It is well known that tropical forests and savannas house a significant proportion of global biodiversity. However, an appreciation of the diversity of interactions among organisms in tropical ecosystems is only just emerging. Interactions among species are important because they affect the growth, survival and reproduction of individuals, but also because they have a key role in structuring communities and in the functioning of ecosystems. A sound knowledge of these interactions is therefore fundamental to understanding how tropical ecosystems work, as well as informing important practical concerns such as conservation, management and carbon sequestration. The aim of this book, and the meeting from which it derives, is to synthesize the current state of knowledge of biotic interactions in terrestrial communities in the tropics. Each of the 22 chapters of this volume provides a review or a case study of interactions among organisms from tropical ecosystems, with a perspective drawn from the organisms and sites with which the individual authors work. Our aim was to draw on research conducted in both Old and New World tropics and to include biotic interactions among taxa at all trophic levels. Most authors have taken plants (typically trees) as their starting point, but taken together the chapters consider interactions of plants with other plants, with micro-organisms and with animals, and the inter-relationships of human-induced disturbance with interactions among species.

Introduction

Herbivore attack is hypothesized to contribute to the high level of plant species richness common in many tropical terrestrial communities (Janzen 1970; Connell 1971). Janzen and Connell independently proposed that seed and seedling predation in tropical forests would prevent any one highly competitive plant species from excluding other species. Two sets of observations served as the basis for the initial hypothesis. Firstly, high seed and seedling predation is common for many tropical tree species, potentially influencing the population dynamics of those species (Janzen 1970; Connell 1971). Secondly, predation on a superior competitor in an adjacent lower trophic level in temperate marine systems was shown to reduce the likelihood of competitive exclusion (Connell 1971) or actually to reduce exclusion (Paine 1966). Since this hypothesis was first proposed (Janzen 1970; Connell 1971), herbivory and seed predation, at least for temperate systems, have been shown to influence the richness and diversity of plant species (Dirzo 1984; Augustine & McNaughton 1998; Howe & Brown 1999), vegetation structure (Brown & Heske 1990) and plant succession (Brown & Gange 1992). In turn, herbivory (Rausher & Feeny 1980; Doak 1992; Ehrlen 1995) and predispersal seed predation (Louda & Potvin 1995; Ehrlen 1996; Kelly & Dyer 2002) have been demonstrated to influence plant population dynamics in temperate systems.

Support for the impacts of herbivores on plant diversity in tropical systems is much less complete. At least five hypotheses have been proposed.

Introduction

Arbuscular mycorrhizal fungi (AMF) (Zygomycetes) are an ancient group, dating back to the invasion of land surfaces by plants. Currently, they are perhaps the most abundant soil fungi, and they form intimate relationships with the roots of the vast majority of terrestrial plant species across the planet. These fungal symbionts generally play a mutualistic role, aiding the host plant primarily by enhancing the acquisition of soil nutrients, particularly phosphorus (P). In addition, AMF species often affect plant hormone production/induction (Allen et al. 1980), resistance to root pathogens (Newsham et al. 1995); water uptake (Kyllo et al. 2003) and soil structure (Andrade et al. 1998; Rillig & Allen 1999). In return, all AMF species obligately depend on the host plant for photosynthetically fixed carbon. Given their obligate dependence, AMF are influenced by their hosts at essentially every phase in their life history – hyphal development, sporulation and spore germination (Hetrick & Bloom 1986; Sanders & Fitter 1992; Bever et al. 1996). On the other hand, the degree of mycorrhizal dependence often varies widely among the host plant species in a community (Janos 1980a; Azcon & Ocampo 1981; Hetrick et al. 1992; Kiers et al. 2000).

A central and still largely unanswered question is the degree to which host plant and AMF species influence each other's community composition in natural systems. Fundamentally, for community effects to occur, different combinations of host and AMF species must produce different outcomes of survival and growth.

Introduction

All plants need the same resources (water, light and nutrients) for their survival, growth and reproduction. Paradoxically, water is a limiting resource in many tropical rain forests. Variation in community structure, composition and functioning is driven to a large extent by variation in rainfall (Medina 1999; Poorter et al. 2004). Species composition changes continuously and gradually along the rainfall gradient (Hall & Swaine 1981; Bongers et al. 1999; Bongers et al. 2004), and within forests species tend to sort out along slope gradients in water availability (van Rompaey 1993; Webb & Peart 2000; Harms et al. 2001). Within a given forest a range of species coexist that differ in their drought tolerance. These species face to a greater or lesser extent a bottleneck during the dry season, when water is in short supply. Seasonal droughts (with low soil matric potentials) occur annually in tropical moist forest and occasionally in tropical wet forest (Veenendaal et al. 1996a). Even in seemingly aseasonal wet forests dry periods of 15–30 days may occur every other year (Walsh & Newbery 1999).

