil. Where these roots did exist, they were fewer in number and thinner than the laterals. Preliminary tests: buttressed trees Excavation revealed that of the two buttressed trees examined in the preliminary tests only the larger Anchorage mechanics of buttressed trees 1707 Table 1. Root system morphology of mature buttressed trees (Aglaia and Nephileum species) and non-buttressed trees (M. wrayi) (a) Overall root system morphology, (b) morphology of the tap roots, (c) morphology of the laterals, and (d) morphology of the sinker roots. Means, SE of the mean and the number of samples (n) are given. Note: All trees excavated were analysed for (a) ranging in gbh 30-50 cm in the buttressed trees and 25-40 cm in the non-buttressed trees, whilst only trees ranging in gbh of between 30-40 cm of both species were analysed in (b) (c) and (d) to ensure comparable population classes. Room system morphology Buttressed trees Mean (a) Overall root system morphology Number of tap roots Number of lateral roots Number of sinker roots Percentage laterals with sinkers (b) Tap root morphology length of tap root to a taper of 35 mm diameter (m) Depth of tap root (m) Maximum width at base: MaxWid (mm) Minimum width at base: Min Wid (mm) MaxWidjMinWid (c) Lateral root morphology Height at buttress join (mm) Width at buttress jom (mm) Height/width at buttress join (d) Sinker root morphology Distance along lateral (m) Diameter (mm) 0.9 3.8 1.1 30.2 Non-buttressed trees SE n Mean SE n 0.09 0.38 0.31 8.28 14 14 14 14 1.0 6.0 0.0 0.0 0.0 1.6 0.0 0.0 14 14 14 14 0.61 0.06 7 0.64 0.06 9 0.88 21.6 21.3 1.1 0.16 106 11.5 0.09 7 7 7 7 1.08 13.2 11.9 1.2 8.43 0.9 1.2 0.09 9 9 9 9 24.6 0.99 0.89 31 31 31 33.8 32.4 1.07 1.9 1.7 0.13 58 60 58 - - - 216.5 28.2 7.9 0.30 31.1 Nephelium had sinker roots growing down from its buttresses. However, despite this difference in root morphology the two trees failed in a similar way as the trees were pulled over (Fig. 2). Significant movements in the roots only occurred after the trunk had been displaced by c. 20° from the vertical. Both trees then began to rotate about a point on the leeward side of the trunk, at a depth of about 25 cm (Fig. 2-lb, 2b) and as the tree was pulled over the leeward buttressed lateral (or more accurately, the winchward lateral) was pushed into the soil. The buttress, with its high bending rigidity bent imperceptively as it was pushed into the ground. However, at a trunk inclination of 30-35° from the vertical these buttressed roots broke towards the end of the buttress (Fig. 2-2c). In contrast, behaviour of the windward buttresses differed between the two trees. In the case of the Aglaia affinis the buttress, which did not possess sinker roots to anchor it into the ground, was levered out of the ground (Fig. 2-lb). Snapping of finer roots was heard when the tree trunk was inclined by about 30° from the vertical. The windward buttress of the Nephelium was held securely in the ground by its sinker root (Fig. 2-2b). Consequently, as the test proceeded very little root movement was detected on the windward side. At c. 20°, however, the root began to delaminate, starting in the middle of the buttress about 10 cm out from the join with the trunk (Fig. 2-2b). Delamination propagated up the buttress and 1 m up the windward side of the trunk as the test proceeded (Fig. 2-2c). In both trees the tap roots rotated about the centre of 0.049 4.25 8 8 rotation, coming up on the windward side after originally being confined by the surrounding soil. Non-buttressed trees As the trees were pulled over, the two M. wrayi trees failed in a similar manner to each other but in a dissimilar way to the buttressed trees (Fig. 3a). Significant movements in the roots occurred at small trunk displacements (about 10° from the vertical) and the trunks rotated about a point close to the leeward side of the trunk, at the greater depth of around 50 cm (Fig. 3b). The leeward roots, with their low bending rigidity bent and buckled rather than pushing far into the ground whilst the windward laterals either snapped at their base or remained attached to the trunk and were uprooted. The behaviour of the tap roots also differed from that of the buttressed trees being pushed sideways into the soil on the leeward side, leaving a cavity where they had previously been (Fig. 3b). As the trees were pulled over the tap root continued to be pushed into the ground and the leeward roots buckled further (Fig. 3c). Roots orientated at 90° to the direction of pull (growing out the 'sides' of the tree) probably contribute little to anchorage of either buttressed