Chapter Five
DIVERSITY
Most tropical forests are very rich in species and a single hectare of lowland forest may contain 100-200 species of trees.1 The total number of plant species, including smaller trees, shrubs, herbs, climbers and epiphytes is obviously much higher, and a recent study found 233 species in a plot of just 10 m x 10 m in Costa Rica. This is equivalent to one-sixth of the British flora, for example, on an area the size of half a singles tennis court. The total of plant species in a Sulawesi forest is likely to be lower but certainly of the same order. Caution is required before enumerations of total species are embarked upon, however, because it has been calculated that to study a single hectare would take one person nearly a decade (Whitmore et al. 1986).
The only detailed species count conducted in Sulawesi was in the somewhat seasonal, and therefore rather species-poor, forest at Toraut, Bogani Nani Wartabone National Park2 (table 5.1). The total of tree species in a hectare of forest in the ever-wet centre of the island would probably be higher. It has to be remembered that the figures are derived from relatively small, and not necessarily representative areas, and that the pattern of growth phases within the area enumerated will affect the totals.
The problem of explaining the very high diversity of plants in tropical forests, particularly lowland forests, has taxed many minds. Before detailed investigations of fossil pollen began about 15 years ago it was thought that tropical forests had climatic stability for millions of years and that this stability had allowed time for evolution of so many species. It is now realised that tropical vegetation has experienced considerable changes (p. 29), but those areas that retained a wet, non-seasonal climate have the richest floras and faunas. Some people have suggested that the formation of isolated forest blocks, or refuges, during the peaks of glacial activity, when climates were cooler and drier, would have resulted in the species of each block following a course of evolution slightly different from that followed by species in the other blocks when the climate ameliorated and the forest blocks were reunited, some of the closely-related species would have been incompatible for reproduction (i.e., would have become different species-p. 30), and thus a greater degree of diversity would exist than before the forest blocks were separated. This is generally known as the 'Refuge theory' (Haffer 1982) and has been useful in explaining the high diversity in certain organisms in Amazonia and Africa. It is less convincing as a theory in Southeast Asia, however, because the whole of Peninsular Malaysia, Sumatra, Borneo and West Java would have formed one 'refuge' and the majority of New Guinea the other (Meijer 1982).
In addition to historical factors, species richness is also influenced by the wide range of potential niches available in a forest, variations in site conditions, the changes in physical conditions through the forest growth cycle (p. 359), and the vast array of potential interactions with animals in processes such as pollination and dispersal (p. 379) (Whitmore 1984).
* Figure calculated from numbers in 1.6 ha; tree numbers show a linear relationship with area, but tree species do not.
1 - Whitmore and Sidiyasa 1986; 2 - Whitten 1980; 3 - Kartawinata et al. 1981; 4 - Raemaekers et al. 1980; 5 -Bismark 1982b
Figure 5.1. Percentage occurrence of different tree species from plots in (a) Karaenta Reserve, Maros, and (b) Mopuya, Bolaang Mongondow, to illustrate the large number of rare species. Intervals represent 10% of the trees present. In both cases half of the trees are represented by three-quarters of the species.
From data in Doda 1980; Harun and Tantra 1983
An ecosystem with high diversity has, by corollary, a large number of rare species (fig. 5.1) so, instead of asking "why is lowland forest so diverse?", one could equally well ask "why are there so many rare species?" (Flenley 1980). To answer this one has to consider the different reproductive strategies adopted by living organisms. These can be described as falling within a spectrum within two extremes: r-selected strategies in which as many offspring as possible are produced with minimal parental investment in each offspring, and K-selected strategies in which few offspring are produced but each receives a great deal of care, attention and material resources in order to ensure its survival and success. Thus, elephants, cows, and bats are K-selected, and many (but not all) rats, pigs, fish and insects are r-selected. If an organism adopts the K-strategy in a lowland forest it will almost certainly be relatively rare. It is then necessary to ask "under what circumstances can rarity be an advantage?". It is generally true that the herbivores found on a plant species in lowland forest are restricted to a few species which have evolved as the plant species evolved and have found ways to circumvent whatever form of physical or chemical defence the plants had adopted (p. 370). The chances of a herbivore finding an individual of a particular tree species are greater if that tree is common, and so the pressure exerted by herbivores on an abundant, or otherwise dominant, tree species puts it at a disadvantage and permits less common species to coexist (Janzen 1970).
There have been many attempts to devise and index that would adequately describe in a single figure both the number of species in an area of vegetation (its richness) and the way individuals are distributed between the species (the evenness). A recent evaluation of the available indices found that richness was best represented simply by the number of species, and the evenness was represented with similar degrees of appropriateness by most of the techniques available. The Shannon Wiener function or
where 's' = number of species and 'Pi' = the proportion of the ith species, was the most appropriate measure of diversity for tropical rain forest3 A measure of evenness (E) could be derived from this by
E = H ÷ Hmax
where Hmax = Log2S where 'S' is the number of species, but the usefulness of calculating diversity has been questioned of late because its biological meaning is unclear (Stocker et al. 1985).
The high diversity of plant and other life forms in the tropics reflects the great structural diversity and an enormous range of potential niches. In an attempt to answer the question of how many species there are in an area of tropical rain forest, chemical fogs which cause insects to drop out of the canopy have been blown through a number of forests in America and Southeast Asia. One of the first attempts was in Panama and gave an estimate of 41,000 species of arthropods in one hectare of unspectacular seasonal forest (Erwin 1982), with more than 945 species of beetles recorded from just one species of tree (Erwin and Scott 1980). Animal diversity on Sulawesi, like diversity of plants, is high by the standard of temperate regions, but is relatively low compared with neighbouring Borneo or New Guinea, the largest island to the east (p. 39). Similar fogging techniques were used in the Toraut forest but the results are not yet available (N. Stork pers. comm.).
STRUCTURE AND COMPONENTS
Characteristics
Lowland forests are characterized by the huge amount of plant material they contain, and this can be measured in terms of the amount of carbon present. The tropical forests of the world (i.e., not just lowland forests but also montane, swamp and dry forests) cover 1,838 million hectares or 11.5% of the earth's land surface, yet they contain 46% of the living terrestrial carbon (Brown and Lugo 1982). Lowland forests in Sulawesi would certainly contain an even higher percentage of the carbon relative to their area.
It has often been stated that nearly all the inorganic nutrients in a tropical forest are held in plants and very little remains in the soil, a situation markedly different from forests in temperate regions. Detailed studies of various forests throughout the tropics have now shown that there is great variation between forests in the percentage of nutrients held in different parts of the forest ecosystem, and also that within a forest the distribution of different inorganic nutrients varies. In general, however the biomass of tropical forests does not hold a disproportionately large percentage of the nutrients (Whitmore 1984).
Lowland forests, particularly in the wetter regions, are characterized by the conspicuous presence of thick climbers, large buttressed trees, and the prevalence of trees with tall, smooth-barked trunks. The vast majority of trees have simple mesophyll-sized leaves (p. 498) between 8 cm and 24 cm long. The trees of the lower canopy, including the immature individuals of large tree species, often have large leaves (Parkhurst and Loucks 1972). Lower canopy saplings commonly have pronounced drip-tips (fig. 5.2). Since one theory of the functions of drip-tips is that they allow water to run easily off the leaf surface and thus prevent or hinder the growth of epiphylls (p. 358), it is reasonable that these should be most common nearer the forest floor where the relative humidity is highest (i.e., rate of evaporation lowest).
There is considerable similarity between plant families in the shape of their leaves and this, together with the high diversity of species, and the poor knowledge of the Sulawesi flora, frequently make identification, even to the genus level, somewhat difficult. Characters such as sap, bark type, buttress size, leaf-vein arrangement and arrangement of leaves on twigs are those of most use in the identification of the majority of the specimens but since plant taxonomy is based on the structure and arrangement of flower parts, final identification relies on the examination of flowers (p. 385).
Layering
The crowns of forest trees are usually said to form three 'layers'. While this is a convenient theoretical concept it has limited application in the field. An emergent layer is often quite clear but below that the supposed layers are often difficult to distinguish. Some trees may be referable to a particular layer but many more are generally not. Part of the difficulty in applying the concept of layering lies in the existence of the mosaic of pioneer-, building- and mature-phases in the forest (p. 359), each of which is in a dynamic flux. The strata are clearest in forests where there is a single dominant species (Richards 1983).
Figure 5.2. Examples of drip-tips on leaves of trees in Sulawesi lowland forest, a - Cratoxylon celebicum (Hype.); b - Ficus variegata (Mora.); c - Pterospermum celebicum (Ster.); d - Vitex quinata (Verb).
Borrowing terms from limnology, two clear zones have been described. The canopy, comprising the crowns of the emergent trees together with the crowns of the main tall trees, has been called the euphotic zone (p. 314), for it is in this zone that most of the photosynthesis and production of the forest occurs. The lower level of this zone undulates between the lowest boughs of the larger trees (fig. 5.3). As in a lake, animals are most abundant in this zone which is unfortunately the furthest from a ground-bound observer (p. 416). Below the euphotic zone, in the oligophotic zone, most of the light is reflected or transmitted from above, although flecks of sunlight are found. As a result, productivity is much lower and decomposition processes predominate (Richards 1983). For the purposes of this book, the lower canopy is taken to mean the crowns of trees in this oligophotic zone.
Physical factors other than light vary through the canopy, providing a complex range of microclimates (table 5.2).
The crowns of neighbouring trees of similar heights are rarely in contact and are separated by spaces called 'crown shyness gaps'. These appear to by caused by abrasion of buds and shoots as the trees sway in the wind. These spaces are relevant because they affect movements of arboreal animals and the ability of climbers to reach a tree crown from a neighbouring tree (Putz and Parker 1984).
Figure 5.3. Layers and zones through a lowland forest.
After Moore 1986
• = lowest |
••••• = highest |
After Lock 1980
Basal Area and Biomass
The basal area of trees in a given area is a measure of the 'amount of tree' present. The cross-sectional area (πr2) is calculated for each tree and the total basal area per hectare and the average basal area per tree per hectare are frequently calculated. In most lowland forests the total basal area of trees with diameters of 15 cm or more at breast height is between 25-50 m2/ha (table 5.3).
Biomass can be estimated from the volume of the trees (estimated as height x basal area x 0.5) multiplied by the specific gravity which is generally about 0.6 t/m.3 There is probably considerable error for individual trees but this becomes negligible if a large number of trees are included in the sample (Whitmore 1984). The biomass of lowland forest wood averages about 380 t/ha, generally ranging between 200 and 500 t/ha (Brown and Lugo 1984). Direct measurement of above-ground biomass in two plots of lowland forest in Peninsular Malaysia resulted in totals of 475 and 664 t/ha and an estimated average of 500-550 t/ha (Kato et al. 1978).
Roots
Roots have to compete with soil organisms such as fungi, bacteria, and numerous invertebrates for the organic carbon and the other nutrients in the forest soil. It appears that almost all trees in lowland forests and almost all non-grain crops can develop a mutualistic symbiotic relationship between their roots and fungi (Janos 1980). This association is called 'mycorrhiza'. The most common form found in the tropics has the hyphae (root strands) of the fungus penetrating the plant's roots where they take up carbohydrates from the plant. For its part the fungus channels minerals into the roots (Janos 1980, 1983). Many plant species with mycorrhizae appear not to possess root hairs through which water and minerals generally pass, because these structures are unnecessary given the efficiency of the mycorrhizae. Mycorrhizae are probably poorly developed among plants of early successional stages which normally grow on soil enriched by fire ash, deposits left by flood water, or the decomposition products of fallen trees (Janzen 1975). In such situations a plant would be at a disadvantage if it had to wait for its roots to be infected with fungus before it could start to grow (Janos 1980, 1983). Conversely, some species cannot grow without mycorrhizae even at the seedling stage, and yet others grow better without mycorrhizae where mineral nutrients are plentiful, but better with mycorrhizae in poor soils (Janos 1980, 1983). For trees that do depend on mycorrhizae it is advantageous to have large seeds, because they provide a reserve of food upon which the seedling can draw before infection with the fungus occurs. It is not surprising then, that seedlings of these species kept experimentally without mycorrhizae, stop or slow their growth after attaining a size correlated with the average dry weight of a seed for that species (Janos 1980).
* After Clayton in prep.; other data from EoS teams
Buttresses and Trunks
Buttresses are lateral extensions to the lower part of a tree trunk and are both common and varied on tall tropical trees. Different species are relatively constant in the presence, shape and surface characteristics of buttresses and these characters can be helpful in identification of trees.
Since the decision of whether or not a tree is considered to be buttressed depends on the definition of a buttress, it is difficult to compare sets of data collected by different workers on the percentage of buttressed trees in different lowland forests. Some data were collected, however, by EoS teams in different locations in North and Central Sulawesi (table 5.4) and up to three-quarters of the trees of 15 cm diameter at breast height were found to be buttressed.
Buttresses are sometimes said to act as structural supports for trees whose roots are relatively shallow and where the substrate affords little anchorage (Smith 1972; Henwood 1973). It has been suggested that buttressing increases a tree's resistance to mechanical stress, and reduces the tensile forces on the roots. There should, therefore, be a smaller probability of the trunk of a buttressed tree snapping close to the ground. This hypothesis has not been found to be valid in a Panamanian forest, where there was no statistical difference between buttressed and non-buttressed trees in terms of the number of snapped or uprooted individuals. Further, mortality rates were no different between buttressed and non-buttressed trees (Putz et al. 1983).
Buttresses are also said to be composed of 'tension wood' to reduce the pulling strain on the roots, and so buttresses would be expected to form on the side of the trunk which experiences the greatest tension. Observations by EoS teams along a number of transects indicated that many trees do not, in fact, form buttresses at tension sites such as on the uphill side of a trunk, on the side of a trunk opposite a major congregation of heavy climbers, or on the side from which the wind blows.
Recent hypotheses on the reasons for buttress formation have been unrelated to structural problems. It has been suggested, for example, that since buttresses increase the space of forest floor occupied by an individual tree, they could be viewed as a competitive mechanism (Black and Harper 1979). According to this hypothesis the physical presence of a buttress prevents or hinders the establishment of neighbouring (i.e., competing) trees close by. If this hypothesis is correct then it should be possible to observe that:
• the distances to the nearest neighbours of buttressed trees are greater on average that from a non-buttressed tree to its nearest neighbours;
• the density of trees near a buttressed tree is less than near a tree of similar bole size, canopy form and age without buttresses;
• the density of trees around different individuals of a buttress-forming species bears a relationship to the age of the individual trees. Older buttressed trees should have both more distant and fewer neighbours in the smaller size classes;
• species diversity of trees around a buttressed tree is lower than around a non-buttressed tree because some species are likely to be less competitive against buttresses than others (Black and Harper 1979).
Data were collected in the Toraut forest from 17 mature buttressed trees and six similar trees with no buttresses. It was found that the density of trees around buttressed trees was not significantly different from the density around non-buttressed trees, but the difference in mean distance of trees from buttressed trees was significantly greater than from non-buttressed trees (table 5.5). These results suggest that of the first two predictions, the first is valid but the second is not.
Small trees sometimes establish themselves on the outer fringe of the buttress arms, thus supporting the notion that the presence of a buttress is a physical deterrent to the establishment of other trees too close to the trunk. Often, however, small trees can be found in the area between the buttress arms. It appears that these sites may indeed favour the establishment of seedlings of shade-tolerant trees. It may be that the large buttress arms serve to intercept wind-dispersed seeds, or that rats and squirrels bring seeds there to store or, in the case of rats, to take to their burrows which are often located near the base of trees. For whatever reason, the between-arms areas of buttresses are favourable micro-habitats for shade-tolerant tree species, due to low wind velocities, and increased water and nutrients from stemflow and litter fall. The virtual absence of a large tree growing between the buttresses of an older tree suggests that any saplings growing in such places eventually die, presumably as a result of competition for nutrients and water.
Living trees with rotten, empty trunks are surprisingly common in lowland forest. This may appear to be unfortunate for the tree but it has been suggested that it is in fact a strategy bringing distinct advantages. Empty cores of large trees are inhabited by bats and rats in addition to a host of invertebrates. Faeces and other products from these animals decompose and thus provide an exclusive supply of nutrients for the tree (Fisher 1976; Janzen 1976, 1981).
Climbing and Creeping Plants
Climbing plants abound in many types of lowland forest. In lowland forest at Toraut and at Lombogo,4 44% and 37% respectively of the trees with trunk diameters of 15 cm or more at breast height, carried climbers. Of these trees about 10% had heavy vine loads. These figures are similar to those obtained from an area of lowland forest in Panama (Putz 1984b). The crown of a large climber may be as large as that of a tree, but because they do not possess any commercial value (with the exception of rattan palms), they have been little studied. Despite the value of rattans, however, the Sulawesi species are scarcely known. About 50 different kinds and qualities have been recognized by traders (van der Koppel 1928), but it is not known to what extent these refer to actual species. Given the uncontrolled and intense collection of rattans, even in reserves and national parks, and the increase in demand, a good case can be made for a period of intensive collecting and research into sustainable harvesting and domestication.
Data from EoS team; G. O'Donovan pers. comm.
Identifying climbers is difficult but the coiled or convoluted ones with somewhat flattened stems are generally leguminous species, and the ones with regular hoops around the stem are one of the three species of the pink-fruited Gnetum5 (Gnet.), best known to botanists for possessing some of the characteristics of flowering plants and some of the characteristics of conifers. Gnetum has been described as either an ancient type of plant or a freak of nature (Corner 1952).
When young, many species of free-hanging climbers can look very similar to young trees as they grow slowly through the undergrowth and lower canopy, waiting for an opportunity to rise to the canopy. In the Panama forest 22% of the upright plants less than 2 m high were in fact young climbers (Putz 1984b). Opportunities to rise to the upper canopy are afforded when a gap is formed and it is then that a young climber can attach itself to a rapidly growing tree and be carried upwards. Other species, during a similar period of apparent dormancy, are growing a large tuber below ground. In response to a certain cue, the energy available in the tuber is suddenly used up in a rapid burst of upward growth, whether or not a gap has formed (Janzen 1975).
A considerable proportion of the apparent 'seedlings' of climbers (and a certain proportion of some forest trees) are in fact no more than shoots growing up from a horizontal root of an established plant (Janzen 1975). In the Panama forest mentioned above, 15% -90% of the young climbers less than 2 m tall were offshoots from an underground stem rather than seedlings (Putz 1984b). This is a means by which a plant can increase the effective size of its crown, and since these shoots are genetically identical to their 'parent', the actual number of genetically distinct individuals in a given area may be extremely low.
Climbers compete with trees for light, nutrients and water and can cause mechanical damage to trees. In addition, bole diameter is often greater than average for trees with heavy climbers in their crown, indicating that trees have to divert resources away from vertical growth and reproduction in order to increase the strength of the supporting structure. Trees carrying climbers probably suffer higher mortality and cause more neighbouring trees to fall than trees that are free of climbers. The falls do not usually kill the climbers, however, for they quickly rise to the upper canopy again using the abundant supports available at a treefall site (Putz 1984a). Thus the stems of climbers can sometimes be seen to loop down towards the forest floor and back again. Even without having their growth disturbed by treefall, mature climbers are rarely supported by a single tree but rather grow horizontally through the canopy, sewing together the tree crowns. In the Panama forest, climbers connected an average of 1.6 trees (Putz 1984b) and this tying together of crowns may help to prevent trees from falling over in strong winds. If a tree is felled on purpose, however, the climbers in its crown often pull over other trees as well. For many reasons it would clearly be an advantage to a tree to shed or to avoid climbers.
In an attempt to elucidate whether some tree species do actively shed climbers, 20 trees each of 24 species in Panama were examined and their climber load was recorded. Growth rate of the trees and height of crown appeared to have little to do with load of climbers, but there were significant positive correlations between the length of tree leaves and flexibility of tree stems with absence of climbers. Flexible trees with large leaves are abundant in disturbed, well-lit habitats and these characteristics may help them to remain climber-free by shedding the climbers along with the leaves and by breaking the brittle voting growing points of climbers while swaying in the wind (Putz 1984a). In addition, some trees exhibiting a symbiotic relationship with ants (such as Macaranga) may benefit from the ants removing vines from their branches as they have been shown to do in Central America (Janzen 1973).