Within a certain forest type, variation in canopy openness leads to marked gradients in irradiance, temperature and relative humidity (Brown 1993). The light available in gaps can be as high as 30% of full light, whereas in the understorey it can be as low as 1%. Light is therefore the most limiting resource in many humid tropical forests (Whitmore 1996).

Introduction

In 1980 S. P. Hubbell and R. B. Foster began a long-term, large-scale study of tropical forest dynamics on Barro Colorado Island (BCI), Panama. The objective of the study was to test competing hypotheses about the maintenance of high tree species richness in the BCI forest, and in tropical moist forests more generally. Hubbell and Foster established a 50-ha permanent plot on the summit plateau of BCI, within which all free-standing woody plants with a stem diameter at breast height (DBH) of a centimetre or larger were tagged, measured, mapped and identified by 1982. Subsequent complete censuses of the BCI plot have been conducted from 1985 to 2000 at 5-year intervals. In setting up the BCI plot, Hubbell and Foster (1983) reasoned that whatever diversity-maintaining mechanisms were important, they would have to operate in a spatially dependent manner in communities of sessile plants such as the BCI tree community, which meant that the trees had to be mapped. A decade earlier, Janzen (1970) and Connell (1971) had independently proposed a spatially explicit ‘enemies hypothesis’, now known as the Janzen–Connell hypothesis. They hypothesized that host-specific seed and seedling predators were responsible for maintaining tropical tree diversity by causing dependence on density and frequency (rare species advantage), through an interaction between seed dispersal and density-dependent seed predation.

In 1980, there were essentially just two principal tropical forest diversity theories to test: the enemies hypothesis and its variants, and the ‘intermediate disturbance’ hypothesis (Connell 1977) and its variants that invoked a role for disturbances associated with opening, growth and closure of light gaps (e.g. Ricklefs 1978; Hartshorn 1978; Orians 1982; Denslow 1987).

Introduction

The first part of this paper examines the consequences of an interlocking set of mutualisms, involving ants, plants, bacteria and phloem-feeding insects, for the structure and functioning of herbivore-based food webs in tropical communities. This part draws heavily from important recent work by Davidson and colleagues (Davidson 1997; Davidson et al. 2003) and extends their discussion of community-level implications of their findings. The second part explores how trophic interactions evolve when coevolution produces specialized symbiotic ant–plant mutualisms, and is based largely on our own work on interactions between ants and Leonardoxa myrmecophytes of African rainforests. The paper complements a recent general review of ant–plant protection mutualisms (Heil & McKey 2003).

Ant–plant–herbivore interactions and tropical food webs

How food webs function, and how trophic interactions shape communities, have long been central questions in ecology. Interactions between organisms at adjacent trophic levels – predators and prey, parasites and hosts – and competitive interactions among organisms at the same trophic level, all occupy major roles in theories to explain the great species richness and other traits of tropical forest ecosystems (Wright 2002). Following the lead of classic studies like those of Hairston et al. (1960) and Paine (1966), investigations of how communities function have increasingly taken into account not only these direct interactions, but also indirect interactions that extend across several trophic levels. Do natural enemies of herbivores have measurable impacts on fitness of individual plants, on relative abundance of plant species, on primary productivity or on plant species diversity?

Introduction

The production of offspring has long been thought to be predominantly limited by the availability of resources rather than mating opportunities, a view based on ‘Bateman's principle’ espoused by Bateman in 1948. Applied to vascular plants, this principle predicts that fruit production is limited by maternal resources rather than pollen transfer. However, in the past 20 years limitation of seed production by pollination has been reported in numerous studies, and reviews suggest that more than 50% of plants studied show increased fruit production following experimental pollen supplementation (Burd 1994). Of course, short-term pollen supplementation studies fail to capture the lifetime success of the whole plant, leading some critics to maintain that lifetime reproductive output remains resource-limited. Nevertheless, population-wide declines in reproductive success, at least in the short term, owing to reduced pollen availability or ineffectual pollination have now been recorded among a range of plant species and geographic locations.