or non-buttressed trees because any restoring force generated will act close to the centre of rotation of the tree. Anchorage mechanics: soil shear strength Throughout the sampling period the clay soil remained at or close to field capacity. This meant that the shear 1708 Crook et al 1) a) 2) a) Fig. 2. Trunk and root movements during anchorage failure of buttressed and non-buttressed trees, (la) Buttressed tree without sinker roots (Aglaia affinis), (2a) buttressed tree with sinker roots (N ramboutan). Note: sinker roots may be present or absent on both Aglaia and Nephelium species. (la) Buttressed tree without sinker roots The tree is anchored into the ground by the thick buttressed lateral roots and the tap root. (lb) 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 laterals 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. ( l c ) As the test proceeds the leeward buttress finally fails, breaking towards end of the buttressing The centre of rotation 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 (2b), the tree rotates about a point just on the leeward side. Initially the roots firmly anchor the tree in the ground, the leeward laterals 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. a) b) c) \ Crevice • C.OB Fig. 3. Trunk and root movements during anchorage failure of non-buttressed trees (M. wrayi). (a) Non-buttressed tree. The tree is anchored into the ground by the tap root and to a lesser extent the lateral roots, (b) 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 c.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 (c) 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. Anchorage mechanics of buttressed trees 1709 trees failed in their roots rather than by the trunk breaking. In contrast to the buttressed trees, noticeable movement of the roots began at the onset of the tests and the maximum anchorage moment was reached at 27 ±9°, Anchorage tests: buttressed trees later than for the buttressed trees; by this time the tree All the trees that were pulled over failed in the root had already begun to uproot (Fig. 4b). After the windsystem rather than by the trunk breaking. Despite the ward roots had come up, the anchorage resistance began large number of different buttressed tree species used in to fall only slowly because the resistance of the tap root this study the mechanism of anchorage failure of all these was still high. trees was similar to those initially observed, with the The shapes of the anchorage moment/trunk displacewindward buttressed laterals either pulling out of the ment graph (Fig. 4b) for all the trees was similar. The ground or delaminating and the leeward buttresses pushmaximum anchorage strength was 4.9 ±1.45 kNm. ing into the ground and breaking at their end (Fig. 2-lb). However, comparing similar sized buttressed and nonHalf of the trees tested failed by delamination along their buttressed trees, the anchorage strength of M. wrayi was windward buttress (all Aglaias) and half had their windjust over half (56 + 5.7%) that of buttressed trees. ward buttresses uproot cleanly (one Aglaia species and The anchorage strength of the non-buttressed trees also two N. ramboutans). One of the N. ramboutans uprooted correlated with tree size with bigger trees being better even though its buttresses possessed sinker roots. Of the anchored (^ = 49.3%, P<0.0\, n = \A). 12 trees originally tested, four of them were too well anchored to be pulled over with the hand winch. These Damage to the root systems: buttressed trees trees were retested after their buttresses had been cut away (see trees without laterals, below). Without exception, all the leeward buttresses were damThe mechanism of anchorage failure was also similar aged during the anchorage tests, with their roots breaking in trees that had asymmetric rooting patterns, with their 30-50 cm out from the tree, towards the end of their buttresses on the leeward or windward side only, as were buttressing. In contrast, fine roots excepted, the windward the shape of the anchorage moment/displacement curves. buttresses without sinkers pulled out of the ground whilst However, the trees with tap roots and windward butthose that had sinkers delaminated. In 2 of the 8 trees tresses rotated closer to the trunk, whilst the trees with which pulled over the tap roots broke cleanly at their leeward buttresses and tap roots only rotated about the base, leaving the root embedded in the ground. leeward hinge and earlier during testing. Two trees, one Aglaia and one N. ramboutan lacked tap roots, but the Non-buttressed trees shape of the anchorage