Creeping plants are those whose roots are attached firmly to tree trunks as well as in the ground. They are generally shade-tolerant, and flower and fruit below the upper canopy. Between 55% -65% of trees, at least 15 cm in diameter at breast height at Toraut and Lombogo were found to have creepers and 13%-16% of these trees had considerable numbers. In Sulawesi two of the most common creepers are Freycinetia (Pand.), a small pandan with rough leaf edges and many-seeded, club-like fruits held within brightly-coloured bracts, and the superficially similar Pothos (Arac.) which has smooth-edged, two-lobed leaves and bright red fruit. Trees can have one, both or neither of these plants.6
A sample of 234 trees with trunk diameters of at least 15 cm at breast height in Bogani Nani Wartabone National Park were scored for whether they had no, one, few or many climbers and creepers. Of these, 33% bore neither, 57% bore creepers, 38% bore climbers, and 21% bore both. It would be expected that creepers could be more common since their progress is not hampered by any of the trees' possible defences against climbers. Trees without creepers are frequently those with flaking bark such as Eucalyptus deglupla (Myrt.), Tristania (Myrt.), Agathis dammara (Arau.), and Dillenia celelnca (Dill.). The trees sampled had an average diameter of 30 cm but those with abundant climbers and those with abundant creepers were, not surprisingly, large and very similar in average diameter (40-42 cm). This is presumably an effect of age and, as stated above, the diversion of resources from vertical growth and reproduction to increase the strength of the tree's supporting structure.
Epiphytes and Epiphylls
An epiphyte is a perennial plant rooted upon, not in, a large host, and which does not have to produce or maintain massive woody stems and branches to live above the forest floor. Epiphytes are common in lowland forest, but even more common in certain montane forests (p. 507), and they are thought to make a significant contribution to the total biomass and species-richness in a forest (Benzing 1983; Whitmore et al. 1986). The major higher plant families of epiphytes include Gesneriaceae, Melastomaceae, Rubiaceae, Asclepiadaceae and Orcidaceae. In addition to these there are numerous smaller epiphytic ferns, lichens and bryophytes (including mosses and liverworts). The small size of bryophytes makes them less restricted than other plants in the ranges of microhabitats they can occupy (fig. 5.4). The most luxuriant bryophyte communities in wet lowland forest are generally found on the bases of large trees (Pócs 1982).
Epiphytes might seem to have an easy life but they live on a substrate which is extremely poor in nutrients. Most of them rely upon nutrients dissolved in rain, in litter fall, and in occasional mineral inputs from animals. Epiphytes also have to contend with a very erratic water supply— being drenched when it rains, and parched when the sun is shining and the wind is blowing.
Epiphytes cope with growing in these nutrient-poor conditions in various ways (although few species will exhibit all these features):
• their own tissues may have very low nutrient concentrations;
• juvenile stages are prolonged even though little growth is required to become mature;
• their vegetative parts are reduced in size;
• they may obtain nutrients from unconventional sources such as ants (p. 463);
• their leaves tend to be long-lasting (due perhaps to some form of defence against herbivores) so that the replacement of nutrient-expensive leaves is required only infrequently (Janzen 1975) (p. 370) ; and
• their flowers are pollinated by animals because wind dispersal of the pollen would result in 'expensive' losses (Benzing 1983). In addition, it has been observed that some epiphytes concentrate their roots on the underside of boughs where water and nutrients would remain longest (Whitten 1981).
Some epiphytes, notably the superficially-similar Alyrmecodia (Rubi.) and Hydnophytum (Rubi.) (p. 463), harbour ants in their stems and these may well protect their leaves from caterpillars and other arthropod herbivores. To test this hypothesis would not be difficult if ants on one set of plants were killed with insecticide and herbivore damage then monitored on that set and on a control set of plants.
Figure 5.4. Bryophyte microhabitats in a lowland forest. 1 - bases of large trees, 2 - upper parts of trunks, 3 - macro-epiphyte 'nests', 4 - bark of main branches, 5 - terminal twigs and leaves, 6 - bark of climbers, shrub branches and thin trunks, 7 - pandan stems, 8 - tree fern stems, 9 - palm trunks, 10 -rotting trunks, 11 - soil surface and termite mounds, 12 - roadside cuttings, 13 - rocks and stones, 14 - submerged or emergent rocks in streams.
After Pócs 1982
Although epiphytes (by definition) do not take nutrients directly from the host plant, they have nonetheless been called 'nutritional pirates' (Benzing 1981, 1983). This is because they can tie up nutrients in their own biomass from the dust, leaf leachates and rain which would otherwise have fallen through to the soil to be utilized by the trees that support them. While this is true, host trees can turn the tables at least partially back in their favour by producing small roots beneath the mats of dead and living epiphyte tissue (Nadkarni 1981). These roots thus give the tree access to sources of nutrients in the canopy. The main boughs of some tree species are almost invariably covered with epiphytes but it would appear that physical or chemical features of the bark of other species actively deter epiphyte growth, presumably because the relationship is disadvantageous.
Special forms of epiphytes are strangling figs. These look like large trees with latticed trunks when mature, but they started life high in the forest canopy as a seed deposited in the faeces of a bird, bat or civet. The seed germinates and grows like an epiphyte, but roots grow down to the ground. These roots and the leafy part of the plant grow in size and an increasing number of roots grow down and around the trunk. With time these roots fuse with each other, completely enclosing the tree trunk. The host tree eventually dies, probably as much by competition for light in the canopy as by strangulation.
Epiphylls are those mosses, liverworts, algae and lichens that grow on the living surface of leaves in shady situations where the air is almost continually saturated, conditions also found in moss forest (p. 500). Thus they are generally restricted to the leaves of young trees, shrubs and long-lived herbs of the lowest layer of forest up to about 2-3 m above the ground, as well as near streams. Over ten species of epiphylls can occur on a single leaf and some of these are obligate; that is, they are found only on leaves (Pocs 1982). Epiphylls are restricted to certain types of leaves and are possibly more common on leaves lacking drip-tips (p. 500).
DYNAMICS
Succession and the Growth Cycle
Different types of tropical rain forests, and indeed forests throughout the world, have many fundamental similarities because the processes of forest succession (the changes over time in the occurrence of species in a given area), and the range of ecological strategies of tree species are more or less the same (Whitmore 1984). These strategies form what is known as the forest growth cycle: the events of a large mature tree falling over in a closed forest thereby forming a gap,7 the gap filling with a succession of plant species until the final, large, mature tree falls over again. Many species require gaps for their growth and others are stimulated by them. The size of a gap depends on its cause; a single dead tree falling over forms a smaller gap than a fire, landslip or drought, and the biological effects of a gap are related to its structure and size. Many of the important life history features of a particular plant species are related to the strategies it employs to exploit different types of gaps (Pickett 1983). For example, vigorous, light-tolerant (shade-intolerant) species grow up in the gap and these create favourable conditions for seeds of shade-tolerant (but not necessarily light-intolerant) tree species to germinate, for the seedlings to grow and for these to eventually supersede the species which initially filled the gap. The formation of mature forest from a gap, is known as 'secondary succession'.
Three basic ecological principles can be formulated from the study of both primary8 and secondary succession:
• succession proceeds in only one direction, with fast-growing, tolerant colonizers replaced by slower-growing species with more specific requirements, and great competitive ability where those requirements are met;
• as new species colonize an area, they will inevitably alter the environment by their presence;
• the new conditions are generally less suitable for their own seedlings but more suitable for those of other species, hence the succession continues.
The forest growth cycle can be divided into three phases—the gap-, building- and mature-phases—which together form a mosaic throughout the forest that is continually changing in state and shape (Whitmore 1984). For example, a patch of forest in the building-phase may return to the pioneer phase should a tree in a neighbouring mature patch fall across it. The floristic composition of a forest will depend on the size of former gaps because a large gap will be filled initially with tree species requiring light whereas small gaps will be filled by shade-tolerant tree species. Many of the latter may have grown slowly for many years to become part of the lower canopy but have been unable to form part of the main canopy until a small gap was formed. It should be noted that since a single species of tree can contribute to pioneer-, building-, and mature-phases, it is incorrect to refer to 'pioneer-phase' or any other '-phase' species (Whitmore 1982). Terms describing the characteristics of a species, such as pioneer, shade-tolerant or light-intolerant, are more useful. In addition, because a forest comprises a mosiac of gap, building and mature areas, it is a matter of debate whether the term 'climax' vegetation has any meaning in the field.
The regeneration of trees in a one-hectare botanical plot at Toraut seems to be in gaps caused by single tree falls. The peculiarly small crowns, absence of a middle layer and an apparent low density of climbers sewing together adjacent crowns in that forest, results in very neat tree falls causing little damage to surrounding trees. The pattern of the different phases is thus very fine-textured and in some places trees had fallen without really forming a gap at all (fig. 5.5) (Whitmore and Sidiyasa 1986).
Small gaps created by dead standing trees initiate short cycles without a gap phase. That is, immature trees beneath the dead tree are able to grow up as soon as the light intensity increases. Large gaps created by the fall of several adjacent large trees allow the full forest growth cycle to occur. The size of gaps clearly increases with the maturity of the forest (larger trees, more crowns interconnected by climbers), thus there are more pioneers in an area of generally mature forest than in the same area of forest in the building-phase because of the larger gaps (Brokaw 1982b).
A sample of 310 fallen trees was examined in Panama and it was found that 70% had snapped at some height above the ground, 25% had been uprooted, and 5% had broken off at ground level. Presence or absence of buttresses appeared to have no connection with the manner in which the trees fell (p. 351). Compared with trees that had snapped, uprooted trees were generally larger and shorter for a given bole diameter, and had denser, stiffer and stronger wood. The trees with less dense wood grew faster and compensated for their tendency to snap by frequently sprouting from their base, particularly those trees in the smaller size classes (Putz et al. 1983). Records of treefalls in that forest showed that most occurred in the middle of the wet season, possibly due to stronger winds or increased weight due to water in the bark and on the leaves, and this may affect germination and/or dispersal of plants that need light gaps in which to grow (Brokaw 1982b).
Determining the duration of the growth cycle in temperate regions is greatly assisted by being able to age the trees simply by counting the rings in the wood that form because of differential growth in different seasons. In the tropics, where the rate of wood growth is more or less constant through the year, other more roundabout methods have to be used to examine the dynamics of forests through time. One such study was conducted in East Kalimantan and it was found that it took 60-70 years after the formation of a large gap for the number of growth-phase species to reach a maximum, and as long again for mature-phase species to dominate and for gap formation to begin again. Biomass continues to increase for 220-250 years after disturbance, some trees live for over 500 years, and it may take this long for a stable and dynamic system to form (fig. 5.6) (Riswan et al. 1985). Similar longevities are known from elsewhere in the tropics (Hue and Rosalina 1981; Brokaw 1982b; Lieberman et al. 1985).
Figure 5.5. Mosaic of pioneer-, building- and mature-phase forests in 1 ha of forest at Toraut, Bogani Nani Wartabone National Park.
After Whitmore and Sidiyasa 1986
Work on Sumatra has helped to elucidate variations in the length of the forest growth cycle between sites. It was found that with increasing age of river terraces (from 40-10,000 years), soil pH and fertility decrease. With decreasing soil fertility, production of leaf litter (p. 363) and fruit decreases although the forests on the poorer soils were mature phase, and the trees were longer-lived. It would appear that differences in soil fertility have caused trees to adapt different forms of nutrient economy and hence patterns of life history, and this is reflected in forest structure. The trees growing on poor soils clearly have an advantage if they retain scarce nutrients, and for those on richer soils there are fewer benefits from hoarding nutrients, although the cost of doing so remains similar. As trees hoard nutrients, so the soil quality will continue to decline and this in turn favours trees that hoard still further. This trend will favour later reproductions and greater longevity since it will benefit the tree to consolidate its position and retain its competitive advantage before diverting resources to the production of flowers and fruit. This helps to explain why large trees are so often found on ridges even though the soils there are generally poorer than on the slopes. Also, forests that contain tree species typical of pioneer- or building-phases, such as those on recent river terraces, are in fact merely forests that happen, by virtue of their richer soil, to have species possessing shorter cycles of regeneration than trees growing on poorer soils (van Schaik and Mirmanto 1985).
Figure 5.6. Estimates of composition, biomass, and age expectancy of trees after the creation of a large gap in lowland forest.
After Riswan et al. 198S
Within each of the different phases of the forest growth cycle, the trees have a number of ecological characteristics in common. For example, the pioneer species that grow up in a gap generally produce large numbers of small seeds frequently if not continuously. These seeds tend to be easily dispersed (often by wind, squirrels or birds) and, probably because of this, these species tend to have wide geographical distributions (Hue 1981). They have high viability even in full light, and the seedlings grow rapidly.
The building-phase is more or less built by the animals, such as pigeons, monkeys, bats and hornbills, that benefit most from it. This is not pure chance because the trees of this phase bear fruit which tend to be co-evolved with a particular disperser either in size, nutrient content, or phenology. For example, different individuals of Dracontomelum dao (fig. 5.7) fruit at different times such that the monkeys which disperse them have fruit available to them all year round. Both tree and frugivore thereby benefit by this arrangement (Anon. 1980). Many species of trees common in the building-phase are unable either to colonize open sites or to regenerate beneath their living parents (Whitmore 1982).
The forest on the young volcanic soils of Tangkoko-Batuangus Reserve, Minahasa, is largely in the building-phase of development and this, combined with the relatively rich soils and high productivity, is reflected in the composition of the fauna which comprises large numbers of relatively few species and high populations of frugivores. For example, the density of crested macaques Macaca nigra is about 300/km2 and of red-knobbed hornbills Rhyticeros cassidix about 60/km2. This is the highest-known density of macaques in any Indonesia study area (Anon. 1980).
Trees found in mature-phase forest are shade-bearers (that is, they tolerate growing in shade, but if more light is available they will grow faster), and thus can regenerate beneath a closed canopy. They tend to be found in patches but throughout an area (Lack and Kevan 1984). These species often have large seeds with rather specialized means of dispersal and a relatively restricted geographical distribution.
Changes in species composition through the growth cycle may be due to availability of nutrients, availability of seeds, chemical suppression of some species by the same or other species (allelopathy), crown or root competition, and chance, but there is still very little information available on how regeneration actually occurs. There are three ways in which trees can grow: from shoots growing out of roots, stumps or fallen trunks; from established seedlings; or from seeds. Beneath a mature forest there are clearly no seedlings of light-demanding species and, conversely, in gaps, light-intolerant species will die. Seeds can be brought into an area in the 'seed rain' (seeds dispersed from their parents) or may lie dormant in or on the soil surface. These dormant seeds are known as 'seed banks' and their numbers are quite surprising. Topsoil from four forest types in northern Queensland harboured 588-1,068 viable seeds/m2 which is similar to other tropical lowland sites examined. The most seeds were found in coarse-grained soils, which could more easily retain a wide range of seed sizes. Very few trees characteristic of the mature phase had viable seed in the soil except those that were actually fruiting at the time of sampling (Hopkins and Graham 1983). These trees tend to fruit infrequently, have poor powers of dispersal and clearly little or no potential for extended dormancy. Their continued existence in an area therefore depends on fruit from mature trees falling into a spot with the correct conditions for immediate germination. Some seeds of pioneer species found 20 cm below the soil surface were still viable, although they must have arrived many years before.
Figure 5.7. Dracontomelum dao whose fruit is sought by macaques and frugivorous birds. Scale bar indicates 1 cm.
After Soewanda n.d.b
Litterfall, Nutrient Cycling and Productivity
The first stage of nutrient cycling is the fall of leaves and other plant material from the trees to the forest floor. From various studies it would seem that the rate at which fine litter9 falls is quite similar between tropical areas at about 7-14 t/ha/yr, of which leaves constitute about 60% -80% (Lim 1978; Ogawa 1978; Kartawinata et al. 1981; Spain 1984; van Schaik and Mirmato 1985). Five litterfall traps in the Toraut forest were monitored by an EoS team for varying periods (subject to disturbance by pigs) and extrapolation of the data indicated an annual litterfall of 12.7 t/ha/yr, about 80% of which was leaves. The methods used in different studies vary-somewhat, however, and detailed comparisons are difficult. Standardized methods are proposed elsewhere (Proctor 1983).
Calculating the rate at which the coarse litter10 falls is much harder since it occurs irregularly and unevenly. Very few studies have attempted to determine the quantity of coarse litter falling but estimates are of the order of 9-15 t/ha/yr (Yoneda et al. 1978; Brasel and Sinclair 1983).
Most decomposition in lowland forest usually occurs in the soil or on the soil surface, although a relatively large percentage of trees and branches start decaying while still standing or attached to the main trunk respectively (Edwards 1977; Kira 1978). The litter reaching the forest floor has been assumed by many to decompose much faster in the tropics than in temperate regions, although some field data suggest otherwise. The rate of decomposition is determined, at least in part, by the quality of the leaves themselves and the composition of the litter-feeding macro-fauna on and in the forest soil (Anderson and Swift 1983). Decomposition rates of tree leaves appear to depend on leaf thickness (and consequently toughness), and weight loss over six months can vary from 20% to 60% (Takeda et al. 1984). Some leaves lose only 72% of their initial weight after 14 months (Gong and Ong 1983).
The time required for leaves to decay, and thus for the minerals to be released to enter the cycling pathways, varies between species depending on the nature of the leaves. A recent study in Puerto Rico on decomposition rates of six species (three genera of which are indigenous to Sulawesi), showed that the leaves of the three species of mature-phase forest retained only 20%-50% of their original organic matter after 16 weeks and 10%-20% after 32 weeks. The three pioneer species decayed more slowly. The difference was related most closely to structural characteristics such as percentage lignin and percentage of fibre. Given that the more open habitats where pioneer species grow have potentially greater rates of nutrient loss because of the greater quantity of rain falling harder on generally more compacted soil, leaf structures which bind elements for longer periods would represent a valuable means by which trees could ensure a more sustained release of nutrients (La Caro and Rudd 1985).
Some of the nutrient cycling occurs rapidly, such as when nutrients in leaves are absorbed into a tree which grows new leaves every year, before the leaves are shed. Nutrients in leaf litter and rain may be quickly taken up by roots or they may be adsorbed into the soil particles and released slowly. Nutrients in the trunk of a tree, however, may be held for hundreds of years and even when the tree dies, a decade may pass before all the nutrients are released.
The major external input of inorganic nutrients into a forest above the flood level is from rain. These nutrients have probably been released into the atmosphere from fires. Up to 30% of the rain is intercepted by leaves and evaporates or soaks into the bark. A small amount (<1%) flows down tree trunks and the rest reaches the forest floor (Arsyad 1985). The percentage reaching the forest floor also depends on rainfall intensity, being greater when rainfall is heavier (Edwards 1982), and contains, not just nutrients from fires, but also from bark, canopy liter and leaves with which it has been in contact. As a result, the rain reaching the forest floor is significantly richer in nutrients than that reaching the soil in an open field nearby. The forest probably does not have the ideal closed or leakproof cycling of nutrients attributed to it because some nutrients are lost in the flow of water to the rivers. Potassium does seem, however, to be cycled tightly within the forest.
Very few studies of total forest primary production (p. 132) have been conducted anywhere in Southeast Asia, largely because of the many methodological difficulties involved (Whitmore 1984). At a site in Peninsular Malaysia three methods were used (Kato et al. 1978; Kira 1978; Koyama 1978; Yoda 1978a, b). The results suggested a net primary production of 30 t/ha/yr and a gross primary production11 of 80 t/ha/yr (Whitmore 1984). The net primary productivity is lower than for rubber and oil palm plantations (Kato et al. 1978), but a mature forest would not be expected to have as high a productivity as a managed, forest plantation which is harvested before the rate of production increase begins to fall.
Flower, Fruit and Leaf Production
No studies of the cycles of flower, fruit and leaf production in Sulawesi forests have been published although the fruiting of trees was monitored in Tangkoko-Batuangus Reserve for nearly two years (Anon. 1980). Studies of canopy trees elsewhere in the tropics generally find wide variation between, and even within, species in the patterns of flower, fruit and leaf production. Some species flower at consistent but non-annual intervals, some flower more or less annually, whilst others flower at irregular intervals in excess of one year and irrespective of weather. Weather can, however, influence whether the flowers will form fruit, such that a tree may flower and yet bear no ripe fruit. Abortion of flowers and young fruit appears to be due to a lack of necessary resources such as inorganic nutrients, water and products of photosynthesis (Stephenson 1981). Patterns of leaf production are similarly variable.