The cause of the declining efficiency in pollination has often been traced to changes in the spatial distribution of plants in the population, which in turn has affected the abundance of pollinators or led to changes in their foraging behaviour. Plant spatial distribution has become increasingly relevant following rapid changes in land use and landscape structure driven by anthropogenic activities. Logging and land clearance for agriculture and development has caused degradation of forest habitats either through the partial removal of economically important species, or by wholesale clearance and effective fragmentation and isolation of remnant forest patches.

Introduction

In many habitats ants form a major part of the arthropod fauna found on vegetation, and recent studies have shown that the abundance and diversity of ant–plant associations is particularly remarkable in the tropical region. For instance, one-third of the plant species in a Panamanian forest (Schupp & Feener 1991) and over 20% of the woody species in a Brazilian savanna (Oliveira & Oliveira-Filho 1991) were found to produce ant rewards. Furthermore, 312 ant–plant interactions were recorded in one Mexican coastal site (Rico-Gray 1993), and the ant–plant community in an Amazonian rainforest comprised 377 plants per ha (Fonseca & Ganade 1996). In the tropics many ant species use plant surfaces as a foraging substrate to search for both live and dead animal prey, as well as for different types of plant-derived food products (Carroll & Janzen 1973). Ant activity on foliage can be promoted by the occurrence of predictable and immediately renewable food sources, in cluding extrafloral nectar, honeydew from phloem-feeding hemipterans, and secretions from lepidopteran larvae (see Way 1963; Bentley 1977; Buckley 1987; Koptur 1992; Pierce et al. 2002). In fact, plant- and insect-derived liquid foods appear to provide a large amount of the energy supply of foliage-dwelling ants (Tobin 1994; Davidson et al. 2003). Although food resources located on foliage are probably more often found and exploited by arboreal species, ground-nesting ants frequently extend their foraging areas onto the plant substrate as well (Rico-Gray 1993; Blüthgen et al. 2000; Davidson et al. 2003).

Introduction

The neutral vs. structure debate

We can define a neutral community as one in which all species, and so all individuals, are equivalent, in the sense that they are interchangeable at all times and under all conditions. In contrast, we can define a structured community as one in which species are not equivalent, and species-specific differences affect the population dynamics, and therefore the behaviour, of the community.

This distinction is an important one, because in a neutral community the biodiversity, as measured by species richness and abundance patterns, has nothing to do with the biogeochemical functioning of the community (e.g. carbon fixation and nutrient-cycling). In fact, in a truly neutral community one could eliminate all but one species without affecting the biogeochemical functioning of the community at all.

In contrast, much of the species-specific variation in biological traits observed in reality (see below) has direct relevance for the functioning of the community. For example, the short-term carbon uptake of a forest depends on the growth rates of the individual trees, and the long-term carbon storage depends on adult life-span and wood density, and there is wide species-specific variation in these traits. In niche-structured communities, the biodiversity and functioning are intimately linked, and some combination of at least some species is required to maintain the functioning of the community.

Biodiversity as a process

Biodiversity is dynamic, with species richness and composition changing over time and space in response to ecological, evolutionary and physical processes (e.g. Todd et al. 2002; Prieto et al. 2001; Hart et al. 1989; Menge et al. 1983). The scale of changes in biodiversity can be large, such as changes in geological time from plate tectonics (e.g. Crame 2001), variable, such as expected from global climate change, or as small as those responding to localized disturbance or heterogeneity (e.g. Clark et al. 1982; Louton et al. 1996; Jansen 1997). At each of these scales, biodiversity can be conceptualized as a process responsive to particular biotic and abiotic factors rather than as a static attribute of a particular location. The role of biotic interactions in maintaining biodiversity in tropical ecosystems, then, can be elucidated by studies that show how biotic factors can singly or in combination, directly or indirectly, change biological diversity in those systems. Using a model system in Costa Rica we will highlight indirect trophic interactions that cause changes in biodiversity within a rainforest food web.

Whereas theoretical studies in ecology and evolution are often on the mark with respect to their appreciation of dynamic processes, their application in efforts to conserve biodiversity has been subject to shortcuts. Specifically, over the last century, many conservation efforts have focused on saving particular species at certain locations. Broader goals now target particular habitats and hotspots of endemism.

The generality of the role of trophic cascades in creating trophic structure is still in debate (Strong 1992; Polis et al. 2000; Holt 2000). In part this is because spatial (van Noughys & Hanski 2002) and temporal effects (Dyer & Coley 2002; Dyer & Letourneau 2003) make it difficult to pinpoint patterns in the impact of intertrophic effects and cascades on community structure. For example, predators may exert effects at some times but not others (Sinclair 2003), and in some locations but not others (van Noughys & Hanski 2002). Sinclair (2003) and Sinclair and Krebs (2002), working with vertebrates (hares, lynx, predatory birds and others), found that predators regulated their prey populations at high but not low densities in the boreal forest of Canada, and that this effect did not cascade down to plant populations.