strength/displacement curve was similar in shape to the other trees, probably because of Unlike the leeward roots of the buttressed trees only a their symmetrically distributed buttresses. small proportion of the leeward laterals of non-buttressed trees were damaged noticeably during the tests. Only four The maximum anchorage strength of the 8 trees which (10%) of the leeward roots broke; these roots were the were successfully winched over was 10.6 + 3.19 kNm at a largest and leastflexibleand the damage was always close trunk displacement of 18 + 4.8°. The shapes of the anchorto the join with the trunk. The majority (90%) of the age moment/trunk displacement graph (Fig. 4a) for all leeward laterals, being thinner and moreflexible,buckled the trees was also similar, with the anchorage strength instead of breaking, and appeared to be undamaged. The initially rising rapidly and then falling rapidly as the roots windward roots either snapped at their base (one-fifth of uproot or break. Figure 4a shows the relationship between roots) or pulled out of the ground. Only three of the 14 anchorage strength and trunk displacement for individual trees. The anchorage strength of the tree correlated tap roots broke during uprooting, and where damage did with tree size with bigger trees being better anchored occur it was always close to the base. strength of the soil did not differ significantly with time or between locations. The average strength (and sd) was 62±18kPa. Relative anchorage strength and components Non-buttressed trees Unlike the buttressed trees all 14 M. wrayi had an evenly distributed lateral root system with roots emerging from all around the trunk. The mechanism of anchorage failure of all the non-buttressed M. wrayi trees was the same as that initially observed, with the windward laterals breaking or uprooting, the leeward laterals buckling and the tap root pushing into the soil on the leeward side. All To highlight the differences between buttressed and nonbuttressed trees in the relationship between their anchorage moment and the angle of inclination (Fig. 4) the relative anchorage moment of each tree at each inclination were calculated by dividing the raw figure by the anchorage moment at 18%, where buttressed trees usually show the greatest moment. Although the buttressed trees belonged to different species, this procedure shows that 1710 Crook et al. b) Nonbuttressed a) Buttressed 15 £ z E o 10 en o A luxonensis A elliptica A affinis A affinis A affinis 0 10 20 30 40 50 60 Trunk displacement from the vertical (°) 0 10 20 30 40 50 60 Trunk displacement from the vertical (") Fig. 4. The anchorage moment generated as trees are pulled over for (a) buttressed trees (Aglaia species solid lines, N. ramboutan dashed lines, n = 8) and (b) non-buttressed trees (all M. wrayi, n= 14) grouping them together is appropriate because the shapes of the original curves were similar (Fig. 5). Trees without laterals All trees whose laterals had been removed failed in a similar manner, with the tap root pushing into the ground on the leeward side, just as in intact Mallotus. Figure 6 shows the relationship between the anchorage strength and the displacement of the trunk. The shapes of the curves of both tree types were also similar in shape to the intact M. wrayi, with the anchorage strength plateauing rather than falling after the maximum anchorage strength had been reached (Fig. 4b), and the anchorage strength of the two types was similar. Paired Mests revealed that for both tree types the anchorage strength was significantly different from the anchorage strength of trees with an intact root system (/> = 0.04, n = 3 for the buttressed trees, P = 0.015, n = 5 for non-buttressed trees). However, the maximum anchorage strength of the cut buttressed trees was only 32 ± 12% of intact trees; in contrast, cut non-buttressed trees had 75 ± 12% of the anchorage strength of intact trees. These results suggest that tap roots make a far more important contribution to anchorage in the non-buttressed than the buttressed trees. anchorage moment and the strains generated along the trunk and roots of both the buttressed and non-buttressed trees. However, assessing the strain curves of all trees showed that general patterns of strain behaviour did occur. Figure 7 shows the strain responses of individual trees that best highlight these responses. Buttressed trees For buttressed trees the strains on both sides of the trunk rose and fell with the moment applied by the winch; when the overturning force was at its peak so was the strain along the trunk. As the tree came out of