Figure 5.8. Monthly percentages of trees with unripe fruit, ripe fruit, with flowers, and with new leaves or shoots in an area of lowland forest in Peninsular Malaysia.
After Medway 1972
Despite these variations, the tree community as a whole does generally show relatively clear patterns of flower, fruit and leaf production (fig. 5.8). In areas with two wet and two dry seasons each year, two flushes of new leaves seem to be produced with a major peak just after the driest time of year and a minor peak just before, and extending into, the wettest time of year (Medway 1972). Flowering generally peaks after the driest time of year and fruiting, not surprisingly, peaks after that, just before the wettest time of year. The flowering of many lowland forest trees seems to be initiated by waterstress (Whitmore 1984). It should be stressed however that this is the trend in a forest and individuals may differ significantly from the above pattern.
Monitoring of cycles of flowering, fruiting and leaf production in lower canopy shrubs and small trees (<10 m tall) in Peninsular Malaysia and Costa Rica did not reveal such clear patterns as found in upper canopy species (Opler et al. 1980; Wong 1983), possibly because the microclimate near the forest floor is less seasonal that in the canopy.
The main distinguishing feature of forests in western Indonesia is the predominance of trees of the Dipterocarpaceae. Although these are more or less absent in Sulawesi (p. 386), their effects on forest ecosystems are relevant. A remarkable characteristic of the dipterocarps is that they all tend to fruit together at long but irregular intervals of five to seven years, at which times there is often exceptional flowering and fruiting activity in other tree families too (Appanah 1985). The primary advantage of this is probably similar to that of new leaf flushes in canopy trees, simultaneous spawning of coral (p. 221), or gregarious nesting habits of turtles (p. 155); that is, to satiate the appetites of predators so that at least some leaves, larvae or baby turtles escape early death. It avoids specialization of predators on those organisms and keeps the population of such predators down to relatively low levels.
It has been suggested that the large numbers of dipterocarps and the infrequency with which they produce fruit may be a factor in the relatively low biomass and diversity of animals in the forests of western Indonesia and Malaysia (Janzen 1974). In Sulawesi, the place of dipterocarps is taken by trees that do not appear to fruit gregariously but the biomass and diversity of animals is not noticeably greater than to the west, with local exceptions (p. 363), which suggests that the generally low production of fruit (when compared to the America tropics), possibly as a result of poor soils, may be the major factor in this discussion. Long-term studies (at least two years) of flower, fruit and leaf production in everwet areas of Sulawesi would help to reveal any differences with forests in Peninsular Malaysia, Kalimantan and Sumatra and would assist in understanding the role of dipterocarp trees in those other forests.
Many of the trees in lowland forest are pollinated by insects, and it is not surprising that peaks in abundance of certain insect species coincide with peaks of flower production. These peaks occur just after the drier times of the year and usually coincide with, or come slightly after, peaks of leaf production which are exploited by butterfly and moth caterpillars (Coomans de Ruiter 1951). Other insect species, particularly those associated with rotting wood, or those dependent on pools of water for breeding, become most abundant during the wetter months (McClure 1966; Hails 1982). Dramatic increases in insect numbers are occasionally observed, perhaps in response to climate and availability of food. For example, in June 1982, an expedition driving up the logging road on Mt. Roroka Timbu near Lore Lindu National Park encountered an enormous cloud of butterflies between 400 m and 1,000 m altitude. The cloud was so dense that it was necessary to use windscreen wipers in order to see the road (K. D. Bishop pers. comm.).
These variations in insect abundance are probably reflected in the behaviour of insectivorous or partially-insectivorous birds12 which, in Sarawak and Peninsular Malaysia at least, breed and moult all year round but experience major peaks in these activities, coincident with peaks of food availability (Fogden 1972; Wells 1974). Similar patterns have been found for insectivorous bats (Gould 1978).
In East Kalimantan, birds such as large hornbills and pigeons which eat large-seeded fruits with flesh rich in lipids,13 and seed-predating parrots, left the study area when their foods became scarce. Frugivores with fixed home ranges, such as macaques, respond to low availability of fruit by decreasing the distance they travel daily, increasing the proportion of non-fruit items such as insects, spiders, and small vertebrates such as frogs and geckos in their diet, and also by exploiting the few tree species that fruit outside the period during which most trees are fruiting. These trees, now known as 'pivotal species', and those that stagger their fruiting over a period, are extremely important in maintaining sedentary populations of arboreal frugivores (Leighton and Leighton 1983; Howe 1984).
Studies elsewhere in the tropics have shown that the peak of pregnancy in forest rats coincides with the seasonal peak in fruiting (Harrison 1955; Dieterlen 1982) (fig. 5.9) and there is no reason to believe that this would not also apply to Sulawesi rats that include fruit as a major item in their diet.
Herbivory
About 50% of a leaf consists of cellulose, a complex carbohydrate molecule which makes up the outer cell walls. Most animals lack the necessary enzymes to break down the cellulose into easily digestible molecules, but some use certain bacteria, protozoa or fungi (p. 401) to conduct this first stage of digestion on their behalf. The possession of these micro-organisms gives the animal opportunities to exploit an abundant food source. In mammals, the bacteria themselves form an important source of protein for the host and can also synthesize most vitamins except A and B.
There are two major forms of bacteria-assisted digestion: foregut and hindgut fermentation in vertebrates. Foregut fermentation is found in, for example, anoa, deer, and leaf-monkeys, and hindgut fermentation is found in, for example, rabbits, horses, and rodents (Bauchop 1978; Muul and Lim 1978) and probably cuscus. Leaves are eaten by a wide range of mammals and larval and adult insects, but not by amphibians and by very few reptiles (p. 301) or birds (Morton 1978; Rand 1978). It is estimated that 7%-12.5% of leaf production in forests is eaten by insects but only 2.4% by vertebrates (Leigh 1975; Wint 1983). These figures are usually calculated. however, from the remains of fallen leaves and so do not necessarily include leaves or shoots in their entirety (Janzen 1978a).
Figure 5.9. Relationship between rainfall (columns) and reproductive activity as a percentage of 13 rodent species (upper line) and percentage of trees in fruit (lower line) in a lowland forest in Zaire.
After Dieterlen 1982
It is generally considered that ferns suffer little from grazing by insects but this view is not supported by firm data. It appears, in fact, that they suffer as much damage as flowering plants (Hendrix and Marquis 1983), and one insect specializing on eating certain Sulawesi ferns is the primitive wingless grasshopper Karnydia (R. Butlin pers. comm.).
In ecological studies of herbivory, the actual amount of leaf material eaten is not the most important measurement because each part of a plant is different in its value to the plant, 'cost' for replacement, food value and concentration of defence compounds. For example, the loss of a small shoot from a seedling is considerably more costly to the individual plant than the loss of a great many leaves on a mature plant. Leaves pay back their cost of production by photosynthesizing and contributing the products of photosynthesis to the plant. So, once a leaf has paid back its 'debt', its value to the plant decreases with time. Thus, it is clear that for advanced analysis of an ecosystem, a list of plant species eaten by a particular animal species is not nearly so useful as knowledge of what parts of each species are eaten and the likely cost to the individual plants of the material lost.
There is probably no plant whose defences preclude attack by all herbivores, and there is certainly no animal that can eat all types of leaves. A herbivore gains less usable food per mouthful when eating a leaf rich in defence compounds than when eating a leaf low in defence compounds. An animal which eats defended leaves probably has more plant species to choose from and experiences relatively less competition from herbivores unable to cope with the chemicals.
To the eyes of a human observer the leaves of a lowland forest, or indeed any other tropical vegetation type, vary in shape and shade of green. To an animal dependent on leaves as a food source, however, they are 'coloured' "nicotine, tannin, lectin, strychnine, cannabinol, sterculic acid, cannavanine, lignin, etc., and every bite contains a horrible mix of these" (Janzen 1981). These phenolic and alkaloid compounds (Walker 1975) are just some of the toxic and digestion-inhibiting chemicals used by plants to defend their leaves against herbivores (Edwards and Wratten 1980). To balance this, however, it must be stressed that not all food plants are chosen by herbivores on the basis of their chemistry alone; other important factors include the effects of carnivores and microclimates (Janzen 1985b).
Leaves of forest trees in Costa Rica were analysed for phenolics, fibre and alkaloids at three stages of growth: when the leaves were young but fully expanded; middle-aged (two months later); and mature (six months later). Concentrations of phenolics and fibre per gram of dry leaf remained more or less constant with growth, but water content decreased considerably. Alkaloids were much more common in leaves of deciduous species and trees appeared to defend their leaves with either alkaloids or tannins rather than both, probably because of some interaction which reduces their respective effectiveness. The distribution of two major groups of leaf-eating moth caterpillars on different trees was most striking and reflected some of these differences; large emperor moths (Saturniidae) preferred leaves with phenolics and no alkaloids in the upper canopy, and hawk moths (Sphingidae) preferred leaves with low levels of phenolics which are likely to contain alkaloids in a variety of habitats (fig. 5.10) (Janzen and Waterman 1984; Janzen 1985b).
The concentration of defence compounds may also vary between different parts of a single tree. Evidence is beginning to be gathered which shows that a tree is not really a single individual but rather something akin to a colonial organism. The tip of each growing shoot contains within it a group of undifferentiated dividing cells, called the meristem, which is capable of forming both germ cells for sexual reproduction and normal tissue cells. A large branching tree might have 10,000 to 100,000 meristems and even with conservative mutation rates at least some of the branches will be genetically distinct from the rest. Thus, if one branch contains a novel and effective cocktail of defence compounds its leaves may last longer than its neighbours' and its flowers may have a better chance of setting fruit with the chance that the genes for that defence will be passed to the next generation (Cherfas 1985).
Figure 5.10. Typical adults and larvae of emperor (saturniid) moths (a) and hawk (sphingid) moths (b), major consumers of leaves in lowland forest, each of which has a different feeding strategy.
The bear cuscus Ailurops ursinus (p. 429) is probably very selective in the plants it eats, choosing particular species, ages ofleaves, parts (stem, tip or entire blade), and quantities. Studies of food selection by leaf-monkeys have shown that leaves selected as food are consistently lower in lignin (fibre), and usually higher in protein, than leaves that are not selected. Tannin levels do not show consistent significant correlations with food selection (Oates et al. 1977, 1980; Bennett 1984), and it is now generally felt that the role of tannins in food choice has been overemphasized (Waterman 1983). They do, however, probably remain a potent force in ecosystems where poor soil conditions produce some form of stress in the trees growing on them, such as on ultrabasic soils (p. 466).
Tannins and lignins are classed as digestion-inhibiting phenolic compounds (although lignin is 10-20 times less effective than tannins), whereas others, such as the glucosides of Antiaris toxicaria, are classed as toxins. The adjective 'toxic' is really a measure of the energy expended by an animal in utilizing a food relative to the energy value of usable materials gained from the digestion process. Toxicity is therefore an outcome rather than an inherent property of a chemical, although some chemicals are more likely to be toxic than others. Whether a certain percentage composition of a potentially dangerous chemical is toxic or not will depend on the food value of the material eaten and on the efficiency of the animal's detoxification system. Thus, a leaf with a certain weight and concentration of defence compounds may be more toxic than a seed of the same weight and the same concentration of defence compounds but with higher nutritive value (Freeland and Janzen 1974; Janzen 1978a; McKey 1978). Defence compounds should not be thought of as purely disadvantageous from the herbivore's point of view because they can have decidedly beneficial effects. It appears that certain large animals, at least, seek leaves or bark for the stimulatory or anti-diarrhoea effects they possess (Janzen 1978a).
Pioneer plants (pp. 359-360) are generally thought to be low in defence compounds; their growth strategy is rather to grow and reproduce quickly because, in the nature of succession, they are limited in both space and time. As a result, resources are not diverted into possibly unnecessary defence.
A study at Toraut looked at plant characteristics and leaf damage by insects in three stages of succession: a field of herbs cleared of forest three years previously with few if any pioneer trees, a forest edge with pioneer trees, and a shaded lower canopy of mature forest (table 5.6).
Preliminary results show, as have these from Costa Rica (Coley 1983a, b), that both nitrogen (in protein) and water content of mature leaves decrease through the succession sequence, while leaf toughness (determined by the concentration of lignin) increases (fig. 5.11).
Various authors have suggested that the amount a plant 'invests' in leaf defence is a function of its ability to produce new leaves. Thus a plant growing in low light or on poor soils is less able to produce new leaves and will therefore invest more heavily in defences for its leaves that a plant growing on richer soils (Janzen 1974; Coley 1983a, b). This ability, measured at Toraut as the rate of leaf production over the six-week study period, clearly decreases through the successional stages and reflects the 'quality' of the habitats. The likelihood of these leaves being discovered by a herbivore depends in part on the life span of the leaves, with long-lived leaves risking discovery more than short-lived ones. To examine this, 10 leaves of each of the 26 species studied at Toraut were tagged and after monitoring over four months it was found that life spans were as follows:
After Greenwood in prep.
Figure 5.11. Changes in average (and range) of (a) nitrogen content, (b) water content and (c) toughness of mature leaves among early herbs (EH), pioneer trees (PT), and shade plants (SP), at Toraut. Toughness is measured as the force in newtons required to punch a rod 6 mm in diameter through a leaf. The shaded area in the water content histogram demonstrates the effect of removal of the fleshy-leaved aroid plants from the sample.
After S. Greenwood pers. comm.
Early herbs: 43-116 days
Pioneer trees: 64-169 days
Shade plants: all leaves of all plants retained.
The shorter-lived leaves of the herbs had fewer of their leaves discovered (even partially eaten) than those in the later stages of the succession and the area of leaf removed was also less (fig. 5.12) (S. Greenwood pers. comm.).
Figure 5.12. Rate of leaf production (number of new leaves produced in six weeks divided by original number of leaves) (a), percentage of leaves discovered by herbivorous insects (b), and percentage of leaf area removed (c) among early herbs (EH), pioneer trees (PT) and shade plants (SP) at Toraut.
Alter S. Greenwood pers. comm.
The differences in percentages of leaves discovered and leaf area removed are just what one would expect given that the leaves of shade plants have a much longer life, and it was felt important to measure the actual rate of damage rather than just its occurrence, to determine whether this decreased through the succession stages. There were great differences in the rate of damage both between and within species, but it did indeed decrease for the mature leaves, although the rate remained at about 4% of leaf area for the young leaves over the six-week period. The difference between the rates for mature and young leaves is significant, however, only for the shade plants. Similarly, the rate of leaf discovery by herbivores is higher for mature leaves of the field herbs than of shade plants, whereas the opposite is true for young leaves (fig. 5.13). In a study of leaf damage in pioneer and shade plants in Costa Rica, young leaves of both types were found to suffer more damage than the mature leaves, but the rates were ten times higher for the shade plants than for the pioneer species (Coley 1982).
Figure 5.13. Percentage of leaf area removed (a) and leaves discovered over a six-week period (b) among early herbs (EH), pioneer trees (PT) and shade plants (SP) at Toraut.
After S. Greenwood pers. comm.
Finally the various measures of leaf damage used at Toraut were tested against both the nitrogen and water contents and the leaf toughness, but no significant relationships were found. In the Costa Rican study, however, leaf toughness, fibre content (lignin) and nitrogen content correlated significantly with damage although phenol-to-protein ratios did not (Coley 1983a, b). Clearly more work is required to understand the processes at work but it would be expected that the effects of straightforward physical and chemical factors will be complicated by, among other factors, the availability of other foods, unmeasured chemicals and microclimate. For example, some species growing in gaps have been observed to suffer less herbivore damage than the same species in shaded habitats (Maiorana 1981). Strangely, when leaves from both habitats were presented to snails14 under laboratory conditions, it was found that the plants from gaps were preferred. It has consequently been suggested that the 'defence' used by pioneer plants might in fact be the hot, unshaded, low humidity of the gap environment itself, which is less favoured by some herbivores.
Interestingly, all the pioneer trees examined at Toraut bore abundant ants. Three of the trees had extra-floral nectaries15 but on Trema orientalis (Urti.) (fig. 5.14) the ants farmed large colonies of aphids from which they collect the sticky, sweet secretion known as honeydew. Indeed, the ants do not seem to have any influence on the activities of leaf-eating insects. The insects most attracted to the extra-floral nectaries of one of the pioneer trees, Mallotus sp. (Euph.) appeared to be small jumping flea beetles (Haticinae), but these were also major consumers of the leaves, thereby making any adaptive argument somewhat redundant. Further research would probably prove to be rewarding.
Figure 5.14. Trema orientalis. Scale bar indicates 1 cm.
After Meijer 1974
In the Tangkoko-Batuangus Reserve at the northern tip of Minahasa, the squirrels Prosciurillus murinus and P. leucomus were common and fed entirely on bark and insects. EoS teams at Toraut also observed unidentified Prosciurillus eating bark from the red-coloured Pometia pinnata (Sapi.) (fig. 5.15), and the darker bark of Dracontomelum dao (Anac.) was scored by squirrel tooth marks which could easily be seen. Bark is not a rich food source but it has been recorded as a component in the diet of orangutans, deer, gibbons, elephants, rhinoceros, as well as squirrels. The physical and chemical features of bark eaten by small squirrels in Sumatra have been analysed and it was found that feeding tended to occur on large, smooth-barked trees which were relatively free of climbers, and in whose bark certain tannins were absent. The food value of barks eaten were, however, no diff erent from those not eaten (Whitten and Whitten in press).
Figure 5.15. Pometia pinnata, the reddish bark of which is eaten by small squirrels. Scale bar indicates 1 cm.
After Soewanda n.d.b
Seed Dispersal and Predation
The dispersal of a seed is not simply the event of a mature fruit being taken by an animal or being released by its parent.16 Dispersal implies the carrying of a seed by some agent to a place where the seed will, perhaps, germinate, grow and eventually reproduce. In many cases, seeds die because the places they are dropped are unsuitable for growth; that is, they are not successfully dispersed. Seeds found directly below the parent tree have probably simply dropped from the branches. Most seeds falling beneath their parent are able to germinate and do so, but few if any have much chance of survival beyond the germination stage, because they experience severe competition from others of their species, because they represent a clumped food resource for seed predators (Becker and Wong 1985), or because environmental conditions beneath the parent simply are not suitable for the developing plant. Dispersal of seeds can be envisaged, therefore, as a means of escape from seed predators (Howe and Smallwood 1982) as well as a method of colonizing widely-separated available sites. In the species-rich lowland forest where most species are rare (p. 345), a seed dispersed away from its parent will stand a relatively low chance of being adjacent to a tree of the same species. A further advantage of dispersal away from the parent is that the biological and physical environment of the location where the parent itself germinated has, in almost all cases, changed by the time that tree has matured (Janzen 1970). This is the same as saying that the environment below the parent tree is now unsuitable for its own seeds. Pioneer, building and mature forest areas (p 359) occur in more or less predictable proportions and patterns in a forest and the means of seed dispersal will aim to maximize the colonization of suitable areas with seeds.
One particular tree, Syzygium lineatum (Myrt.), was found to be common in the lowland forest of Morowali National Park. All the individuals over 1 m in height were measured and mapped in 1 ha of alluvial forest and a total of 527 were found, only 10 of which were large, mature, flowering trees. The distribution of the trees was clumped but very few seedlings were found within 5 m of the parent trees despite the fact that most of the fruit fell in that area; indeed, most immature trees were at least 10 m from their parents (fig. 5.16) (Lack 1984; Lack and Kevan 1984).
The area over which seeds come to rest on the forest floor is known as the 'seed shadow'. The seed shadow of a tree is generally most dense close to the parent tree and much less dense overall when dispersed by an animal than when dispersed by wind. Whereas a wind-generated seed shadow is quite homogeneous, a seed shadow generated by animals will usually be heterogeneous, being concentrated in dung piles, along animal paths, and in certain vegetation types. An exception may be fig trees whose fruit is fed upon by so many species of animals, and whose modes of living are so varied that a relatively homogeneous seed shadow can result. There is also likely to be variation in the seed shadows of neighbouring individuals of the same species because access for the disperser into one tree may be easier than into another, and because other attractive food sources may be close to one, but not to the other (Janzen 1970; Janzen et al. 1976). These seed shadows are not simply of academic interest but are of major importance in determining future forest structure and composition (Fleming and Heithaus 1981).