To date, support for the existence of trophic cascades and their role in influencing trophic structure comes primarily from biologically simple ecosystems with two or three interacting organisms and trophic levels (see the recent volume by Tscharntke and Hawkins 2002). Although there are suggestions in the literature (Tscharntke & Hawkins 2002) that it will be difficult to find evidence in support of strong trophic cascades in high-biodiversity terrestrial systems with complex food webs, Terborgh and co-workers have found correlations that support their occurrence in predator–herbivore–plant systems involving both vertebrates and invertebrates (Terborgh 1992, Terborgh et al. 2001).

Introduction

The advent of the Homogeocene (Putz 1997), Homogecene (McKinney & Lockwood 1999; Lockwood & McKinney 2001) or Homogocene (Lodge & Shrader-Frechette 2003) – the era of human domination of the world – is both a challenge and an opportunity to test the ingenuity of humans. Will we be able to establish a new and sustainable balance with the rest of the world's biota? To do so requires active management of biodiversity based on understanding the function and dynamics of ecosystems. Appropriately, the approach to the study of the biota is undergoing a shift from a taxonomic, distributional and evolutionary focus, to a paradigm that considers biodiversity and ecosystem function (Naeem 2002). This new approach is holistic and quantitative, and helpful in understanding the role of biodiversity in the Homogeocene (Lugo 1995, 2002a).

Much of the current literature on tropical biodiversity focuses on the negative effects of non-indigenous or alien species, particularly invasive ones.

Introduction

Some interactions between plants are uniquely conspicuous elements of certain tropical forests; the giant lianas that wend through the canopy and the epiphyte-laden branches of cloud forests are striking examples. Nevertheless, the fundamental processes involved are no different from those in extra-tropical communities, even though diverse, sometimes uniquely tropical, mechanisms may be involved. An individual of one plant species interacting with an individual of a second plant species can lead to any of the same five outcomes at any latitude, and these consist of all combinations of negative, positive and neutral effects (except the non-interaction described by the mutually neutral interaction, 0/0). But interactions among plants in forests seldom involve such simple one-on-one relationships. More commonly, multiple players are involved and the interactions change with time: the liana binds crowns of several trees, the fallen palm frond damages multiple seedlings, and the solum is shared by roots of many species. Furthermore, positive and negative interactions occur simultaneously, so the observer sees only an integrated net effect of multiple interactions (Holmgren et al. 1997).

Most symbiotic (mutually positive) interactions in tropical forests involve relationships between plants and animals or between plants and microbes – fungi, bacteria, algae – described elsewhere in this volume. What, if anything, distinguishes plant–plant interactions from plant–microbe, plant–animal or animal–animal interactions? Even though they employ different biotic services (pollination, dispersal; see Ghazoul, Chapter 10, this volume; Muller-Landau & Hardesty, Chapter 11, this volume), higher plants (with a handful of exceptions) all use the same abiotic resources: water, carbon dioxide, photosynthetically active solar radiation and the same suite of 13 mineral elements.

Introduction

Early observations of the remarkable diversity in life history and morphology exhibited by tropical plants have strongly influenced our views on how these species coexist. Initially plants were classified according to size and life form, and subsequently into ecological species groups that recognized the importance of variation in light availability associated with the forest growth cycle (Richards 1952; Swaine & Whitmore 1988; Burslem & Swaine 2002). These ecological species groups reflect the existence of adaptive strategies that trees adopt during regeneration, and are the consequence of unavoidable trade-offs among suites of traits influencing growth, survival and fecundity.

In tropical forests, light availability has typically been identified as the primary limiting condition to growth. Adaptive strategies influencing the colonization of canopy openings and the capture or use of light are manifested as combinations of traits influencing dispersal, germination, seedling establishment, and allocation of resources to growth, storage and defence (van Steenis 1958; Budowski 1965; Whitmore 1975; Bazzaz & Pickett 1980; Coley et al. 1985; Kitajima 1994). This review addresses two issues: first, the extent to which individual trade-offs linking these traits can lead to the differentiation of tree species accounting for interspecific differences in the requirements for light and other resources; and second, the extent to which the action of multiple opposing trade-offs has a contrary effect of equalizing overall recruitment success so that no particular trait combination provides a recruitment advantage.