the ground, and the anchorage moment was reduced, so were the strains. The maximum strains on the trunk differed considerably between the trees, ranging from —1200 to over — 2100 ^Strain on the compression side and between 2300-2820 ^Strain on the windward, tension, side. At the same applied moment, strain 1 m up the trunk was negatively correlated with trunk diameter (^ = 43%, P<0.0\). The strains up the trunk were also higher than those about the base of the tree. This indicated that the force flow is concentrated along the buttresses rather than down the trunk (Figs 7-lb; 8). The strains along the windward buttress roots were highest where the lateral joined on to the trunk Anchorage tests: strain measurements along the trunk and (Fig. 7-lb) falling only slightly further along the root. roots When the roots were pulled out of the ground the strains fell because the roots no longer anchored the tree and Because of large variation in root morphologies and had only to support their own weight. orientation there was no 'typical' response between the Anchorage mechanics of buttressed trees a) Buttressed 1711 b) Nonbuttressed 10 c E o 0.8 CD o 0.6 o c .> 0.4 en 0.2 • 0.0 10 20 40 50 60 10 20 30 40 50 60 Trunk displacement from the vertical (°) Fig. 5. The mean relative anchorage moment of (a) buttressed (n = 8) and (b) non-buttressed trees (n= 14). Bars represent ± S E of the mean. Filled symbols represent trees with an entire root system and open symbols, the component tests, where the trees had their laterals cut away, leaving only the tap root to anchor the tree (n = 3 for buttressed trees and n = 5 for non-buttressed trees). Values are calculated relative to the anchorage moment at 18° of each individual tree, except the relative anchorage strength of the component test trees which were calculated relative to the anchorage strength of trees of the same species of similar size. Hence, the shapes of curves and contribution of the tap root can be compared between buttressed and non-buttressed trees, but not their anchorage strength. The pattern of strain development in the leeward buttress was similar, with the strains along the trunk and join following the shape of the anchorage strength curve (Fig. 7-la, b). However, unlike in the windward buttressed laterals, the strain further out along the root continued to rise after the anchorage moment began to drop; this corresponded to the centre of rotation moving further to the leeward (Fig. 2-1 c) and the catastrophic bending at the ends of the leeward buttresses. Therefore, at small trunk displacements the strains generated at the join and along the root were similar on both the leeward and the windward sides; however, at maximum anchorage strength, strains on the leeward roots were up to seven times greater than those on the windward roots. Non-buttressed trees Figure 7-2a shows the anchorage moment produced by the root system of an individual non-buttressed M. wrayi tree when pulled over, together with the strains generated along a windward root and a leeward root (Fig. 7-2b). Just as in the buttressed trees the strains on both sides of the trunk were proportional to the anchorage moment of the tree (Fig. 7). Again the maximum strains on the trunk differed considerably between the trees, ranging from —1500 to over — 3760/^.Strain on the compression side and between 1760-2900 ^Strain on the windward side. Unlike the buttressed trees, however, the strains about the bases of the trees (both on the tension and compression sides) were higher than they were up the trunk (Fig. 8). Figure 8 compares the ratio between the strains 1 m up the trunk and the base of the trunk of both buttressed trees and non-buttressed trees. Strains at the base of the buttressed trees were lower than up the trunk whilst in the non-buttressed M. wrayi the strains were greater at the base than up the trunk, showing force flow is concentrated down the tap root rather than the laterals. The strains along the windward laterals were highest where the lateral joined on to the trunk (Fig. 7-2b) and fell along the root, just as in the buttressed trees. However, the difference was far more marked, the strain at the join reaching over 30 000 ^Strain (and off the scale of the strain amplifier). The pattern of strain development in the leeward laterals was also similar to those of buttressed 1712 Crook et al. b) Nonbuttressed a) Buttressed 15 10 o o u c 5 - 0 A affinis A odoratisTma A affinis A affinis A crassinervia 10 20 30 40 50 60 Trunk displacement from the vertical (°) 0 10 20 30 40 50 60 Trunk displacement from the vertical (°) Fig. 6. The anchorage moment generated as the tree is pulled over after the lateral roots have been cut away (component tests), (a) buttressed trees (Aglaia species solid lines, N. ramboutan dashed lines, n = 6) and (b) non-buttressed trees (all M. wrayi, n = 5). trees and to the windward roots (Fig. 7-2b), with the strain rising throughout the tests. Statistical analysis of anchorage strength and root attributes Stepwise multiple regression was used to determine which roots contributed most to anchorage strength of the buttressed and non-buttressed trees. The analysis correlated the total areas of windward, side, lee laterals and tap roots 10 cm along their length with anchorage strength. In the case of the buttressed trees none of these components were correlated with the anchorage strength of the tree. This was partly because of the small sample size and large variation in root orientation; both the leeward and windward roots are likely to contribute substantially to anchorage and the analysis could not separate these out. To overcome this difficulty a further multiple regression examined the correlation between the area of the tap root and the total area of all buttress roots and anchorage strength and revealed that of the two components only the area of the buttress roots were significantly correlated with anchorage strength (^ = 0.46 P<,0.05, n = 8) and the regression equation is: Anchorage strength (kNm) = 7.0 + 297 x area of the buttress roots (m2) The constant is high (P<,0.0\) indicating that the buttresses are not the only important factor and suggesting that the tap root contributes to anchorage after all. Whilst this analysis was of limited use for determining importance of root type and orientation on the anchorage strength of buttressed trees, the technique proved to be far more successful in the case of the non-buttressed species: only the area of the tap root (P<,0.00l) and the area of the windward laterals (P<,0.05) were significantly correlated with anchorage strength (^ = 0.82): Anchorage strength (kNm)= —0.11 +319 x area of the tap root (m2) + 742 x area of the windward sinkers (m2) Therefore, using average values for the areas of the tap root and the windward laterals (calculated from Table 1), their relative contribution to anchorage around 72% and 28%, respectively, comparable to the value obtained from the anchorage component tests. Discussion Although the study was limited by the number of suitable buttressed trees available it was still possible to identify how buttressing contributes to tree stability. Most of the methods described worked well, though one criticism of the initial trenching tests could be that, because the root system was damaged and less confined by the soil, the root movements may have differed from untrenched trees during uprooting. However, no noticeable differences were observed between them and intact (i.e. unexcavated) trees. Attaching strain gauges proved a useful addition because they can reveal how the roots are behaving during overturning without first needing to damage to the root system. Anchorage mechanics of buttressed trees 10 r- 1713 2a) H 1 1 h BODO over 30000 >tStrain 4000 • 1b) 2b) 2000 / ^ ^ ^ ^ 0 ..•• •• a. • *"•. .-••••• » • • • * *• .».•••* • " • - • • • • -2000 • » •.. . • • • • • " • ' -4000 A'. -6000 c "o -B0O0 V) -10000 * '. '. '. 4 •' -12000 -14000 -18000 K 1 m up the trunk Base of tree "Join" 0.5 m along root -18000 -20000 ! • • • » . ' -A. , -22000 20 40 60 20 40 60 Trunk displacement from the vertical (°) Fig. 7. (1, 2a) the anchorage moment generated by individual trees, both buttressed and non-buttressed, as they are pulled over and (1, 2b) the associated strain generated along the trunk and the lateral roots. All the strains are negative on the leeward side (in compression, dotted lines) and positive on the windward side (in tension, solid lines), ( l a ) shows the anchorage moment generated by two buttressed Aglaia affinis trees. The dashed line shows the anchorage strength of a tree that had strain gauges positioned along its leeward buttress only and the solid line a buttressed tree that had gauges positioned along a windward buttress only and (2a) the anchorage moment generated by a non-buttressed M. wrayi tree, with gauges positioned along both windward and leeward laterals simultaneously. 1714 Crook et al. 1.8 D 3 1.6 1 " 4 Z. 1-2 3 E 1.0 c o O.B "K 0.6 o .