Figure 5.16. Numbers of seedlings of Syzygium lineatum at different distances from the parent tree.
After Lack and Kevan 1984
Wind-dispersed fruits are generally borne by tall trees for these are the only forest trees to be affected by wind to any extent. Wind-dispersed fruits are either very light or have wings, and in both cases their descent to the ground is delayed. Although wind does not disperse tree seeds particularly far from the parent, wind can at least be relied upon, whereas animals cannot. The fate of seeds dispersed by wind from one tree of each of nine species was followed for one year in Panama and it was clear that dispersal away from the parent tree was advantageous to all species. For eight of the nine trees, dispersal also resulted in lower mortality of seedlings, although for a variety of reasons. Pathogens or micro-organisms, such as fungi and bacteria, were the major cause of seedling mortality. Some seedlings suffered from pathogens irrespective of distance from the parent or density of surrounding seedlings, but in all cases the mortality caused by pathogens was less in light gaps largely because fungal pathogens are favoured by high humidity and low temperatures (Augspurger 1984).
The majority of forest tree seeds are dispersed by animals that eat the fruit and carry away the still-living seeds in their guts to be deposited some distance away when the animal defaecates. Birds such as pigeons, many of which are obligate fruit eaters, in New Guinea at least, appear to spend more time in a fruit tree than birds that include fruit as only part of their diet. These longer visits by birds remaining in a fruit tree between feeding bouts, probably result in many seeds being dropped beneath the parent tree since rates of gut passage are about 30-45 minutes. This behaviour might make them open to predation, but the birds tend to be quite large and cryptic in behaviour or colouration (Pratt and Stiles 1983). It is noteworthy that the backs of most pigeons in Sulawesi are green, and this is the part of the birds seen by the eagles that might prey on them.
Fruits can usually be identified as being adapted for dispersal by a particular group of animals (Payne 1980). The dispersal agent for a particular fruit depends on the nutritional requirements of the animals, the accessibility of the fruit, its shape and its size. The fruits produced by a tree vary in size and, for some species, in the number of seeds they contain. Thus different portions of the fruit crop may be dispersed by different animals over different distances (Howe and van de Kerckhove 1980; Howe and Small-wood 1982; Janzen 1982).
It is obviously to a plant's advantage to ensure that fruit are dispersed to the right type of location and numerous means are used to selectively advertise their presence and to encourage certain dispersers whilst discouraging others. These characteristics have been shaped over evolutionary time by reciprocal interactions between animals and plants but, unlike the process of pollination, few species-specific relationships exist (Wheelwright and Orians 1982; Janzen 1983b). Edible fruits are basically only seeds covered with a bait of food and this bait can vary greatly in nutritional value. Plants tend to adopt one of two strategies:
• they can produce large numbers of 'cheap' (small, sugary) fruit, most of which may be wasted or killed by the many species of obligate and facultative frugivores attracted to them, or
• they can produce smaller numbers of 'expensive' (large, lipid-rich or oily) fruit that have to be searched for by those few specialized species of frugivores which gain a balanced diet almost entirely from fruit (Howe 1980, 1981; Howe and Smallwood 1982; Howe and van de Kerckhove 1980).
An example of 'cheap' fruits is figs which grow on plants of numerous life forms, occurring as free-standing trees, epiphytes, shrubs and stranglers of other trees (p. 358). Examples of 'expensive' fruit include nutmegs (Myristicaceae) and laurels (Lauraceae).
The pollination of figs is of special interest because it directly affects seed predation. The flowers themselves are within a narrow-mouthed cup formed by the developing flower structure arching over and around itself. The hole at the mouth of a fig is partially closed by scales or modified bracts. Fig flowers are pollinated exclusively by tiny fig wasps (Agaonidae). and both figs and wasps are entirely dependent on each other for their survival since one species of fig plant will generally have just one species of pollinator wasp. Female wasps fly to the figs in which the female flowers are ready for pollinating and one or more of these wasps climb inside the fig by squeezing their way through the scales and lose their wings and antennae by so doing. Once inside, a female pushes her ovipositor down into the ovary of one of the female flowers and lays an egg (fig. 5.17). As she moves around inside the hollow fig looking for a suitable ovary, the female deposits pollen from the fig in which she hatched onto flowers in the fig into which she has crawled, and so effects pollination. Although the wasps pollinate the flowers, they are clearly also seed 'predators' and in a sample of 160 figs from four species in Costa Rica, 98% of the figs had more than 30% of their potential seeds killed by pollinating wasps (Janzen 1979).
The larvae develop and pupate and the wingless males emerge first. They search around the fig and mate with young females before the latter emerge. The males' role is not yet completed, since they then tunnel through the wall of the fig to the outside. The carbon dioxide level inside the fig is initially about 10% but when the hole is made the concentration drops to atmospheric levels (i.e., 0.03%). This change appears to stimulate the development of the male flowers, the emergence of the females, and the process of ripening. The cycle continues with the females flying off carrying pollen to a developing fig and the males dying within the fig.
A fig tree studied in Peninsular Malaysia had a crop of about 41,000 figs, some of which were ripe over a period of eight days. The consumption by bats and other nocturnal mammals was not estimated but among the diurnal animals, macaques took the most, birds took half as many, gibbons half as many as the birds, and squirrels just a quarter of this (MacKinnon and MacKinnon 1984). A more detailed study in Costa Rica monitored the fate of over 100,000 figs over five days. Birds took about 65% of the figs consumed each day (the rest were taken by bats and other mammals) and most of these were eaten by parrots which are seed predators.17 Only about 4,600 figs were eaten by seed dispersers. Taking into account the seed predation by both parrots and the pollinator wasps, there was clearly an enormous wastage and only about 6% of the enormous crop was dispersed unharmed by birds (Jordano 1983).
Figure 5.17. Female fig wasp inserting an egg through her long ovipositor into an ovary of a fig flower. Her wings and antennae were lost as she struggled into the fig.
Fruits advertise their ripeness to dispersers using colour, texture, taste, conspicuous shapes, and odour. Between ripening and being taken by a disperses however, the fruit is also exposed to seed predator animals and micro-organisms. Seed predators include both animals which consistently destroy seeds of that type, and disperser animals which eat the fruit before the seeds are mature. Thus a fruit needs to be as unattractive as possible before the seed is viable and this is commonly achieved by the presence of defence compounds whose concentrations reduce as ripening progresses (p. 532).
Some fruit do not lose all their defence compounds when they are ripe, however, and this is thought to be a means of protecting fruit from being eaten or destroyed by non-disperser animals or micro-organisms (Herrera 1982; Janzen 1983). Defence compounds can be divided into two groups: toxins which can debilitate or kill, and quantitative defences or digestion inhibitors such as resins, gums, volatile oils and phenols which deter potential seed predators or cause some mildly adverse physiological effect (such as nausea) (Maiorana 1979; Waterman and Choo 1981; Waterman 1983). A review of defended ripe fruits concluded that the species producing them were in fact at a competitive disadvantage for dispersal relative to species without defended ripe fruit because of the resources which were diverted to make the defences. If the defences are viewed as a means by which undesirable seed predators can be avoided, then extreme toxicity may be a 'last resort' (Herrera 1982). One extremely toxic seed is that of the tree Antiaris toxicaria (Urti.) which contains poisonous glucosides. The latex from the tree contains the same chemicals and is used by many peoples in the preparation of poisoned arrows and darts, but it is actually not as deadly as popularly believed (Burkill 1966). Macaques almost certainly eat the A. toxicaria fruit together with the seeds which pass through unharmed, thus demonstrating that the tree probably uses its toxins to deter the 'wrong' frugivores.
A sample of 33 seed pods of Pericopsis mooniana (Legu.)18 were examined in Lamedae Reserve south of Kolaka in Southeast Sulawesi. Over-exploitation of the species has resulted in this being one of the few remaining Pericopsis stands in Sulawesi. The pods each contained 2-3 seeds and 69 seeds were examined in the sample. Of these, however, 11% had been destroyed by small beetles as evidenced by round holes in the pot shell. The average number of holes in the five pods attacked was three (L. Clayton pers. comm.). Larvae of small beetles are common predators on seeds and they can show a high level of specificity. Of over 975 species of shrubs and trees in an area of Costa Rica, at least 100 species regularly had beetle larvae in their mature or nearly-mature seeds. Three-quarters of the 110 beetles species found (primarily from the family Bruchidae) were confined to a single species of plant; if a beetle species was found on more that one plant species, the plants were found to be closely related. Of the 100 species of plants attacked, 63 were legumes and 11 were from the family Convolvulaceae (Janzen 1981). This 'preference' for legumes might be because most plant families have had strong defences against attacks from bruchid beetles whereas a few families have not. It seems more likely, however, that bruchid beetles became legume seed predators many millions of year ago and were able to counter whatever forms of defence the plant used. As the legumes evolved and diversified so the beetles evolved to meet the changes—a case of coevolution. Since plant defences appear to be so complex, the specializations required to predate on a particular species or genus are such that a beetle species would be unlikely to be able also to become a predator on seeds in another plant family.
Destructive micro-organisms (fungi and bacteria) represent a form of competition for frugivores (Janzen 1977, 1979). To a microbe, a ripe fruit is a considerable food resource that can be digested and converted into more microbes. Having started it is obviously preferable to exploit the whole of the fruit, but to do so its competitors have to be deterred. The microbe produces antibiotic substances to prevent other micro-organisms from colonizing the fruit, as well as toxic and unpleasant-tasting substances to deter frugivores. Producing tasteless toxins alone would not be a successful strategy because an animal may still eat the fruit, kill the microbe and later die itself to the benefit of neither party. For microbes to succeed, the fruits they invade have to look, smell or taste different from untainted fruit. Some fruits are known to produce their own antibiotics against decomposer micro-organisms (Janzen 1975) but this field of study has received little attention.
COMPOSITION
Composition of Mature-phase Forest
No part of Indonesia yet has a comprehensive guide to or account of its forest trees, but the first stage, annotated checklists with line drawings of the major timber species,19 have recently been prepared by the Forest Research and Development Centre, Bogor, although the Sulawesi volume (Whitmore and Tantra in press) has not yet appeared. When it does, the way will be open for the writing of keys and detailed descriptions.
Vernacular names allow differentiation of some of the trees and represent a common and useful means of conducting a general forest survey. It should be realized, however, that vernacular or commercial names generally greatly underestimate the number of species present, although some languages have more than one name for a single species (Soewanda n.d.a, b). There is a misconception, too, that any villager employed on forest surveys from a village near the forest edge will know the names of all the trees in the forests around the village.
Numerous surveys of Sulawesi lowland forests have been conducted for forestry evaluations but these are of limited use for ecological work or for general descriptions, because they enumerate only those trees that provide profitable timber, and they concentrate on areas where timber trees are likely to be most common. Consequently, there is very little ecologically useful information concerning the composition of Sulawesi's forest arising from such studies.
Variation in the floristic composition of a given phase in the growth cycle of lowland forest can be quite considerable but, because the variation is often continuous (that is, there are no sharp boundaries, and forest of one composition changing gradually into a forest with another composition), it is very hard to study. It has been proposed that floristic composition is largely determined by chance factors, particularly at the time of fruit dispersal and seedling establishment (Poore 1968). Within a given area, however, composition is related to large-scale habitat features such as soils and topography (Ashton 1964; Franken and Roos 1981; Baillie and Ashton 1983). A detailed study of species composition and various soil and other site variables in Sarawak showed that soil texture, levels of iron and aluminum oxides, and the acidity of the soil parent material were the most important factors determining species composition (Baillie and Ashton 1983). These effects are very subtle and result more in a variation in the relative abundance of species, rather than in the presence of a species. This is partly because a particular soil type will not always exclude ill-adapted species, but adults of such species will be relatively rare and have reduced rigour.
In western Indonesia the forest trees are often dominated by members of the Dipterocarpaceae, whilst in Sulawesi this family is represented by just six species (fig. 5.18; table 5.7). Other important families in western Indonesia are the Lauraceae, Euphorbiaceae, Annonaceae, Myristicaceae, Rubiaceae and Sapotaceae. With the virtual absence of dipterocarps, Sulawesi forests are not dominated by any one family and, from what information is available, there is considerable variation between sites in the major species and families found (table 5.8). For example, the most common families at Toraut were the Lauraceae, Guttiferae and Anacar-diaceae (Whitmore and Sidiyasa 1986), whereas in the Lore Lindu area Sapotaceae and Burseraceae are prominent (Meijer 1984). Some of the variation can be interpreted. For example, the presence at the Toraut site, Bogani Nani Wartabone National Park, of Garuga floribunda (Burs.), Tetrameles nudiflora (Dati.) and Kleinhovia hospita (Ster.) is indicative of a relatively seasonal climate (Whitmore and Sidiyasa 1986). These and other common species are illustrated below (figs. 5.19-26).20
The most famous lowland trees in Sulawesi are the ebonies Diospyros (Eben.).21 The architecture of the trees is highly distinctive. The branches are horizontal but droop at their tips, and emerge in whorls at intervals up the trunk rather like those of Terminalia (Comb.) and the kapok tree Ceiba pentandra (Bomb.). Various species are present in the forest but the blackest and hardest timber is from D. celebica found mainly in the central and northern regions (fig. 5.27) (Steup 1931).
Fifty years ago in the Onggak-Dumoga area of Bolaang Mongondow, there are said to have been 50 exploitable D. celebica trees per hectare (Verhoef 1938). At this time there were about 1,000 tons of D. celebica ebony exported each year from North Sulawesi, nearly all of which was sent to Japan, and even now this very fine-grained wood is much sought after. Another such ebony-dominated area is Tanoma in Tanjung Peropa Reserve where over 50 trees/ha of D. pilosanihera of 15 cm diameter and over have been found (Bismark 1982b). Most interesting from an ecological perspective is that ebony trees were generally found in dense clumps over several hectares in which 90% of all trees present would be ebony (Steup 1935). In one area south of Poso, the only other species present was the Livistona palm (Steup 1931). This pattern is very similar to that of ironwood Eusideroxylon zwageri., of Sumatra and Kalimantan which also has very hard, resistant, and therefore very valuable timber. It may be, too, that certain ebony species are similar to E. zwageri in producing relatively large, toxic seeds more or less continuously through the year (Whitten et al. 1984). The fruits of D. celebica measure about 4 cm x 2 cm and are amongst the largest in the genus. It would be of great interest to study fruit production and seed dispersal and predation in a population of these trees, but dense stands must now be very rare.
Figure 5.18. The two endemic Sulawesi dipterocarp trees: Hopea celebica (left) and Vatica flavovirens (right).
After Soewanda n.d.a
After Ashton 1984
After Steup 1933; Soeriaatmadja 1977; Anon. 1980; Bismark 1980; Wirawan 1981; van Balgooy and Tantra 1986; Whitmore and Sidiyasa 1986
Figure 5.19. Pterospermum celebicum. Scale bar indicates 1 cm.
After Soewanda n.d.a
Certain other areas appear to be dominated by single species, although not to the same extent. In the dry hills west of Marisa, Gorontalo, for example, Adina fagifolia (Rubi.) is unusually common with about 12 large trees/ha (Steup 1935), and Elmerrilia ovalis (Magn.) appears to be extremely common in parts of Minahasa, eastern Bolaang Mongondow and the area east of Lake Poso (Steup 1931, 1932, 1933).
Palms are a common sight in lowland forest and include the black-spined Oncosperma horridum, the trunkless Licuala celebensis, the thin-trunked and spineless Pinanga, Areca, fishtail palms Caryota, and the wild sugar palms Arenga sp. (fig. 5.28). This sugar palm can be distinguished by the pale undersurface of the leaves and the black 'hair' around the leaf bases. The distinctive Pigafetta filaris with its green trunk and grey rings is generally found in somewhat disturbed sites (p. 396).
Figure 5.20. Tetrameles nudiflora. Scale bar indicates 1 cm.
After Soewanda n.d.b
Figure 5.21. Dracontomelum mangiferum. Scale bar indicates 1 cm.
Alter Soewanda n.d.b
Figure 5.22. Celtis philippinensis. Scale bar indicates 1 cm.
After Soewanda n.d.a
Figure 5.23. Octomeles sumatrana. Scale bar indicates 1 cm.
After Soewanda n.d.a
Figure 5.24. Adina fagifolia. Scale bar indicates 1 cm.
After Soewanda n.d.a
Figure 5.25. Metrosideros vera. Scale bar indicates 1 cm.
After Soewanda n.d.b
Figure 5.26. Manilkara celebica.
After Soewanda n.d.b
Figure 5.27. Diospyros celebica. Scale bar indicates 1 cm.
After Soewanda n.d.a
One of the most common palms in Sulawesi is the tall fan palm Livistona rotundifolia, the leaves of which can reach 130 cm in diameter and have hard, sharp spines on the leaf stem. A total of 88 were found in one hectare at Toraut (Whitmore and Sidiyasa 1986). Observations of two seven-year-old individuals with trunks over 2 m tall in Bogor Botanic Gardens showed that new leaves were produced every 20 days (Mogea 1986). The rate of production of new leaves was in individuals with similar, shorter and no trunks monitored at Toraut, and was found to be much slower than this (EoS team). The actual rate of height increase in palms is complicated because the distance between the leaf scars (the internodes) is greater in shady conditions than in full sun, and also by the faster growth of young palms. For example, a coconut may have 6 leaf scars/m on the lowest part of its trunk but 90 scars/m at 35 m above the ground (T.A. Davis pers. comm.).
Felled palm trees such as Livistona rotundifolia soon develop hollows in the top of the standing stump of their trunks where the softer pith has decayed and these, like pitcher plants (p. 450), soon develop unique animal communities. These were examined at Toraut and the top carnivore was the nymph or larva of a libellulid dragonfly Lyriothemis cleis which fed on the larvae of beetles, mosquitoes and crane flies (Tipulidae) which also lived in the water-filled hole (Kitching 1986).
Details of the plants found in the lower layers of lowland forest in Sulawesi can be found elsewhere (Wirawan 1981; van Balgooy and Tantra 1986), but one small tree worthy of mention is Dendrocnide (Urti.) the leaves and stems of which have stinging hairs. The cells of the hairs contain quantities of silica which make them hard but brittle. When a hair enters the skin and breaks, both the tip and the sap, which acts on nerve endings, remain in the skin to cause irritation. The irritation is long lasting, at least in part, because the hairs remain in the skin.
Composition of Pioneer- and Building-phase Forest
As indicated earlier (p. 359), the plant species found in the younger phases of forest growth are quite distinct from those in the mature phase. The actual composition follows a certain trend, but is virtually impossible to predict because factors such as past history and treatment of a site as well as chance tend to complicate any trends predicted from the measurement of soil conditions and microclimate. Plants in these phases appear to have been studied only once in detail in Sulawesi,22 along the boundary of Lore Lindu National Park (Wirawan 1981; Rombe 1982) and much of the information below is taken from that work.
In areas where fires occur more or less annually, the vegetation is dominated by grasses such as Arundinella setona, and Themeda triandra. The well-known sword grass or alang-alang Imperata cylindrica is not as widespread as is often stated and it is unjustly associated with the worst of land management. It is generally found on roadsides and at sites where fires are less than annual. When I. cylindrica does start to grow in grassland dominated by other species, it can be an indication of an improvement in soil conditions. Also, because it is relatively tall, certain tree seeds have a greater chance of germinating and starting growth under I. cylindrica than under shorter grasses. Where fires are relatively infrequent, the grasses are invaded by the tough fern Dicranopteris linearis and the cosmopolitan bracken Pteridium aquailinum. Grasslands may contain trees which are resistant to fires such as Morinda tinctoria (Rubi.), Lagerstroemia speciosa (Lyth.) (a tree with masses of violet flowers, frequently planted in cities), Fagraea fragrans (Loga.) and Albizia procera (Legu.) (Steup 1931, 1939a, b; Jacobs 1977). These trees are more resistant to fire than most others but the young trees are susceptible. The height to which flames reach obviously depends on the height and nature of the grass but scorch marks have been found up to 5 m on Anthocephalus (Rubi.) trees near Kendari (Jacobs 1977).
