Seedling Density Dependence Promotes Coexistence Of Bornean Rain Forest Trees - Statistical Data Included

Ecology, Sept, 1999 by Campbell O. Webb, David R. Peart

INTRODUCTION

Elucidating mechanisms that promote coexistence is of special interest in tropical rain forests because of their extraordinary species richness. A fundamental question is whether there are ecological mechanisms that buffer the many rare species against local extinction, or limit the dominance of the more abundant species. Any such mechanisms are, by definition, density dependent. Our goal in this paper is to assess evidence for density-dependent dynamics that promote species persistence in a rain forest community.

High species richness does not necessarily imply that density-dependent mechanisms exist. Species-rich assemblages could, in theory, result from the random process of community drift, with rates of local extinction balanced by dispersal and speciation rates (MacArthur and Wilson 1967, Hubbell 1979, 1997), especially if there are processes that slow the rate of random species exclusion (Hutchinson 1961, Hurtt and Pacala 1995). However, models that predict long-term species persistence (sensu Connell and Sousa 1983), rather than eventual random-walk extinction, must contain density dependence, either explicitly (Janzen 1970, Connell 1971) or implicitly, e.g., due to resource limitation (Tilman 1982, 1988) or recruitment fluctuation (Chesson and Warner 1981).

Analyses of density dependence relevant to coexistence in rain forests must take account of the complexities associated with high tree species richness, spatial scaling, and the great range in individual sizes. We argue that the appropriate emphasis is on analyses that: (1) compare common and rare species in a multispecies statistical model, (2) treat as large a spatial scale as possible, and (3) use adult abundance as the independent variable. We further suggest that measures of response at the early life stages are likely to be most informative; in this study we focus on seedlings.

Inclusion of rare tree species is vital in analyses of density dependence, not only because most species are rare (Whitmore 1984, Richards 1996), but also because their populations are, by definition, closest to local extinction. Yet, even the very largest data sets available are limited. With a sample of [approximately]240 000 trees, Wills et al. (1997) were able to test for density dependence in single-species analyses only for the most abundant 84 species out of 300. For each species analyzed, there must be adequate variation in density and adequate samples of individual response for statistical analysis. A demonstration that the abundant species experience negative density dependence does not imply that rare species' populations will be maintained. Hubbell (1980) demonstrated that local density dependence in abundant species (via distance dependence) could explain only a small fraction of the observed forest species diversity.

A means of including rare species and comparing their responses directly with those of more common species was proposed by Connell et al. (1984). If species differ in abundance, but all share a similar density-dependent response to population abundance, a community-level "compensatory" trend (Connell 1978) will emerge: species' population growth rates will decline as a function of their abundances. Hereafter, we will refer to this as a community-level compensatory trend (CCT). We suggest that tests of population-level density dependence (which are also important for assessing scale effects) augment, but do not replace, community-level analyses.

Tests of density dependence require a measure of population abundance as the independent variable. Individual genets range from undispersed seeds to large canopy trees. Proximally, seed mortality could depend on seed density, and seedling mortality on seedling density. Alternatively, performance of a life stage may be influenced by density at another stage; e.g., seedlings may experience more herbivore damage when there are more conspecific canopy trees nearby (Janzen 1970, Shaw 1974). Thus, various life stages and spatial scales may be appropriate for identifying the agents responsible for density dependence (if any) and the mechanisms by which those agents act. However, we suggest that to evaluate the implications for coexistence, the scale of mechanisms is not primary. Indeed, density-dependent mechanisms can occur without promoting coexistence. Consider, for example, local aggregations of seedling progeny around widely scattered adults, with seedling survival being negatively density dependent on local seedling density. Such density dependence does not translate into density-dependent control on adult abundance or subsequent seed production, and thus may not act to promote coexistence of adults in the canopy. This hypothetical seedling density dependence could promote species coexistence, however, if seedling aggregations around different parents were to overlap, with seedling survival thereby becoming dependent on local adult density. Thus, although ecological interactions can lead to complex combinations of effects of size class and spatial scale (Schupp 1992), there are compelling reasons to focus mainly on adults for measures of population abundance. Further, it is appropriate to obtain these estimates of adult abundance on the largest practical spatial scale in the community of inference, i.e., approaching the scale on which we choose to record species as present in the community, and consider them rare or common.