* 0.4 t? 0.2 0.0 Fig. 8. The strain at the base of the trunk relative to the strain 1 m up the trunk (on the windward side with an applied overturning moment of 2.2 kNm) for buttressed and non-buttressed trees. Different letters indicate significant differences between the groups (analysis of variance with Tukey multiple comparison test). The study revealed two unexpected findings about root morphology. The first was that the majority of the buttress roots did not have sinkers to anchor them firmly into the ground and second was that the buttressed trees did have tap roots. These findings are in disagreement with Mattheck's model of tree anchorage (Mattheck, 1991, 1993). Mattheck also suggested that the buttresses strengthen the anchorage by preventing delamination at the join of the windward roots and trunk. In these studies windward buttresses that did possess sinker roots often remained securely in the ground and did delaminate rather than uprooting. Presumably, without the buttresses this would have occurred at much lower forces. Despite these surprises the anchorage mechanics of buttressed trees does appear to be similar to the model proposed by Mattheck; the leeward buttresses resisting being pushed into the ground and the well anchored windward roots with sinkers also resist uprooting. Although the Aglaia and Nephelium trees without sinker roots along their windward buttresses uprooted, the trees that did possess sinkers on their windward buttresses were more securely anchored, fitting Mattheck's model better. In contrast to the windward buttresses which require sinker roots to be effective, the leeward buttresses resist being pushed into the soil whether they have sinker roots or not. This finding, therefore, is inconsistent with Henwood (1973), who believed that buttresses were primarily tension members; buttresses without sinker roots will be most effective as compression members. It must be remembered, however, that in order to be most efficient at anchoring the tree, the buttresses must withstand both uprooting and being pushed into the soil because of the dynamic loading of the wind. One thing that was clear was the effectiveness of buttresses at increasing anchorage strength and efficiency. The buttressed Aglaia and Nephelium species were almost twice as well anchored as similar sized non-buttressed M. wrayi. This was clearly due to the contribution the buttressed laterals make to the anchorage strength of the tree; and component tests showed that the buttresses were responsible for generating around 60% of the total resistance to uprooting compared with only 40% for the tap roots (Fig. 6). In contrast, in non-buttressed trees, though the tap root contributed a similar anchorage moment as tap roots of buttressed trees, its thin laterals contributed much less; only around one-sixth that of buttresses. Unlike the non-buttressed trees that reached their maximum anchorage strength around 25° trunk lean from the vertical, buttressed trees reached their maximum anchorage strength earlier, at around 18° trunk displacement (Figs 4, 6). This meant that the greatest resistance of buttressed trees to overturning was produced before anchorage failure began. In contrast, anchorage failure had already begun in the non-buttressed M. wrayi before the maximum anchorage strength was reached, reducing its efficiency. There was a trend for larger trees to possess a greater proportion of sinker roots along their buttresses. Although this was not statistically significant, late development of sinkers is not inconceivable because as a tree grows taller and the anchorage requirements increase, the tree may no longer be adequately anchored by the tap root and leeward buttresses alone. Developing sinkers along the laterals would increase the anchorage strength of the tree by increasing the anchorage efficiency of the windward buttresses. In taller tree species that develop extensive buttresses, like many of the Dipterocarpaceae for example (Wood and Meijer, 1964), one might expect all the buttresses to possess sinkers. Extensive excavation of many tree species of varying sizes is needed to test this hypothesis. So how does buttressing confer increased anchorage strength? One main factor is that whereas both leeward and windward laterals of buttressed trees contribute to anchorage strength it is only the windward lateral roots of non-buttressed trees that contributed to anchorage. This is because whereas even thin roots are strong in tension and resist uprooting, laterals need to be thick to resist bending forces (Gordon, 1978). The laterals of M. wrayi are thin and flexible and will generate only a small restoring moment when they are bent. Even the windward roots contributed little and this is probably because of their thinness which makes