Figure 5.28. Silhouettes of the distal ends of some Sulawesi palm leaves, a - Caryota mitis, b - Areca vestaria, c - Pinanga sp., d - Gronophyllum selebicum, e - Arenga sp., f - Licuala celebensis. Note that to date, the tree palm Gronophyllum is known in Sulawesi only from Southeast Sulawesi and the area between Malili and Poso.
Dransfield 1981
Some trees, such as Ficus miquelli (Mora.), Nauclea sp. (Rubi.) and Evodia sp. (Ruta.), can be found immediately after the garden or ladang is abandoned until the building-phase forest is formed. Other trees, like Trema (Ulma.), Orophea (Anno.) and Macaranga (Euph.), tend to grow from under a low shrub layer (Wirawan 1981; Rombe 1982). It should be noted that some of the trees of the early succession stages are valued and of great use to nearby villagers. Trees most notable in this respect for use as firewood, fence poles and in house construction are Orophea sp. and Nauclea sp. (Wirawan 1981).
In ladang areas that are reoccupied every decade or so a succession of species and of dominant life forms can be observed from herbs, through shrubs to trees although there is considerable overlap. Some herbs such as Curculigo latifolia (Amry.) and the fern Nephrolepis bisewata are able to maintain themselves under increasing tall vegetation but most of the herbs and grasses that invaded the ladang initially are not present even under low forest. Similarly, almost all the shrubs and small trees such as Grewia sp. (Tili.), Homalanthuspopulneus (Euph.), Malastomapolyanthum (Mela.) and Blumea balsamifera (Comp.) die as the vegetation grows above them. Soil fertility is not necessarily associated with length of the fallow period, but rather with the kinds of plants growing in the area. Thus, plant species are good indicators of soil fertility and are used as such by traditional shifting cultivators (p. 570) (Wirawan 1981; Dove 1985).
Pigafetta filaris, a majestic palm of Sulawesi and the Moluccas, is distinctive in having lines of shiny golden-brown spines along the bases of its leaves, and a dark-green trunk with light grey rings where the leaves have fallen off. It is a palm that favours altitudes between 300 and 1,000 m. It appears to be unusual among the palms of Southeast Asia in being adapted to growth in secondary habitats, in that it is fast-growing, tolerant of sunlight, and produces enormous numbers of small fruit (p. 363). For example, the three specimens by the front door of the Bogor Herbarium have trunks 15 m tall but were planted as seed just twelve years ago. During this time they have fruited 15 times, and new leaves appear to be produced about every 16 days (Mogea 1986). It is rarely found in mature forest, although it can be found at the site of a treefall or along a riverbank where new sediment and light penetrating to the forest floor have resulted in suitable conditions. Seedlings can sometimes be found growing together in clumps on the forest floor, as though they have been deposited in animal faeces, but these plants rarely progress beyond the one-leaf stage; it appears that the seedlings are not tolerant of shade. In the past this palm was presumably much rarer than at present, being found primarily on landslips, volcanic screes and flows, riverbanks and very steep-sided ridgetops (Dransfield 1976). Its trunk is used as supports for the traditional houses and rice barns in Tana Toraja, and is also used in the same area to make water conduits (Sneed 1981; S.C. Chin pers. comm.).
All the gaps in the forest of Toraut (p. 360) were colonized by two species of Macaranga and a number of gaps were colonized by Piper aduncum (Pipe.) and even grass (Whitmore and Sidiyasa 1986). P. aduncum is a wide-ranging tree of open sites and forest edges in the New World tropics from Mexico to the West Indies and down to Argentina (W. Burger pers. comm.). It was introduced to Bogor Botanic Gardens one hundred years ago (C.G.G.J, van Steenis pers. comm.), and is now common in many parts of West Java. In North Sulawesi it seems to have found conditions eminently suitable to the extent of out-competing the relatively few indigenous species of pioneer tree. In certain areas it grows in pure stands of secondary forest adjacent to undisturbed forest. Its competitiveness is quite remarkable and it will be interesting to see whether it extends its range even further. This success is partly because P. aduncum grows very quickly and mature specimens flower and fruit continuously both in its native habitat and in Indonesia. The minute crowded flowers on the spike-like inflorescence appear to be pollinated by small sweat bees Trigona (p. 418), and the ripe infructescence is taken by small fruit bats23 (p. 421). Each tree appears to produce at least one ripe infructescence each night. As has been found in Central America, almost all the ripe fruits are taken by bats.
Other trees that can be found in almost pure stands in secondary growth are Nauclea and Anthocephalus macrophyllus (fig. 5.29) (Steup 1939a, b). Where such trees are established, reforestation is unnecessary because the shade they cast will kill the grasses and create conditions favourable to other tree seedlings. The planting of suitable shade-tolerant seedlings may be appropriate, however.
A scheme of forest succession on Sulawesi is shown in figure 5.30.
Figure 5.29. Anthocephalus macrophyllus.
After Soewanda n.d.
Figure 5.30. Schematic pathways of succession on Sulawesi.
After steup 1931, 1933, 1939a, b; Bloembergen 1940; Wirawan 1931; Rerung 19S3
ANIMAL COMMUNITIES
Soil and Litter Communities
The soil community lowland forest at 130 m above sea level in Sarawak comprised 2,579 visible invertebrate individuals/m2 and this declined with even small increases in altitude. The dominant group of detritus feeders was the termites (62% of the total individuals) followed by earthworms (4%) and beetles (3%), and the major predators were ants (20%). The relative contribution of the groups to the biomass is somewhat different with termites accounting for 33% of the 4.1 g total, followed by earthworms (29%), ants (11%), millipedes (6%) and beetles (5%) (p. 523) (Collins 1980a). As for the animals not visible to the naked eye, 1 m2 of soil in Peninsular Malaysia was found to have 10,000 mites and 56,000 nematode worms (Chiba 1978).
Termites24 may look superficially like ants (Hymenoptera) but they are classified in a completely different order (Isoptera) which is more closely related to cockroaches (fig. 5.31). Termites are like ants, however, in that they form enormous colonies with (at least in some parts of the world) possibly a million or more members. A colony is not a simple community of distinct individuals because all except two of the colony members are siblings. The two exceptions are the 'royal pair' or parents. The parents originated in other termite nests from which they flew, along with thousands of others, to seek a mate and a new nest site. Some manage to escape the hordes of predatory ants, amphibians, reptiles, birds and mammals to which such a swarm is a food bonanza. After landing the termite wings drop off and if a male and female meet they will look for a suitable crack in the ground or a tree (depending on the species) and here they build their 'royal cell'. They copulate and the female begins laying eggs. The larvae, unlike the helpless larvae of ants, bees and wasps, are fully able to move around and undergo several moults (like cockroaches), rather than a single metamorphosis before becoming adults. These first larvae have to be fed by their parents but as soon as they are large enough to forage for food and to build walls for the nest, the royal pair devote themselves entirely to the production of eggs. The abdomen of the female, or 'queen', grows enormously and a production rate of thousands of eggs per day is common.
A termite colony can be likened to a single, although sometimes disparate, organism because none of its components is capable of independent life. The workers are blind and sterile, and the soldiers which protect the columns of foraging workers and guard entrances to the colony's nest have jaws so large that they can no longer feed themselves and have to be fed by the workers. The king and queen also have to be fed by the workers. In the same way that communication between different organs or parts of a single body is effected by chemicals flowing through a body's tissues, so chemicals—pheromones25—link the different members of a termite colony into a coordinated and organized whole. All of the colony members continually exchange food and saliva which contain pheromones. Workers pass these materials from mouth to mouth and also consume each other's faeces to reprocess whatever partially-digested food remains. Workers feed the soldiers, larvae and the royal pair as well as collect the queen's faeces. There is thus a continual interchange of chemicals through the colony and this coordinates the operations of the colony. For example, all the larvae are potentially fertile termites of either sex but the food with which the workers feed them contains quantities of a pheromone from the queen which inhibits the development of larvae and produces sterile workers. The soldiers also produce a pheromone which prevents the larvae from developing into soldiers. When, for some reason, the number of soldiers falls, the level of the 'soldier repression' pheromone circulating through the colony members falls and some of the eggs develop into soldiers. On other occasions the queen will reduce her repressive pheromones and her eggs will develop into fertile winged termites which, when the air is moist, will leave the nest in a swarm and the cycle of colony formation begins again.
Figure 5.31. Typical worker and soldier termites (left) and ant (right).
Most invertebrate decomposers of the lowland forest soil depend on free-living saprophytic fungi and bacteria to break down indigestible plant material into a digestible form before they can play their role. Most types of termites feed on already decomposing litter and dead wood, and their digestion is assisted by symbiotic gut protozoa or bacteria. One subfamily of higher termites, the Macrotermitinae, represented in Sulawesi by Macrotermes and Odontotermes, maintain a 'garden' of fungus in their nests which releases them from competing for partially-decomposed material outside the nest. The workers of the above genera build complex frameworks, or combs, from their own round faecal pellets. Their main wood and leaf litter, and consequently the combs, look like fragile pieces of moist, spongy rotten wood. These combs are the food of a form of fungus called Termitomyces which is unknown outside termite nests. If a nest is cut open, small white dots can sometimes be seen on the combs and these are clumps of asexual fungal spores (Collins 1980c).
This relationship with the fungus confers several advantages on the termites. No animal has the necessary digestive enzymes to digest lignin, a largely inert compound which is the main component of wood and thus one of the major items in a termite diet, and so lignin is excreted unchanged in the termite faeces. The fungus can, however, digest lignin. The termites are also unable to digest cellulose—the main component of plant cell walls—but the fungus produces the necessary enzymes for digesting it. When the termites eat the fungus-permeated comb, these cellulose-digesting enzymes persist in the termite gut, thereby improving the efficiency of its own digestion (Collins 1980c). The fungus releases carbon dioxide as a result of its respiration. This has the advantage to the termite that the carbon-nitrogen ratio in the combs increases roughly four-fold when processed by the fungus (Matsumoto 1976, 1978). The tremendous contribution the fungus makes to the nutrition of the termites has resulted in these termites being unable to live without Termitomyces—an example of obligate symbiosis. It has made them so successful that they have been reported to be responsible for the removal of 32% of leaf litter in a forest in Peninsular Malaysia (Matsumoto 1978). The role of termites in the decomposition of fallen trees was studied in the same forest and it was found that the rate of wood removal was higher for small branches (3-6 cm diameter) than for trunk wood (30-50 cm diameter) which is harder and more difficult to cut and remove. In a species of dipterocarp, 81 % of the tree's dry weight of small branches was removed in the first 18 months after it fell compared with just 18% of the trunk in the same period (Abe 1978, 1979). Termites thus play a major role in the breakdown and cycling of plant material (Wood 1978).
The percentage of forest litter production consumed by termites in Southeast Asian forests appears to depend on whether the fungus-farming termites are present. For example, in an area where they are present, consumption of organic matter by termites was calculated to be 155-174 g (dry-weight)/m2/yr, equivalent to about 15% of litter production, whereas at another site the figures calculated were only 7-36 g (dry-weight)/m2/yr, equivalent to just 0.9%-3.4% of litter production. At the first site, where annual rainfall was 2,000 mm, the fungus-termites were responsible for the consumption of 75% of the total consumed by all the termites together, and at the second site, where about 5,000 mm of rain fell each year, the fungus termites were absent or rare. The difference may be caused by the leaf litter in the wetter area being dominated by free-living decomposer organisms which rob the fungus termites of their advantage (Collins 1983). Thus the decomposition processes should be strikingly different between the drier and wetter forests of Sulawesi.
Figure 5.32. Termitodiscus indonesiensis beetles from a termite nest near Rantepao.
After Kistner 1984
Various animals live in termite nests, among which are Termitodiscus beetles (Staphylinidae) (fig. 5.32). The first species known from Sulawesi, the reddish-brown T. indonesiensis with a total length of about 2.5 mm, was found near Rantepao in a nest of the fungus termite Odontotermes sundaicus. The beetles feed on fungus in the fungus gardens and their main social adaptation is avoidance of the workers and soldiers of their termite hosts (Kistner 1984). In addition to these specialized beetles some termite nests also have phorid flies in the galleries where they breed in the fras. Ant nests also have aliens in their midst and a peculiar phorid fly was found in one such at Toraut (Disney 1985, 1986a, b, c, in prep.).
The number of termite species known from a given area of lowland forest has not yet been ascertained in Sulawesi, but there are at least 55 species of termite at Pasoh Forest Reserve in Peninsular Malaysia (Abe and Matsumoto 1978, 1979). For population studies of these, the four most abundant species with conspicuous mound nests, part of which projected above the soil surface, were chosen (Matsumoto 1976, 1978). Approximately 100 nests were found per hectare for three of the species but there were only 15 large nests of Macrotermes. The population size of a Macrotennes nest was about 88,000, however, at least twice that of the other species.
Forest Floor Community
The forest floor community is taken to include terresuial animals that spend a major part of their life walking on the forest floor. Among the smallest visible members are certain beetles which are exceptionally successful at utilizing dung, carcasses, rotting wood and fruit. Dung or scarab beetles (Scarabaeoidea) can comprise 50% of the arthropod biomass on the forest floor and they have an enormous diversity. The principle roles they play in the decay of organic matter are in fermentation, burial, parital assimilation of the organic matter, transportation of micro-predators and parasites, and aeration of the soil. In addition, they may be active in dispersal of seeds by rolling seeds into their balls of dung (J. Krikken pers. comm.).
The decomposer animals also include the rarely seen land crabs (p. 523). Only one species Gecarcoidea lelandii is known so far from the lowlands of Sulawesi (fig. 5.33) (Turkay 1974) although unidentified species have been caught in Bogani Nani Wartabone and Morowali National Parks. Four species of land crabs were caught in flood-prone alluvial forest in Sarawak where soil invertebrates in general are poorly represented, and near a river one species of crab with a carapace width of 20-35 mm reached the high density of 0.32/m2, suggesting it must play a major role in litter breakdown (Collins 1980b).
Beetles and other invertebrates utilizing the shrubs and herbs of the forest floor fall prey to birds and skinks. There are at least 19 species of skinks that live on the floor of lowland forests but how these divide up the available resources is not known. Perhaps the most startling species is Emoia cyanura with its iridescent, light-blue tail.
From the point of view of a human, two of the more unpleasant animals in the ground community of lowland forests are mites and leeches. These are locally abundant but rarely seem to occur together at the same place at the same time. Whether this is spatial or temporal (seasonal) separation is not clear. It may be that mites are relatively more active than leeches in dry conditions, or that certain habitats are generally more favourable to them.
Mites are one of the world's most ubiquitous groups of animals, being found near the North and South Poles, in deserts, hot springs, as well as in rather less extreme habitats. A handful of leaf litter from the forest floor can contain hundreds of mites from many species and many are parasitic on forest animals.
Figure 5.33. Gecarcoidea lelandii, the only land crab so far described from Sulawesi.
After Turkay 1974
The land leeches Haemadipsa are a harmless but not entirely welcome component of many lowland forests. Until recently, very little was known about their habits, mainly because little money has been expended on research concerning animals whose bite, albeit painful and liable to infection, does not transmit disease. Haemadipsa leeches will, if left undisturbed, feed for about 80 minutes and large individuals may feed for longer. The blood ingested exceeds the initial body weight of the leech by a factor of about 6 (although 14 times the initial body weight has been recorded), and after feeding the leeches are somewhat sluggish and only become active again after half the newly-gained weight is lost. Larger leeches tend to lose their weight more quickly but at a slower rate relative to their body weight compared with smaller leeches. A single meal may provide sufficient sustenance for at least three to eight months (Fogden and Proctor 1985). These data help to explain why leeches are so rarely found on wild animals: a one-hour feed every three months is equivalent to spending only 36 seconds feeding over 24 hours. They appear to be common on humans, but possibly only because humans tend to walk along paths. If leeches do not travel far from where they drop after a feed, and if they are repeatedly attracted to a path by warm bodies passing by, it follows that they will have a very non-random distribution.
One or two hours in a forest is long enough to hear the sound of branches, boughs or even a tree falling to the ground. Falling branches will occasionally land on a termite nest, breaking part of it open. When this happens, termite predators such as ants, devil's coachmen beetles (Staphylinidae) and skinks are rather quick to take advantage of the easy meal as are those phorid and muscicapid flies that suck up juices from termites crushed in the accident. At Toraut, females of two species of interesting scuttle flies Diplonerva (Phoridae) were observed at such a scene at a nasutine termite nest. The female would arrive at the damaged nest and diligently seek out worker termites which would normally not be found in exposed situations. Having found one she would prod its abdomen with her head which caused it to 'break rank' and turn toward the fly. The fly would quickly run round to the worker's head and run away closely followed by the worker. Should the worker lose interest the prod catch-me-if-you-can routine is repeated. After they have walked a safe distance from the termite nest, the worker is put into a coma by the fly which then lays her egg in the worker's abdomen. The mechanism by which she does this is not yet understood but if she killed the termite, it would putrefy and kill the larvae. The fly probably guards the comatose termite during the larva's development (Disney 1985, 1986b).
There are three main groups of birds on the floor of lowland forest— the pittas, the scrubfowl and maleo, and the ground pigeons. There are three pittas known from Sulawesi26 and they are among its most beautiful birds. They feed by tossing leaf litter aside with their beaks and snapping up any adult or larval insects that may have been uncovered. They are rarely seen except by patient and quiet observers, but their clear 2-3 note whistles, trilled or slurred, are distinctive. Their nests, in which they lay 4-5 eggs, are constructed in tree stumps, at the base of bushes, among the roots of trees or simply on the ground within some herbaceous cover. In Papua New Guinea certain pittas decrease in abundance during the dry season (Bell 1982b) possibly because abundance of suitable food in the litter drops at that time, or because it is too hard to dig in the soil.
The habits of the maleo have been described elsewhere (p. 155) and its nest sites can be found in lowland forest. Like the nest sites around the coast, the eggs are often plundered in a quite unsustainable manner. For example, an EoS team visiting the site near Hungayono, the enclave within the western end of the Bogani Nani Wartabone National Park, found all of the 22 nest pits dug out. Its close relative, the scrubfowl Megapodius cumingi is a brown partridge-like bird, whose most distinctive feature is the bare red skin around the eye. They are shy and generally solitary birds but they congregate for calling displays which are followed by pairing and breeding. Their nesting habits are somewhat similar to those of maleo birds but they lay their eggs in a wider variety of sites and use not only heat from the sun and from volcanic activity, but also from the decomposition of vegetable matter which they scrape over the eggs after they have been laid at the bottom of a deep (± 1 m) hole (MacKinnon 1978). As with maleos, they disperse widely through lowland forests when not engaged in breeding activities, avoiding swamps, major disturbance and cultivated areas (Broome et al. 1984).
Figure 5.34. Pseudoscorpion Megachernes grandis from the fur of a rat.
After Durden 1986
The ground pigeons Gallicolumba trisligmata, Chalcophaps indica and C. stephani are shy and little is known of their habits, but they appear to supplement a fruit diet with insects from the forest floor.
The other major group of animals on the forest floor are the rats such as species of Rattus, Maxomys, and Paruromys. In Tangkoko-Batuangus Reserve it has been estimated that they live at a density of about 20/ha (Anon. 1980). Studies of forest rats in Peninsular Malaysia found that the diameter of the lifetime range of individual rats was 250-500 m (equivalent to 0.05-0.2 ha) and that a 'lifetime' for a rat was an average of only 3-6 months (Harrison 1955, 1958). Some species are not confined to the forest floor but also climb.
The rats of the forest floor also support an interesting assemblage of parasites among their fur. One particular Musschenbroek's rat Maxomys musschenbroekii caught in the Toraut forest was infested with no less than 129 mesostigmatid mites, 7 astigmatid mites, 4 ticks, 79 sucking lice and, most interestingly, three pseudoscorpions (fig. 5.34) (Durden 1986). Pseudoscorpions usually live in or on the soil. They were first described from a living mammal when a new genus and species Chiridiochernes platy-palpus, 3 mm long and morphologically adapted for life in fur, was found on a summit shrew rat Bunomys penitus on Mt. Lompobatang in 1969 (Muchmore 1972). This pseudoscorpion, and those from Bogani Nani Wartabone National Park, are incapable of feeding directly on mammals and probably prey on the smaller mites.