them flexible in bending and poorly orientated to resist upward movement of the root plate (Coutts, 1983a, 1986). In contrast, the Anchorage mechanics of buttressed trees buttressed laterals of the Aglaia and Nephelium species are highly resistant to bending not only because the roots are thick but much more because their height is almost eight times the width of the root (Table 1); this geometry makes the roots more rigid in bending, because their second moment of area (/), is high, just like as in the 7 ' beams used in civil engineering (Rigg and Harrar, 1931; Gordon, 1978; Crook and Ennos, 1996). In fact, using values of height and width of the roots from Table 1, it can be calculated that the second moment of area of the laterals of buttressed trees is over 20 times greater than that of the laterals of the non-buttressed trees. Assuming that the material stiffness of the two woods is the same this means that the bending rigidity of buttressed roots is also over 20 times greater than the non-buttressed roots of M. wrayi. Even this, however, is likely to be an underestimate because bending stiffness increases with wood density and the wood of the buttressed species is denser than that of M. wrayi (Burgess, 1966). In conclusion, therefore, this study has shown that buttresses do have a clear anchorage function. Both leeward and windward buttresses are effective and act in compression and tension, respectively (provided that the windward buttresses are mature enough to have sinker roots), and together produce more anchorage than the tap root. As a result the buttressed trees examined had almost twice the anchorage of similar sized trees without buttresses. So why do some trees have buttresses while others do not? Clearly, size is important, since larger trees tend to be more frequently buttressed than smaller ones. This might be because the anchorage efficiency of tap roots falls as trees get bigger and so larger trees require additional support (Ennos, 1993). A further study performed on the scaling of anchorage of Mallotus will help to test this idea (in preparation). Wood density may also be an important factor. The buttressed trees used in this study had wood of higher density than the M. wrayi (Burgess, 1966). Trees of denser wood may have thinner trunks and tap roots which will be less effective at anchoring the tree and may require additional support. Clearly, a survey correlating tree morphology and wood density of a wide range of species would be instructive to test this theory. The factors that mediate the growth of buttresses is also unknown. It is possible that differences in trees may result purely from their initial pattern of rooting which is related to factors involving water and nutrient uptake (Scott-Russell, 1977). The Mallotus wrayi trees studied had a larger number of thinner roots compared with the buttressed trees (Table 1). According to the Constant Stress Hypothesis (Metzger, 1893; Mattheck, 1991, 1993) trees grow adaptively by laying down wood faster in highly stressed regions. It is possible that the roots of M. wrayi will be less mechanically stressed during growth 1715 than those of the buttressed trees and so will remain thin. In contrast, the thicker roots of the buttressed trees will be more highly stressed particularly at their junction with the trunk and buttresses will automatically be produced (Mattheck, 1991, 1993). This possibility is also being tested by instrumenting young trees, of contrasting morphology, with strain gauges around their stem base and laterals and investigating the differences in strains along the root as the trees develop in natural breezes. Acknowledgements We are indebted to the Malaysian Economic Planning Unit and the Danum Valley Management Committee of Yayasan Sabah Forestry Upstream Division (RBJ) for permission to work at the excellent Danum Valley Field Centre. We thank Ismail Mohammed Bin Samant for invaluable and superb field assistance, Leopold Madani (of the Forest Research Centre) for identification of the trees, our internal collaborators Robert Ong and Anuar Mohammed (also of the FRC, Sepilok, Sandakan), the Royal Society and the NERC for funding (Grant Number GR3/9446). This paper is based on material collected (in part) whilst the authors were participants in the Royal Society's South East Asian Rain Forest Research Programme (program publication No.A/141). References Baker HG. 1973. A structural model of the forces in buttressed tropical rain forest trees (Appendix to Henwood, 1973). Biotropica 5, 89-93. Bailie IC, Mamlt JD. 1983. Observations on rooting in mixed