Different groups of rat parasites encounter their hosts in different ways. Most fleas and the larger mites are common in rat nests, but sucking lice do not form a reservoir in the nest, while ixodid ticks and chiggers search for their hosts in the forest. These differences in ecology are reflected in the rate of reinfestation for the different groups (table 5.9; fig. 5.35). The figures show a remarkable consistency in the composition of the parasite fauna, indicating that the relative numbers are dictated by conditions on the rat rather than simply by chance.
Parasites are defined as any organism that is intimately associated with, and metabolically dependent upon, a host organism for the completion of its life cycle, and which is typically more or less detrimental to the host (Lincoln et al. 1982). They are to some extent similar to herbivores for these too depend on 'host' plants for survival and cause the plants stress. Even so, there remain a number of subtle differences (table 5.10).
Frogs and toads27 are conspicuous components of the ground community and they feed on termites, small flies and other invertebrates. When toad tadpoles hatch from pools on the forest floor they are by no means cryptic; instead, they shoal together, indicating perhaps that they are distasteful to predators. At least one of the narrow-mouthed toads Oreophryne celebensis is found in the lowlands. Its eggs are laid not in water but in moss on tree trunks, hollow old tree fern trunks, cavities in ant-plant stems, etc., where eggs develop directly into miniature toads without going through a tadpole stage (J. Dring pers. comm.). These and adult frogs are subject to predation by lizards, snakes and bats (p. 548).
The largest Sulawesi mammals are members of the ground community: the anoa Bubalus depressicornis, babirusa Babyrousa babyrousa and the Sulawesi civet Macrogalidia musschenbroecki (p. 532). The anoa is a close relative of the domesticated water buffalo B. bubalis and the small buffalo of the Philippines B. mindoroensis. The lowland anoa is relatively large (nearly 1 m at the shoulder), has a relatively long tail, white legs and rugged horns, whereas the mountain anoa B. quarlesi is smaller (about 75 cm at the shoulder), has a shorter tail and smooth conical horns (Groves 1969). Their ferocity and unpredictable behaviour is attested to by villagers and scientists who have spent long periods in the forest, and some of them bear the scars made by the sharp, stout horns. The anoa used to be caught by Toraja people in attempts to breed it for its meat but its aggressiveness, even after several years in captivity, prevented it from being used directly as a domestic animal. Considering the close relationship between anoa and water buffalo, it is conceivable that they would interbreed and produce fertile offspring which might have some potential as stock animal. Such a cross has not yet been attempted. Of all the five species of wild cattle in Southeast Asia, the anoa are exceptional in being the only ones whose major habitat is undisturbed forest. Their food includes a variety of fruits, leaves of shrubs and young trees, grasses and ferns (table 5.11). Dung samples collected by an EoS team are being analysed but the results are not yet available. In and around ladangs, plants with latex, such as cassava, are said to be favoured.
After Durden in press
Figure 5.35. Examples of rat parasites. a - flea, Siphonaptera, b - tick Haemaphysalis, Ixodidae, c - mite, Laelaps, Mesostigmatidae; d - chigger mite, Leptotrombidium, Trombiculidae; e - mite Listrophorus, Listrophoridae, f - louse, Anoplura.
After a - Barnes 1968; b - Hoogstraal et al. 1963; c - Evans et al. 1961; e - Fain and Hyland 1974; f -Barnes 1968
Anoa appear to prefer to feed in well-drained, rugged areas where the ground is not so thick with herbaceous plants but with a variety of available food species. They generally lie down to ruminate along dry ridge tops in forest which is relatively open compared with wetter slopes (Wirawan 1981).
After Janzen 1985a
Large herbivores throughout the world seem to experience a shortage of sodium in their diets: this element is scarce in green plants (with the exception of seagrasses and other marine species) but is essential to mammals. Carnivores do not face the same problem because flesh comprises about 0.1% sodium. Many herbivores, including deer, pigs and feral water buffaloes, make up for the shortage by licking rocks or soil that are relatively high in sodium. Wildlife surveys in Central Sulawesi failed to find any such licks frequented by anoa, although anoa themselves were common enough. Licks that were found appeared to be frequented by deer and pigs rather than anoa. It seems, therefore, that anoa either did not need to make up any salt deficiency or obtained salts from other sources. For example, anoa have been seen to drink seawater. Their normal diet appears to comprise growing tips from a wide variety of plants although signs of intensive feeding in a single area were rarely seen. Anoa tracks are frequently seen, however, around springs of water (Anon. 1979; Wirawan 1981) and it might be that these supply mineral needs. Water analysis showed that although some springs had high sodium concentrations (Wirawan 1981), their mineral content in general was barely different from that of adjacent rivers which were apparently not used intensively. The frequent use of the springs is seen rather as a means by which anoa, which live mainly solitary lives at low densities, can judge the size of the local population; it also helps bulls to trace oestrous females (Watling in press). Hunters take advantage of this behaviour, although anoa hunting has been prohibited since 1931.
After Wirawan 1981
Babirusa are primarily nocturnal although they are occasionally seen in daylight hours, too. Observations suggest that babirusa do not dig in soil to eat roots as do most pigs but rather eat fruit and break open rotten wood to obtain beetle larvae. The major fruit eaten by babirusa in the west of the Bogani Nani Wartabone National Park is the potentially poisonous pangi Pangium edule (Flac.)28 (fig. 5.36). A major fruit eaten in certain coastal regions is coconut but the babirusa is not regarded as a coconut pest because it takes only sprouted coconuts or broken pieces of flesh. They can inflict great damage, however, by rooting in ladangs; it is for this reason that people set traps for them and kill them. In Islamic areas the meat is not eaten but given to dogs.
Only one study of the babirusa has been conducted and that was on Pangempan Island in the Togian Islands in 1978 and in 1979-80 (fig. 5.37) (Selmier 1983). This small island (22.5 ha) was chosen for the study after surveys were made on all the Togian Islands on which babirusa were thought to live, although it had only one small herd comprising an adult female, two adult males and two piglets. This composition was maintained over three years, suggesting that the young babirusa swam to the mainland or died. They are said to be able to swim strongly.
Babirusa have no teeth when born and young males initially develop canines in the normal way. Soon, however, the sockets of the upper pair turn upside down such that these tusks grow up through the skin (Groves 1985), and curl around towards, but not into, the skull.
The function of the curly tusks of babirusa has intrigued many people. Wallace wrote:
It is difficult to understand what can be the use of the extraordinary hornlike tusks. Some of the old writers supposed that they served as hooks, by which the creature could rest its head on a branch. But the way in which they usually diverge just over and in front of the eye suggested the more probable idea, that they serve to guard these organs from thorns and spines while hunting fallen fruit among the tangled thickets of rattans and other spiny plants. Even this, however, is not satisfactory, for the female, who must seek her food in the same way, does not possess them. I should be inclined to believe rather that these tusks were once useful, and were then worn down as fast as they grew; but that changed conditions of life have rendered them unnecessary, and they now develop into a monstrous form. — WALLACE 1869
The question of function has recently been addressed again as a result of examining the patterns of wear on the tusks of a sample of skulls (MacKinnon 1981). On the skulls from Sulawesi, all the lower tusks had been sharpened and half of them were chipped or broken. All the upper tusks on skulls of adult animals were chipped or broken and most of these had deep grooves and scratches beneath them, probably caused by collisions with hard, sharp objects. In most pigs, the sharpening of the lower tusks is achieved by the upper tusks growing across them and rubbing against them as the mandible is moved, but in babirusa the sharpening appears to be caused by rubbing against small trees, as anoa sharpen their horns and civets sharpen their claws. It must take male babirusa a considerable time to sharpen their tusks, indicating that these sharp weapons are extremely important for them. In addition, the fact that the upper tusks suffer so much damage indicates that they are highly functional, rather than simply decorative to attract mates. Since females do not have large tusks, it is reasonable to suppose that the males use them in fights between males over rival claims for females or territory. The upper tusks could not be used for butting opponents, first because the height of the lower tusks would require the head to be held very low to butt and, second, because the tusk tips would hit against the frontal bones of the skull which do not appear to be thickened as they are in the giant forest hog of Africa which does fight by butting.
Figure 5.36. Pangium edule. Scale bar indicates 1 cm.
After Meijer 1974
Figure 5.37. The Togian Islands to show Pangempan Island, site of the only long-term study of babirusa.
When male babirusa fight they stand side by side and push each other with their shoulders. They then stand facing each other and rear up on their hind legs, jabbing their heads upwards to gore each other with the lower tusks. At this point the upper tusk can be hooked over the sharp lower tusk of the opponent, effectively disarming it, while the lower tusk of the advantaged male can be used to jab or cut the other's face, neck or throat. The eyes are to some extent protected from damage by the upper tusks which could deflect strikes by the lower tusk away from the forehead (fig. 5.38). Interestingly, since the form of the upper tusks is that of a logarithmic spiral, a form frequently found in nature, the tip of the tusk is always pointing inside the curve and so it is very difficult for a captured tusk to slip out once caught.
The only Sulawesi animals that could prey on babirusa, and then only on the young, are large pythons Python reticulatus and P. molurus and the endemic civet, but probably neither has a very potent effect on the population. This is rather different from the pressures on most pigs which have to contend with predation from wild cats, dogs and bears. Perhaps in response to this, the litter size in babirusa has no need to be large and the tusks which are used to defend the piglets in most female pigs, including the species endemic to Sulawesi, are poorly developed in female babirusa (MacKinnon 1981). The endemic pig Sus celebensis which arrived in Sulawesi relatively recently has a litter size of two to eight piglets with an average of five (Anon. 1983), compared with one or two slow-developing young for babirusa.
One of the reasons the babirusa is no rarer that it is, is because it is not eaten by those who adhere to Islam. The law for Moslems (and for Jews) states that animals with cloven hoofs that do not chew the cud may not be eaten. All other pigs have simple stomachs and do not chew the cud, but one of the many peculiarities of the babirusa is its complex stomach. Its digestive apparatus suggests that it may be able to break down cellulose with the aid of bacteria, but this has yet to be confirmed. Suggestions that babirusa actually do chew the cud are founded in wishful thinking rather that fact. It was hoped by some that babirusa would become a 'halal' or kosher pig, thereby proving its usefulness and ensuring its continued survival, but it is more likely that if religious prohibitions against eating its meat were ever lifted, then the wild populations would suffer dramatically.
Figure 5.38. Male babirusa locked in hypothetical combat. The upper tusk of the male babirusa on the left is used to hook over the sharp lower tusk of his opponent, thereby disarming him. The advantaged male can now jab the throat of his opponent. The long-term field study of babirusa by Lynn Clayton has never observed fighting in this way.
Lower and Upper Canopy Communities
For the purposes of this section, the lower and upper canopy refer to the oligophotic and euphotic zones of a forest respectively (p. 348), but the lower canopy does not include the forest floor or short herbs and shrubs. Since there are considerable differences in microclimate through the canopy (p. 348), one would expect to find differences in the animal communities.
Traps were set up at four levels through the canopy of lowland alluvial forest at Morowali National Park to investigate differences in insect abundance through the canopy as part of Operation Drake. Low-power ultraviolet lamps were placed in the traps and the catches were compared. It was clear that some of the insect groups caught were far more abundant in the upper canopy than the lower (fig. 5.39), but the mayflies were most common in the middle layers (Sutton 1983).
Trapping, using the same techniques and equipment, has also been performed in Zaire, Panama, Papua New Guinea and Brunei, and the concentration of insects in the upper canopy is more marked in Sulawesi than in any of the other sites. The marked gradients observed at Morowali seem to be a feature of flat sites, particularly where tree height is relatively uniform. Insects which are dependent on leaves for some aspect of their life would be expected to be concentrated in the tree crown and one would also expect there to be clusters of insects focused in the different crowns. Thus traps set to attract insects in rugged country or in a forest with a complex canopy would not produce such a clear gradient (Sutton 1983).
In forests with marked dry seasons, arthropods in the lower canopy tend to be most common in the wet season, but where the dry season is relatively mild, the structure of the forest buffers the drying effects and the greater frequency of sun flecks reaching the forest floor increases the productivity of plants in this zone with a consequent rise in arthropod abundance. This was the case in an everwet forest in Peninsular Malaysia, but the fourfold variation in abundance was not correlated with the amount of rainfall or flower, fruit or leaf production. Ants were the most abundant of these arthropods, followed by beetles, wasps and bees. The relatively low level of variation is probably due to the dependence of these groups on resources other than the immediate products of plant productivity (Wong in press a). Since foliage is available continuously in the lower canopy, however, caterpillars and their predators are not so uncommon (Wong in press b).
The relative abundance of different groups of birds also differs through the canopy. In Queensland forest in Australia, the major factor in niche separation of birds is differential use of the vertical strata. This is true both within and between the different guilds.29 Some groups of species had similar vertical ranges and their niches were determined by differential utilization of food resources, different feeding behaviour and changes in preferred vertical strata through the day (Frith 1984). There are also differences through the day with many species of bird feeding high in the canopy during the early morning and moving down through the forest as the temperature rises. In some cases, even males and females of a single species have different preferred height strata (Bell 1983c). Among Sulawesi birds one of the most striking divisions of canopy use is found in the hornbills. The large red-knobbed hornbill Rhyticeros cassidix is found in the upper canopy and the garrulous dwarf hornbill Penelopides exarhatus below (Watling 1983).
Figure 5.39. Abundance of six insect groups at four height levels through lowland forest in Morowali National Park. Note that the frequency scale on the top and bottom rows are different.
Data from Sutton 1983
Seven species of hornbills were studied in an area of East Kalimantan and a wide range of social systems was found (Leighton 1982). Extrapolating from the results of that study, the large Sulawesi hornbill is probably nomadic, occasionally forming large, loose flocks of up to 50 individuals, but breeding30 in monogamous pairs away from others of the species (fig. 5.40). The smaller hornbill is found in small groups at all times and is probably territorial and breeds monogamously but remains in groups. In East Kalimantan such birds live at a density31 of about 5/km.2 It is quite likely that young birds are fed not only by their parents, but also by other 'helper', non-breeding members of the group. Such helpers are known for many kinds of birds (Brown 1978).
Figure 5.40. Adult female hornbill in her tree hole nest, the entrance of which has been reduced to a mere slit using mud brought by the male. Until the young are able to fly, both the female and the young are entirely dependent on the male (and 'helpers' in the dwarf hornbill) for food.
Colonies of stingless sweat bees nest in cavities in moderately large trees (about 1 m diameter) with entrances at 0.5-4 m above the forest floor. In Costa Rica and Panama, the most common species have nest densities in the order of 20 nests/km2 (Johnson and Hubbell 1984). Nests are sometimes built in cavities in the base of the trunk, but can also be found underground around the tree roots. The nest mouth is sometimes no more than an oval opening with black wax around its edge, but it can develop into a brittle horn perpendicular to the trunk. These bees are catholic in their choice of pollen and nectar, and on a single day twenty species of flower may be exploited. Despite this, however, a single bee tends to concentrate on a single plant species (Johnson 1983). The bees are small, less than 1 cm in length, but they can range up to 2 km from their nest and still return successfully (Roubik and Aluja 1983a, b). The workers leave the colony at dawn, and start returning in the middle of the morning. The number outside continues to decline until late afternoon when a minor burst of activity occurs. Foraging32 has ceased by dusk.
Niche differentiation has been studied in a pair of sweat bee species which were more or less the same size, collected pollen in the same way from the same type of flowers, and lived in the same area. As explained elsewhere (p. 230) the manner in which these species are able to coexist would therefore be of considerable interest. It was found that one of the species foraged in large groups in areas with a high density of a particular flower, whereas the other foraged in smaller groups where the flower was comparatively scarce. The second species lost in aggressive encounters with the first species (Johnson and Hubbell 1975; Hubbell and Johnson 1977).
The most remarkable member of the lower canopy community is the large-eyed jumping tarsier Tarsius spectrum, one of the world's smallest primates, with a head-and-body length of just 10 cm, a tail 20 cm long, and a weight of 100 g. The tarsier social unit comprises an adult pair, which forms a stable, long-term monogamous relationship, together with their one or two immature offspring. They all sleep together in a tree hole within a more or less permanent territory which is defended from other tar-siers by song. Every morning, just before these nocturnal animals retire to their nest, the entire family sings a complicated territorial call. The male generally initiates the song with a regular series of squeaks, and the female joins in with a descending series of squeals which then rise in pitch to a fast climax. These calls declare to surrounding groups that the mated pair is present, is fit and claims the immediate surrounding area as its living space. When neighbouring tarsier families come close to the territory the resident group will chase the intruders away, frequently shrieking loudly as they do so.
The calls made by tarsiers are excellent clues for finding and plotting sleeping sites and so observations, at least in the early morning and evening, are relatively easy. The sleeping sites include plant thickets, tangles of vines or ferns, tree holes (with more than one exit), and the criss-cross roots of strangling figs (p. 358). Tarsiers have been found in a wide range of habitat types from urban areas, secondary growth, mangrove, lowland forest, riverine forest and montane forests, with densities ranging from 3-10/ha (MacKinnon and MacKinnon 1980b).
After leaving the nest hole, typically 20 minutes after sunset, the family spends only a few minutes interacting before the members leap off in different directions. In a single night a tarsier may leap between trees for a total of 1 km. Virtually their entire diet comprises insects; other types of animals such as crabs, shrews, small rats and frogs seen close to the tarsiers during the detailed study at Tangkoko-Batuangus, attracted little interest from the tarsiers. Most hunting occurs in the lowest 1.5 m of the forest although it also occurs on the ground and up to 9 m above the forest floor (fig. 5.41).
The home range of tarsier families at Tangkoko-Batuangus is about 1 ha, an area determined by the furthest points of ranging of the adult male who is the most active in terms of calling and marking branches with urine. Within this area, the animals regularly use the same routes. The home ranges of different families overlap to some extent but the sleeping sites are generally in areas used exclusively by the family (fig. 5.42).
Figure 5.41. Differential use of the canopy by tarsiers for five major activities.
After MacKinnon and MacKinnon 1980b
The social organization of tarsiers is very similar to that of gibbons which have been studied intensively in various parts of western Indonesia. The tarsier call has considerable similarities with the structure and function of the 'great call' of gibbons, and the song structures of both vary geographically. Similarities between gibbons and tarsiers are also found in their specialized (though different) forms of locomotion, their active nature, their longevity (at least 25 and 10 years for gibbons and tarsiers respectively), their specialized diets composed of foods which are rich in nutrients and reliable, but very widely and thinly spread, and the stable nature of their environment. Their monogamous, territorial lifestyle confers advantages such as:
• minimizing reproductive wastage because only one young is born at a time and can therefore be protected and cared for to the greatest extent possible (i.e., they are K-selected [p.345]);
• minimizing time spent in conflict with neighbours because each group knows the boundaries of the area it can peacefully live within;
• minimizing time spent in searching for food and its food sources because the relatively small home range and its food sources are known well; and
• minimizing time spent competing for mates because the pair bond is more or less permanent, and is maintained outside the breeding periods by complex, coordinated duetting.
Figure 5.42. Ranges and sleeping trees of tarsier families near the shore in Tangkoko-Batuangus Reserve. White circles represent sleeping trees.
After MacKinnon and MacKinnon 1980b
There are disadvantages too, however, such as the inability to exploit rich food sources outside the home range, the small group size which cannot be exceeded without reducing the efficiency of finding food, and extreme niche specializations (MacKinnon and MacKinnon 1984).
A human observer restricted to walking on the forest floor gets a biased view of the nature of a forest. The opportunity to either climb to the upper canopy using rope slings and mountaineering ascending/-descending equipment (Perry 1978; Perry and Williams 1981; Whitacre 1981), or to walk along an aerial walkway (Mitchell 1982) brings a completely new perspective on forest life.