dipterocarp forest, central Sarawak. The Malaysian Forester 46, 369-74. Black HL, Harper KT. 1979. The adaptive value of buttresses to tropical trees: additional hypothesis. Biotropica 11, 240. Boom BM, Mori SA. 1982. Falsification of two hypotheses on liana exclusion from tropical trees possessing buttresses and smooth bark. Bulletin of the Torrey Botanical Club 4, 447-50. Burgess PF. 1966. Timbers of Sabah. Sabah: Forest Department. Corner EJ. 1988. Wayside trees of Malaya, 3rd edn. Malaysia: The Malaysian Nature Society. Courts MP. 1983a. Root architecture and tree stability. Plant and Soil 71, 171-88. Coutts MP. 19836. Development of the structural root system of Sitka Spruce. Forestry 56, 1-16. Coutts MP. 1986. Components of tree stability in Sitka Spruce on peaty gley soil. Forestry 59, 173-97. Crook MJ. 1994. The mechanics of lodging in wheat. PhD thesis, University of Manchester. Crook MJ, Ennos AR. 1993. The mechanics of root lodging in winter wheat, Triticum aestivum L. Journal of Experimental Biology 44, 1219-24. Crook MJ, Eonos AR. 1996. The anchorage mechanics of deep rooted larch, Lanx europea x L japonica. Journal of Experimental Biology 47, 1509-17. Ennos AR. 1990. The anchorage of leek seedlings: the effect of root length and soil strength. Annals of Botany 65, 409-16. Ennos AR. 1993. The scaling of root anchorage. Journal of Theoretical Biology 161, 61-75. Ennos AR, Crook MJ, Grimshaw C. 1993. A comparative study 1716 Crook et al. of the anchorage systems of Himalayan Balsam Impatiens glandulifera and mature Sunflower Helianthus annuus. Journal ofExperimental Botany 44, 133—46. Fayle DCF. 1975. Extension and longitudinal growth during the development of red pine root systems Canadian Journal of Forestry Research 5, 109-21. Francis WD. 1924. The development of buttresses in Queensland trees. Proceedings of the Royal Society of Queensland 36, 21-37. Gordon JE. 1978. Structures or why things don't fall down. London: Penguin Books. Henderson R, Ford ED, Renshaw E, Deans JD. 1983. Morphology of the structural root system of Sitka spruce. I. Analysis and quantitative description. Forestry 56, 121-35. Henwood K. 1973. A structural model of forces in buttressed tropical rain forest trees. Biotropica 5, 83-9. Jenik J. 1978. Roots and root systems in tropical trees: morphological and ecological aspects. In Tropical trees as living systems. Tomlinson PB, Zimmerman MH, eds. Cambridge University Press, 323—49. Jordan CF. 1982. The nutrient balance of an Amazonian rainforest. Ecology 63, 647-54. Lewis AR. 1988. Buttress arrangement in Pterocarpus officinalis (Fabaceae): effects of crown asymmetry and wind. Biotropica 20, 280-5. Mardia KV. 1972. Statistics of directional data. London: Academic Press. Mattbeck C. 1991. Trees: the mechanical design, Springer Verlag. Mattheck C. 1993. Design in der Natur. Der Baum als Lehrmeister. Freiburg: Rombach Verlag. Metzger K. 1893. Der Wind als massgeblicher Faktor fur das Wachstum der Baume. Mundener Forsthche Hefte 3, 35-86. Navez AE. 1930. On the distribution of tabular roots in Ceiba (Bombacaceae). Proceedings of the National Academy of Sciences (USA) 16, 339-44. Ng FS. 1989. Tree flora of Malaysia, Vol. 4. London: Longman. Petch T. 1930. Buttress roots. Annals of the Royal Botanic Gardens, Peradeniya 11, 277-85. Richards PW. 1952. The tropical rain forest. Cambridge University Press. Richter W. 1984. A structural approach to the function of buttresses of Quaranbea asterolepis. Ecology 65, 1429-35. Rigg GB, Harrar ES. 1931. The root systems of trees growing in sphagnum. American Journal of Botany 58, 391-7. Senn G. 1923. Uber die Ursachen der Brettwurzelbilding bei der Pyramiden-Pappel. Verhandlungen des Naturforschenden Gesellschaft in Basel 35, 405-35. Scott Russell R. 1977. Plant root systems: their function and interactions with the soil. MaidenHead: McGraw-Hill Book Company (UK ) Limited. Smith AP. 1972. Buttressing of tropical trees: a descriptive model and new hypotheis. American Naturalist 106, 32—46. Spoor G, Godwin RJ. 1979. Soil deformation and shear strength characteristics of some clay soils at different moisture contents. Journal of Soil Science 30, 483-98. Stark NM, Jordan CF. 1978. Nutrient retention by the root mat of an Amazonian rain forest. Ecology 59, 434-7. Warren SD, Black HL, Eastmond DA, Wtaaley WH. 1988. Structural function of buttresses of Tachigalia versicolor. Ecology 62, 532-6. Whitmore TC. 1972. Tree flora of Malaysia. Vol.2. London: Longman. Wood GH, Meijer W. 1964. Dipterocarps of Sabah (North Borneo). Kuala Lumpur: Art Printing Works.
© Copyright 2025 Paperzz