An aerial walkway in two spans (41 m and 70 m long) about 30 m above the forest floor was built in Morowali National Park as part of Operation Drake, and was used for a number of studies, one of which was on insect abundance (p. 425) and another on the composition and abundance of the fruit bat fauna.33 When dusk approaches an observer standing on the forest floor frequently sees a large number of small insectivorous bats flitting erratically around the lower canopy, but rarely sees fruit bats, which fly in a much straighter, more determined fashion. Nets set on the aerial walkway and on the forest floor below revealed that fruit bats were 50 times more common above the canopy than below it, and three times more common than in any open habitat, except over water, where bats frequently come to drink water while flying along. The five most common species of fruit bats in the canopy comprised 95% of the total catch and this diversity was considerably higher than in any other habitat or even in all the other habitats of Morowali National Park combined. Netting of bats in upper montane forest showed that the three most-common species, (90% of individuals), were the same as those representing the 5% of rare species at the lowland site. Similar work was conducted in two forests in Papua New Guinea and it was found that the fruit bat fauna over the forest growing on richer soils was consistently more diverse than that over forest on poorer soils. Interestingly, the testes of adult males in one particular species were larger in those individuals caught in the open than in those caught over forest (Gaskell 1984). This difference in testis size suggests that there is some difference in behaviour between those that are reproductively active and those that are not, and that the open areas are chosen specifically, perhaps for some form of display.
Fruit bats rely more on vision than on smell when feeding, and a recent study of bat eye anatomy revealed that not only are insectivorous and fruit bats extremely different in their visual apparatus, but also that fruit bats' eyes are extremely similar to those of primates in their high-acuity binocular vision. This is a feature which, until recently, was believed to be unique to primates. This discovery encouraged f urther investigation and it was concluded that fruit bats are closer in many respects to primates than they are to insectivorous bats. Indeed, it has been suggested that insectivorous bats evolved from flying shrew-like insectivores in the late Cretaceous (70 Ma ago), but that later in the Tertiary (60 Ma ago) an early line of primates developed the ability to glide and then to fly. Indeed, it now seems that to exclude fruit bats from the primate order requires the redefinition of primates to include the feature that they do not fly (Pettigrew 1986a, b). It is of some interest that in the first classification of mammals by Carl von Linne (p. 669), bats were grouped together with the primates.
Flying foxes (large fruit bats) tend to roost in camps with others of their species and hundreds can be seen at a single site. This may, however, be a mere shadow of the numbers that roosted together before man began hunting. For example, a single camp extending 1 km x 7 km was reported from 50 years ago in Australia that may have contained about 30 million animals (Pierson 1984).
The majority of fruit bats feed predominantly on a range of forest and garden fruit34 (Marshall 1985) but they are extremely selective concerning its state of ripeness. Even slightly under-ripe fruit, of a ripeness enjoyed by humans, is often rejected and for this reason the claims that fruit bats are pests of orchards are unfounded. In most cases, the fruit eaten by bats is too ripe to be picked for sale. Fruit bats will normally pick a fruit and then eat it, either while hanging in the fruiting tree, or in a regular feeding roost some distance away. These bats ingest very little other than fruit juice and pulp using their muscular tongue to rub the food across horny ridges in the roof of the mouth (fig. 5.43). If the fruit is too large to be chewed all at once, the bat will hang from one foot using the other to hold the food against its chest. Fibrous pulp and most seeds are spat out; only the smaller seeds are ingested, but these pass rapidly through the gut unharmed and are voided after only about 20 minutes at which time the bat hangs from its thumbs with its head above its feet. It is generally advantageous to the plant to have its seeds moved some distance away from the parent (p. 379), and in the case of fig seeds, germination will occur only if they have passed through an animal's gut. Fruit bats therefore serve trees not just by taking the seeds away to areas of new opportunity (or death), but also by removing pulp in which eggs or young larvae of potential seed predators may be developing before burrowing into the seed (Janzen 1982).
Figure 5.43. Upper palates of fruit bats showing the ridges against which fruit pulp is rubbed to extract the juice.
Fruit bats in Africa and Asia may have to ingest up to 2.5 times their body mass in fruits in a single night, although more commonly it is 1.5 times. This high intake seems to be due to their need to obtain sufficient protein because they do not include insects in their diet.35 Most fruits contain less than 5% protein (dry weight) (table 5.12) (and are thus amongst the most protein-poor plant tissues), and because of this most frugivorous birds supplement their diets with insects, seeds or flesh. A small number of fruit have about 6-7% of their wet weight as protein and birds specializing on these do not need to seek protein from other sources (Thomas 1984).
One fruit which is eaten by bats is notably protein-rich, and is produced by the small introduced tree Piper aduncum (Pipe.) (p. 397) of forest edges and gaps, particularly in North Sulawesi. A similar species of Piper has been found to contain sufficient nitrogen to meet not just normal maintenance metabolism, but also the needs of lactating female fruit bats. It is not surprising, then, that fruit bats, especially lactating females, show a preference for Piper fruit to relieve them from the need to seek large quantities of protein-poor fruit (Herbst 1986).
Fruits eaten by bats tend to be soft, juicy, somewhat musty or rancid and positioned such that they can be easily taken by a bat. The units of food taken away can be quite large but it was a surprise when a Sulawesi rousette bat Rousettus celebensis was caught in a mist nest on the canopy walkway in Morowali National Park, together with a large seed from the tree Gonystylus macrophyllus (Thym.). These seeds are 'naked' in that they have almost no aril, exocarp or appendages, they hang from the ripe fruit by 'ropes' making their harvest easy, and are quite large (3.6cm x 2.8cm x 2.1cm). This is most unlike typical bat fruits and the possibility that it was a chance capture was examined. A total of 297 seeds were gathered from the forest floor and 60% of these bore marks identical in size and distance to the marks made by the teeth of R. celebensis. If the bat flies to a regular feeding roost and then drops the seed, germination may occur but very few dropped seeds would survive the intense competition in the limited area around the feeding roost. It may be that its awkward size and shape may result in a high rate of fumbling by the bats and therefore a relatively even spread of seeds around the parent plant (fig. 5.44). It seems as though it must be the seeds which are eaten, but they taste very bitter to humans and may contain either or both of saponins and alkaloids. Feeding trials with captive bats and investigations of the R. celebensis digestive system might provide some explanations. Interestingly, the seeds of G. macrophyllus from other parts of its range (most of Indonesia) are of different shapes and sizes. This may indicate morphological adaptations in response to different sets of dispersers (Kevan and Gaskell 1986).
From data supplied by the Institute for Research and Development of Agro-based Industry, Bogor, and results of analyses conducted by that institute of samples collected by an EoS team.
Figure 5.44. Sulawesi rousette Rousettus celebensis and the awkward fruit of Gonystylus macrophyllus.
After Kevan and Gaskell 1986
The aerial walkway at Morowali National Park also facilitated studies of the pollination of forest trees. One species, Syzygium lineatum, was studied in depth. The trees flowered and fruited more or less simultaneously and their small, abundant, nectar-rich white flowers predispose them to a wide range of generalist pollinators, a common feature of tropical forest trees. The numbers of flower visitors (short-tongued wasps and flies), and fruit dispersers observed, however, were few. This is not particularly surprising in a species-poor forest, but it would be interesting to observe how the tree is serviced in other parts of its range. The observations also showed a clear pattern of activity among the flower visitors with numbers increasing until about 1400 hrs., after which numbers fell. Very few visits were recorded after 1600 hrs. (fig. 5.45) (Lack and Kevan 1984).
The abundance and activity of plant-sucking insects (Hemiptera), comprising the plant bugs (Homoptera) and the leafhoppers, frog-hoppers, cicadas, aphids and scale insects (Heteroptera), in the upper canopy in the lowland forest at Morowali also varied through the canopy but this appeared to be related to differences in microclimate. The more rain that fell, the more bugs were caught and this was more noticeable within the upper canopy than below it. In addition, with increasing rain the average (modal) size of these insects increased, and the species of these insects flying on wet nights differed from that found on dry nights. Interestingly, fewer bugs were caught on moonlit nights. When the data were examined, it was clear that it was the actual occurrence of rain, not just high humidity, and of darkness that resulted in large numbers of flying hemipterans being caught. The primary predators on these insects are insectivorous bats for which raindrops would greatly reduce the efficiency of echo-location36 because both raindrops and many hemipterans are about 3 mm in diameter and length respectively. With sound having a velocity of about 300 m/s in very humid air37 a sound with a 3 mm wavelength has a frequency of 100 kHz and this is in the upper range of bat calls (Fenton 1983; Hill and Smith 1984). The heavier the rain, and the denser the rain drops, make conditions for larger flying insects 'safe' from bat predation. On dry nights only the smallest (<3 mm) hemipterans fly but even these may concentrate their activity in microhabitats which bats would find hard to enter, such as among leaves and fine branches. The above explanation is elegant, but the abundance of flying hemipterans during rain might just as well be a response to low bat activity at such times because of the possibility of losing too much heat from the water covering and dropping off the wings and body (Rees 1983).
Figure 5.45. Abundance of insects observed feeding from marked flowers of Syzygium lineatum in Morowali National Park.
After Lack and Kevan 1984
Closely-related species living in the same area frequently have different body sizes. This allows them to eat different types of food, and to use different elements of the habitat available. Large birds can take large fruit but smaller birds have access to more types of fruit by virtue of being able to perch on the more slender branches. Some years ago it was proposed that, for birds, the weight ratio between sympatric members of the same genus should be at least 1:1.2, or at least that bill size should differ by this ratio (Hutchinson 1959). A study of two genera of New Guinea pigeons showed that the ratio of weights between one bird and the next smallest closest relative was on average 1.90; it was never less than 1.33 and never more than 2.73. Relatives with a weight ratio of less than 1.33 were too similar to live in the same area and no species pair with a weight ratio in excess of 2.73 was found, presumably because another species of intermediate size could coexist with both the larger and smaller species (Diamond 1973).
An attempt was made to determine whether the same ratios were valid for two groups of Sulawesi upper canopy birds: the parrots and fruit-eating pigeons. A simple listing of the pigeons (table 5.13) shows that most of the weight38 ratios are in fact below the 1.33 found to be the minimum in New Guinea. Closer examination of the likelihood of overlapping niches shows that not all species should be compared in this analysis. For example, the pink-headed imperial pigeon Ducula rosacea differs little in size from the species immediately larger and smaller than itself, but it is found only on the Tukang Besi Islands and the islands south of South Sulawesi and cannot really be considered a mainland species. Also, amongst the remaining species, some are predominantly montane while others inhabit lowland forests. The pied imperial pigeon Ducula tricolor is found primarily on small islands (and occasionally in other coastal habitats), whilst the rather similar silvery imperial pigeon D. luctuosa is found primarily in open woodland, cultivated areas and forest edge habitats. Thus, if the weight ratios of the four species from lowland forests, and the four from montane forests are calculated, it is found that they fall within the figures mentioned above with the exception of the two green pigeons, the grey-faced Treron griseicauda and the pink-necked T. vernans. These species are commonly seen in and around Lore Lindu National Park, and are sometimes seen feeding together in the same tree, and the ecological separation of the two is by no means clear (Watling 1983). It may be that the heavier bill of T. griseicauda is an adaptation to certain fruit that cannot be eaten by T. vernans.
L = Lowland species |
M = Montane species |
Data from Meyer and Wigglesworth 1898; Goodwin 1970; Watling 1983
A similar pattern to that found among the pigeons is found among the parrots (table 5.14). Only two species, the golden-mantled Prioniturus platurus and the yellow-and-green lorikeet Trichoglossus flavoviridis, are found in montane forest, and the two major habitats utilized are lowland forest and open areas such as secondary growth and forest edges. If these are taken into account then, once again, the weight ratios are within the range expected. As with the pigeons, one pair remains an enigma, however, and that is the two endemic racquet-tails. The red-spot racquet-tail P. flavicans is found only in the northern peninsula and is especially common in the Bogani Nani Wartabone National Park, whereas the golden-mantled P. platurus occurs throughout Sulawesi. There is virtually no difference in size and it may be that P. flavicans is able to hold its own against P. platurus but is less well adapted to dispersal. P. flavicans is more common than P. platurus in Bogani Nani Wartabone National Park (Roden-burg and Palete 1981), and Tangkoko-Batuangus Reserve (Anon. 1980). P. platurus probably arose in southern Sulawesi and extended its range northwards, and it would be interesting to study the ecological interactions of the two species.
The applicability of the size ratios does not seem to extend to all bird genera, and particularly not to those which are insectivorous (Simberloff and Boecklen 1981; Bell 1984). In contrast to the pigeons and parrots whose fruit food is concentrated in the upper canopy, the insectivorous birds have a wide range of options available for their feeding niche because their prey are available from the uppermost branches to the forest floor. Thus, many similar-sized species of the same insectivorous genus can coexist by specializing in taking the same or similar food as other species, but in places or by methods unavailable to the others (Bell 1984).
Mixed flocks of mainly insectivorous birds are occasionally seen in lowland forests. Such flocks in Lore Lindu National Park comprised core species which were almost always present, species which were frequent participants, and species which participated occasionally (table 5.15).
These mixed flocks have been interpreted as a means by which relatively specialized insect-eating birds can increase their food supply, particularly during times of food shortage, because of the insects flushed by the activity of the other species in the flock. The different responses of insects to disturbance and the different specializations of the species could ensure benefit for all the flock members (Croxall 1976). Studies of such flocks in Papua New Guinea also found that the participating species modified their normal behaviour to accommodate the activities of the other birds (Bell 1983).
The largest mammals of the upper canopy, although none of them is confined to it, are the cuscuses and the macaques. The large, dark-brown bear cuscus Ailurops ursinus39 eats leaves especially of Dracontomelum dao (Anac.), Finis spp., Syzygium spp. (Myrt.) and Garuga floribunda (Burs.) and is generally found in pairs which are active in daylight. The density of this cuscus at Tangkoko-Batuangus is estimated to be about one pair per four hectares (Anon. 1980). Four hypotheses have been proposed for the feeding behaviour of generalist herbivores such as these (Freeland and Janzen 1974), and they have been shown to hold for the related possum Trichosurus vulpecula of Australia (Freeland and Winter 1975). The hypotheses are:
Key: |
|
O - Open areas |
L - Lowland forest |
Data from Meyer and Wigglesworth 1898; Forshaw and Cooper 1978; Watling 1983
• a generalist herbivore has to ingest several different plant foods in order to meet its energy requirements;
• large amounts of a single plant food are not eaten when a generalist herbivore first encounters it, the animal initially taking small samples in preference to eating a large meal;
• the amount of a single plant eaten can be gradually increased as the animal gains 'experience' with it; this probably is a result of the animal inducing enzymes to detoxify the food; and
• nontoxic foods are recognized quickly, larger amounts of them being eaten than can be eaten of single toxic foods. It is likely that these hypotheses will be found to hold true for Ailurops ursinus if an intensive study of it, a relatively easy task, is ever conducted.
After Watling 1983
The dwarf cuscus Strigocuscus celebensis differs from its larger relative by being nocturnal and frugivorous but is similar in that it lives in pairs and is predominantly arboreal. Both species use their prehensile tails as a fifth limb and the lower surface of the end of the tails is bare, presumably to improve the grip when hanging on to branches. Cuscuses are frequently described as being slow-moving and in this regard loosely resemble the sloths of Central and South America. They are also similar in that they both have relatively thick fur. Experiments on the spotted cuscus Spilocuscus maculatus (which has been introduced to Salayar Island) have shown that its metabolic rate is indeed relatively low for marsupials, as is the metabolic rate of sloths compared with other placental mammals (Dawson and Degabrielle 1973).
Habitat selection by cuscuses is not understood but it is interesting that in a study of their close relatives, the possums, in Queensland, the density of animals was directly related to the concentration of potassium in the foliage. This high-potassium foliage was most common in trees growing on nutrient-rich, basic soils (Braithwaite et al. 1984). This in itself is not exactly news-worthy but it serves to remind planners that if only nutrient-poor or otherwise agriculturally-unacceptable soils are left for conservation purposes, then densities of animals will probably be lower, and hence larger areas will be needed to conserve the animals present.
In terms of relative abundance, macaques are by far the most common large mammal in the forest (Sungkawa 1975). Only one of the four similar-looking species of macaques on Sulawesi (p. 69) Macaca nigra has been studied in detail, and that work conducted in Tangkoko-Batuangus Reserve from 1978 to 1981 has not yet been written up in detail. What is known from that study in Tangkoko-Batuangus and from shorter studies of that and other species indicate that all the species are very similar ecologically. The major exception is the very high density of the macaques at Tangkoko-Batuangus (table 5.16).
After Wilson 1978; Bismark 1979, 1982a, b; Aldrich-Blake 1980; MacKinnon and MacKinnon 1980a; Whitten and Whitten 1982; K. MacKinnon 1983; and EoS teams
Groups comprise some adult males with a greater number of females. Individuals are very hard to recognize, although old animals of M. maura at least develop conspicuous white marks on their heads and forearms (Watanabe and Brotoisworo 1982) and grow somewhat bald. Group size can vary considerably, even within an area, but the social status, exclusiveness and constancy of these 'groups' is by no means clear. At Tanoma (Tanjung Peropa) groups seen ranged in size from 12 to 28 individuals. Various sites in that area were visited and it has been suggested that the relatively high density of macaques in one particular area is due to the high density of fig trees (Bismark 1982b). The few data available do indeed appear to show this but there also other trees favoured by macaques and the relationship is unlikely to be so simple. It is not stated, for example, whether there were any figs in the five locations where no macaques were seen. The group size of the Tangkoko-Batuangus macaques averages 30, but groups numbering 50 were also seen. These groups live in home ranges of about 15 ha in which the animals would move up to 0.5 km each day. The same species (though a different subspecies) in the central part of Bogani Nani Wartabone National Park40 were found to travel about 850 m in a day (Bismark 1982a).
Macaque groups are almost always accompanied in the canopy by the endemic malkoha Phaenicophaeus calorhynchus, a long-tailed bird which flies weakly and more often hops or creeps around the canopy like a squirrel. Its reason for following the macaques around is to catch grasshoppers and perhaps other insects as they are flushed out by the activities of the monkeys. It also appears to follow the somewhat clumsy, endemic coucal Centropus celebensis.
After MacKinnon 1983
The diet of macaques appears to be primarily fruit, with figs Ficus predominating. Other fruit taken includes Dracontomelum spp. (Anac.), Palaquium obtusifolium (Sapo.), Canangium odoratum (Anno.), Syzygium spp. (Myrt.), Spondiaspinnata (Anac.), and Pangium edule (Flac.) (Anon. 1980; Bismark 1982b; Watanabe and Brotoisworo 1982). Pet macaques are often seen to eat insects, so they probably do this in the wild, too. They may also open rotting tree trunks or disturb the litter to find beetle grubs, worms, etc., while on the ground.
The four macaque species differ somewhat in their conservation status, with M. tonkeana being both the most common species and the one with the least percentage of the population within protected areas. M. maura appears to be the most threatened with the smallest population, the smallest population in protected areas, and a distribution that includes some of the most densely-populated land outside Java, Madura and Bali (table 5.17).
Comparison of Mature- and Pioneer-phase Faunas
The high net productivity of the early phases in the forest growth cycle supports large animal populations, although the species are quite distinct from those in forests (Ewel 1983). A comparison was made of arthropod abundance and diversity between cultivated and secondary areas both near a transmigration site in Doloduo and in neighbouring disturbed areas. If the diversity of insects in the forest is taken to be 1, then the diversity was 0.4 at the forest edge and 0.1 in the rice fields (Doda 1980). In the course of collecting mosquitoes in and around Bogani Nani Wartabone National Park, it was found that there were about one-quarter as many catches in forest as there were in neighbouring farmland. Eight of the twelve mosquito species found in forest were also found in farmland but the abundance was far higher in farmland. Two species, Anopheles flavirostris and A. subpictus, are potential serious vectors of malaria and filariasis and were found in farmland but not in forest (Hii et al. 1985).
Many grasshoppers are found in and around forests but three species that might be considered agricultural pests penetrate no further than the forest edge (R. Butlin pers. comm.; K. Monk pers. comm.). Conversely no forest grasshoppers are considered to be potential pests. A similar drop in diversity between forest and disturbed areas has been observed among Collembola (P. Greenslade pers. comm.) and beetles (J. Krikken pers. comm.). The rats found living commensally with humans are rarely if ever found in forests and forest rats are never found beyond the forest edge. Not only that but the parasites they carry are quite different; both groups of rats carry fleas, but the forest species are only infested by fleas from the family Pygiopsyllidae, and the commensal species only by fleas from the Pulicidae (Durden 1986). These observations do not in any way support the contention, frequently heard, that forests are reservoirs of agricultural and human pests.
Studies of birds in forest gaps indicated that the number of species caught in nets in gaps was greater than those caught in surrounding forest. In addition, the assemblage of birds using the gaps was distinct from that in the forest, although there was, of course, some overlap in species (Schemske and Brokaw 1981).
The birds of Sulawesi have not been studied in this regard but it is likely that the situation is similar to that found in Papua New Guinea where about 15 species are more or less confined to secondary forests of the pioneer and early building-phases and about 50% of mature forest bird species might be expected to visit the younger forest. Only a very few species from the mature forest would visit, let alone feed in cultivated areas or in young pioneer-phase forests (Bell 1982d). A recent study in Central America found that migratory birds showed a preference for old pioneer- and building-phase forest due in part to the greater abundance of small fruits (Martin 1985).
There are considerable differences in microclimate between gaps and mature forest (p. 348) and this is reflected in the habits of some animals which are found in both habitats. For example, the orientation of webs of a particular spider was found to be north/south in closed habitats and east/west in open habitats. The longer a spider spends on the web, the more prey it captures (some prey manage to wriggle loose before being killed by the spider) and so presumably an orientation is chosen to allow maximum time on the web. Web orientation seems to be related to the body temperature of the spiders because both too little and too much heat can affect spider activity. Thus, in open habitats, the spiders' webs face east/west to reduce heat load and in closed habitats face north/south to increase body temperature (Biere and Uetz 1981).
THE EFFECTS OF OPENING FOREST
General Effects
The opening up of forest on a scale greater than occurs in natural ecosystems is accompanied by the following changes:
• the creation of open, hot, simple habitats containing relatively few, small, widespread species with broad niches and great reproductive potential (r-selected species-p. 345). These species are rarely, if ever, found in mature ecosystems;
• a huge decrease in biomass (± 30 kg dry weight/m2 in lowland forests to 0.2 kg dry weight/m2 in alang-alang plains);
• the temporary or permanent impoverishment of the soil;
• the creation of increasingly small and isolated patches of natural vegetation whose animal and plant diversity also progressively declines (p. 442);
• virtually all the rainfall reaches the soil surface, and far more rapidly, than in a forest, but less enters the groundwater and more flows in the surface runoff. This can lead to a considerable loss of soil, a decrease in groundwater supplies and an increased propensity to flooding.
These ecological effects can be observed whether the disturbed area becomes wasteland or valuable agricultural land.
The major effect of less intense disturbance (where the forest retains at least some of its former structure) is a simplification of the ecosystem caused by deleterious changes to the soil, hydrology and microclimate as well as by the actual removal of plant material such as logs. As taller trees are removed, so the volume of living space available to the forest biota is considerably reduced (Ng 1983). There is also less substrate available for use as nest sites, aerial pathways or growing sites for epiphytes and climbing plants. In addition, there is obviously also the loss of other resources, particularly food.
It is often stated that after forest clearance the loss of soil fertility and the loss of nutrients in plant biomass pose a serious threat to the integrity of most forested ecosystems. While this is certainly true for montane forests and forests on ultrabasic soils, in other types of forest, disturbance appears to result in succession which is directed towards the re-establishment of mature-phase forest, where continued disturbance is absent. Unfortunately, once disturbed, most forests in Sulawesi experience a long series of destructive disturbances which never allow the succession to progress. Effects which are sometimes attributed to the loss of nutrients from soils may possibly be due to competitive interaction between plants. Studies have shown that the fertility of soil does indeed decline following forest removal, but the evidence of short-term loss in soil fertility is usually accompanied by evidence that the successional vegetation is remarkably good at regenerating soil fertility (Soerianegara 1970; van Baren 1975; Harcombe 1980; Uhl and Jordan 1984). This is not an argument against taking all possible precautions to reduce nutrient loss, but it does indicate that further studies are required to elucidate the situation.
A detailed comparison of soil chemistry, soil respiration, seed storage and plant growth rates between soils of primary forest, cut-over forest, and cut-over and burned forest both before and after rain, has produced many interesting results (Ewel et al. 1981). Amongst these it was noted that after the disturbed vegetation had been burned, 57% of the initial amount of nitrogen and 39% of the initial amount of carbon remained in the soil. The number of viable seeds in the forest soil (the seed bank) was 8,000 seeds/m2 (67 species) but this was reduced to 3,000 viable seeds/m2 (37 species) after the burn. Thus vigorous and relatively diverse growth followed the burn because only a proportion of the seeds were released from the burnt material. The 5,000 seeds that did not survive the fire may have either been killed in the heat of the fire or, as might be expected for primary forest trees, be dependent on high humidity for germination (Ng 1983). If this is not available for a certain length of time, the seeds will die.
A 1 ha plot of Queensland lowland forest (which contains many of the genera found in Sulawesi lowland forest) was first felled and then burned shortly afterwards. Two years later the regeneration of the trees that had reappeared was studied (Stocker 1981). All of the 82 tree species present had regenerated in one or more ways; 74 species had formed shoots from the base of the old trunk, 10 had formed shoots from roots, and 34 had germinated from seed. This last category appeared to have the greatest growth rate, but frequency of shoots from old trunks shows that this is also an extremely important mode of regeneration. The ability of these trees to form shoots is probably limited, however, and repeated cutting or repeated burning may give smothering, light-demanding creepers the advantage and prevent regeneration from taking this form.
When forest is cut selectively for timber, recovery is relatively rapid because the seed bank contains many types of viable seeds and many tree species have the ability to sprout from stumps. When forest is cut, dried and burned and then abandoned, the succession proceeds more slowly because part of the seed bank has been destroyed and because some of the sprouting species are killed by fire (Uhl et al. 1981).
Thus when forests are cut, burned, farmed, weeded or burned and farmed again, recovery of the forest (if allowed) is extremely slow. In places where the only remaining stand of forest (source of seeds) is some kilometres away, full regeneration may take hundreds if not thousands of years. Even if a few mature trees have been left to stand, there is no guarantee that these will regenerate successfully because some species have seeds which have to pass through an animal's intestine before they will germinate (p. 381) (Ng 1983).
Seed banks are large in Imperata grasslands which have not been burned for five years or more, but small in similar but regularly-burned grasslands. In both cases, however, the seeds tend to be of agricultural weeds rather than of pioneer trees. Interestingly, Imperata seeds are not present to any great extent and this suggests that these grasslands are propagated by fresh seeds and vegetative growth from rhizomes (Hopkins and Graham 1983).
Wider Implications of Forest Conversion
The conversion of forests and changes in land use are sometimes suggested as a cause of the rise in levels of atmospheric carbon dioxide. The argument is that forests fix a great deal more carbon dioxide than grasslands or other vegetation types. Most of the Sun's energy passes through the atmosphere to warm the Earth beneath. The warm ground reflects some of this radiation and radiates some upwards at longer, invisible wavelengths in the infra-red. Carbon dioxide absorbs this radiation which is in turn radiated back at the Earth's surface. The more carbon dioxide, the more radiated energy is absorbed, and the warmer the atmosphere becomes. This is known as the 'greenhouse effect'. Levels of carbon dioxide have been increasing over the last 100 years or so and, if present trends persist, the concentration will have doubled from 0.03% in the middle of last century to 0.06% by the middle of the next century. Although still a small proportion of the atmosphere, the increase could well cause major climatic changes.
The global effects of atmospheric pollutants was brought into focus in mid-1986 by a report from the United States Department of Energy. This carried the full weight of establishment respectability behind an idea which, ten years ago, was thought to be no more than an eco-scare story. The fact now has to be faced that the world is getting warmer because of the carbon dioxide greenhouse effect. Rainfall patterns, temperature and sea-levels throughout the world will change, as they have indeed been changing for the last 100 years. Sea level is currently rising by about 0.1-0.25 cm/yr and may become even more noticeable in the next few decades.
Not all the carbon dioxide that should be in the atmosphere, according to calculations, is actually there, indicating that a large amount is being absorbed into natural reservoirs or 'sinks'. No one is certain where these are or how they work, but since plants absorb carbon dioxide from the atmosphere, it is reasonable to suggest that tropical forests may be one of the major sinks. As the area of these forests gets smaller, and the burning of fossil fuels increases, so the situation is aggravated.
Another possible deleterious effect on the atmosphere arising from forest conversion is increasing levels of methane produced as a by-product of termite metabolism. It has been estimated that termites are 8-10 times more common in land disturbed by human activities than they are in forests. Recent research has shown that termites produce about 150 million tons of methane each year, compared with a total global input of between 350 and 1,210 million tons. The level of atmospheric methane is increasing at about 2% per year, and most of the increase appears to originate from tropical areas undergoing development (Zimmerman and Greenberg 1983). The scale of the problem has been challenged (Rasmussen and Khalil 1983) and potential environmental impact of the increased level of atmospheric methane has yet to be assessed, but it is clearly important to be aware of the possibility and to encourage its study.
Effects of Selective Logging on the Forest
Selective logging is the removal from a forest of trees of designated species which exceed a designated size.
While selective logging clearly has many advantages, it also removes the best individual trees, leaving only inferior or younger ones to produce seeds for future 'crops'. Each time logging occurs, plant succession is set back a step, the forest becomes progressively poorer in desirable species and progressively richer in 'weed' species (Wyatt-Smith 1963; Whitmore 1984). For this reason, foresters examining regrowth in the Malili area, for example, suggested that timber trees be planted for the next cropping (Sutisna and Soeyatman 1984). This genetic erosion is potentially extremely serious for forestry in the future (Ashton 1980; Kartawinata 1980; Sas-traprada et al. 1980; Whitmore 1984) but in many logged areas it is doubtful whether in fact a full cycle of regeneration will ever be allowed to occur.
Selective logging is, in theory, a repeatable exercise in any given area. After the largest commercial trees have been extracted the area should be left for about 70 years until the next timber crop is harvested. Logging of a forest at intervals of less than about 70 years is unsound forestry practice but as more species become economically worth exploiting and when timber prices increase, repeated logging occurs. Over fifty species of Sulawesi trees are probably exploited commercially; for example, in an area near Malili, species such as Santiria laevigata (Burs.), Kalappia celebica (Legu.), Agathis alba (Arau.), Vatica flavovirens (Dipt.), and Calophyllum soulattri (Gutt.) are the major trees felled (Sutisna and Soeyatman 1984), but of these only Agathis is of any great value. In certain areas of Sulawesi the logging is extremely selective and only the large valuable Agathis are removed, because it is not felt to be economically worthwhile to take poorer species. As a consequence, the 'value' of the depleted forest to a logging company is greatly reduced and the overall damage and effects of logging in those areas will be less than in areas where a wider range of species is taken.
The term 'selective logging' sounds a very mild and benign activity to many people who have not worked in or visited a logging concession, and it comes as a surprise to learn just how much damage is often caused to the forest as a whole. One estimate is that five times more timber is destroyed or badly damaged than is extracted (Burgess 1971). Another illustration is that for every large tree felled, 17 similar or smaller-sized trees are destroyed (Abdulhadi et al. 1981).
The felling of trees and extraction of logs are the primary causes of disturbance during a logging operation. Felling damages tree crowns, boles and saplings, exposes wood, making it susceptible to fungus damage, and also covers seedlings. Extraction of the logs exposes bare soil and damages large areas of the forest floor (Kartawinata 1980).
Much disturbance is also caused by the access roads which remove all cover from the soil and from channels that are further scoured by runoff during storms. It has been estimated that 20%-30% of a logged area is completely bare, being composed of roads and logyards (Meijer 1970; Kartawinata 1980). Some of the roads cut across small rivers, acting as dams, and the subsequent flooding kills most of the inundated trees and other plants (Anon. 1980; Kartawinata 1980). The lorries used to drag logs out of the forest cause compaction of the soil surface and this disturbance is traceable in some forests by the occurrence of mature pioneer trees even decades after logging.41 By this time other pioneer trees have died or become rare as different genera take over dominance in the succession. The soil compaction caused by lorries is, however, only a fraction of that caused by the bulldozers used to make the roads. Additional disturbance is caused when trees are felled on both sides of the main logging roads to hasten the drying of roads. It is estimated that these 'daylighting strips' accounted for 8 ha/km of logging roads (Hamzah 1978).
There does not seem to have been a detailed study of the effects of logging on Sulawesi forest, but the results would probably be very similar to those found on South Pagai Island in the Mentawai Islands, West Sumatra (Alrasjid and Effendi 1979). A total of 2,416 trees (20 cm in diameter and over), half of which were commercial species, originally stood in the 15 1 ha plots that were enumerated. A total of 194 trees (13/ha) were felled and extracted. On average, about half of the remaining trees had been noticeably damaged, broken or knocked down, and the other half had escaped damage (fig. 5.46). This proportion seems to be relatively consistent between studies (Burgess 1971; Tinal and Palinewan 1978; Abdul-hadi et al. 1981).
The microclimatic changes caused by logging are obvious—it is hotter, lighter and drier in logged-over forest. These changes result in the dieback of crowns, scalding of sensitive trunks and branches, water stress and even an increased likelihood of insect attack, any one of which may lead to the death of a tree (Kartawinata 1980).
There are many ways in which logging operations and their supervision could be improved, such as by yarding or dragging the logs up, rather than down, slopes in order to reduce erosion, avoiding forests adjacent to rivers, felling trees in a direction such that their extraction causes minimal damage, and by mulching areas to be abandoned with forest litter (Marn and Jonkers 1982; Hamilton and King 1983). These do not necessarily increase the overall costs to the contractor.
Figure 5.46. The effect of extracting 8% of the trees by selective logging on an area of forest on South Pagai Island (Mentawai). The other categories are: totally destroyed; seriously damaged; badly damaged crowns and trunks; and undisturbed.
From data in Alrasjid and Effendi 1979
Effects of Selective Logging on Soil and Hydrology
The damage that logging causes to soil has barely been studied in Indonesia but the general subject has been reviewed (Kartawinata 1980). The major problems are soil removal by bulldozers when making roads, soil compaction caused by heavy vehicles, soil loss due to rain striking the ground with its full force, and the greater quantity of rain reaching the soil. The removal or destruction of trees also reduces litter fall and hence the organic inputs to the soil, but in areas where the soil has not been too greatly damaged, soil nutrient levels and micro-organisms can recover within a few years. Since only the larger portions of the tree trunks are removed, and most of a tree's nutrients are in its leaves and roots, the loss of biomass after logging is proportionally greater than the loss of nutrients.
It is obvious from a brief visit to a logging site that a great deal of water runs off the roads, carrying soil with it. This has been quantified by the measurement of the silt content of a river near, but not directly influenced by, a particular logging area; from a river in the logging area, and from a ditch by a logging road. The silt contents were 0.01%, 0.05% and 0.1% respectively—a ten-fold increase (Hamzah 1978). The increased silting of the major rivers in Sulawesi can probably be traced to logging as well as other forms of unprotective land use. In the lower stretches of major rivers the actual volume of water is probably not greatly affected by forest clearance, and it is in the higher river regions that dramatic effects are noticed (p. 630).
Soil erosion and subsequent sedimentation can have serious and expensive impacts on irrigation schemes. For example, the concession granted in 1978 to a logging company in the catchment area of the Gumbasa River south of Palu threatened to reduce the effectiveness of the Gumbasa irrigation scheme. This scheme was begun in 1973 with the intention of irrigating 11,500 ha of rice fields in the Palu valley, in order to make that region into a net rice exporter rather than a rice importer. Thirteen years later, only 5,000 ha were being irrigated, at least in part because of the highly erodible soils in the catchment area which had been washed downstream after the forest cover had been disturbed by logging and inappropriate farming practices. The irrigation canals now have to be dredged every year when about 30,000 m3 of soil is removed (Mattulada pers. comm.).
Many accusations are bandied about concerning whether concession loggers, illegal loggers or upland farmers are to blame for the continual degradation of forest lands (Anon. 1986a). Illegal loggers are significant not just in the opening up of the forest, but also in the loss of taxes and fees to the government. Illegal loggers are also guilty of encouraging the complicity of villagers by buying the wood they extract from the forest (Anon. 1986b). An example of the scale of this activity is given by the seizure in Palu in October 1986 of 668 lengths of ebony Diospyros celebica (Eben.) (p. 386) that had been taken illegally from a transmigration area on the pretext of establishing a cocoa plantation. By the time the wood was seized, it is estimated that 150 other lengths had already been taken to Poso (Anon. 1986c).
The concession holders are supposed to work within the limits imposed by the Department of Forestry but supervision of the operations is difficult and the regulations are open to abuse. Logged forest is easier to clear and burn than undisturbed forest. In this situation the farmer who takes advantage of this in production and other logged forest is most at fault, but the absence of any likelihood of being caught, let alone prosecuted, in these inevitably remote areas, means that he is likely to proceed with scarce worry or concern. The small chance of reprimand is also relevant in the clearance of protection forests. Where social pressures are such that farmers begin to endanger forests and soils, then the agriculture of the area needs close appraisal and assistance. This would not inevitably lead to the adoption of intensive rice culture (p. 572).
Effects of Selective Logging on the Fauna
The majority of mammal and bird species found in Sulawesi are dependent on mature forest. Considering the very large number of insects which are restricted to a single species of plant (p. 346), the percentage of insects dependent on undisturbed forest is probably also extremely high. There is clearly a need to know what effects different types of disturbance have on the fauna, although the trends and general patterns, at least for mammals and birds, are now quite well understood. Thus, the forest fauna is affected by disturbance in at least three ways:
• the noise and shock of disturbance may cause an immediate change in behaviour;
• the actual removal of parts of the forest canopy will alter ranging patterns and diet which may in turn affect social behaviour and population dynamics;
• the slow regeneration rate may cause permanent changes in population density (Johns 1985).
Different species show different degrees of tolerance to disturbance and at least some animals will usually be found even in the most disturbed areas. A study of terrestrial mammals in different habitats ranging from mature forest to an area of alang-alang grass in Peninsular Malaysia showed, not surprisingly, that the total number of species (rats) increased from 0%-100% (Harrison 1968) (fig. 5.47). A lowering of richness and evenness (p. 346) of small mammals species has also been reported from tropical Queensland where the effect was most marked on relatively infertile soils (Barry 1984).
Properly executed selective logging is not disastrous for much of the forest fauna although certain squirrels, bats and birds may fare badly (Marsh and Wilson 1981; Wilson and Johns 1982; Johns 1985; Johns et al. 1985). 'Properly-executed' in the ecological sense used here means an average of 8-10 and an absolute maximum of 15 trunks removed per hectare, no relogging for at least 50 and preferably 70 years, replanting with tree species found in the area, and the preservation of an adjacent area of mature forest from which fruits can be dispersed into the logged area. If such practices are adhered to, there is no reason why nature conservation and forestry should conflict. Logged forest supports a lower species diversity of animals but is able to maintain viable populations of many species.
The initial effects of selective logging on the fauna are probably the most serious and this is illustrated by a lower birth rate and a greater mortality amongst infant primates in logged areas (Johns 1985). These effects seem to be temporary, however, and the populations return to normal eventually. It is likely, however, that the effects of disturbance will last several decades because it takes that long for the lost resources such as food sources and nest sites, and for microclimatic conditions to recover (Johns 1985; Wong 1985).
Figure 5.47. The total number of non-flying mammal species in six types of vegetation, and the percentage of introduced species.
After Harrison 1968
The effects of disturbance depend in part on the social system and diet of the species concerned. Territorial species are worst affected because they are unable, even temporarily, to move out of the disturbed area because they will probably encounter defended territories of other groups of the same species. Non-territorial species may be able to withstand some temporary crowding before moving back into the disturbed forest. Species with relatively specialized diets would be expected to fare rather worse in disturbed forest than unspecialized species.
When forest is cleared completely for large-scale agricultural or other purposes, the animals that once lived in it will eventually die. The territorial species are very unlikely to be able to find unoccupied areas in the neighbouring undisturbed areas, and while the non-territorial species may be able to move into adjacent forests (assuming there are any), sooner or later the food resources will limit the population to more or less its original level.
Some of the displaced animals frequently enter gardens and steal food. The macaques are a major pest, particularly on crops of corn, although they will eat a wide range of other crops. It is clearly hard for them to resist food that has been planted within their range. It is these animals that are caught to become pets (Anon. 1978), and it has been estimated that there are at least 100 pet Macaca nigra along the Domoga valley alone (MacKinnon 1983). Babirusa are one of the first animals to become locally extinct after logging or land opening, possibly because village or feral dogs kill the piglets, a form of predation to which babirusa are poorly adapted (p. 414).