Habitat specificity of two palm species: experimental transplantation in Amazonian successional forests

Ecology, Dec, 1995 by Elizabeth Losos

INTRODUCTION

Two hypotheses have been generated to explain why large-seeded species typically occur in late-seral and mature forests rather than in younger stands. The "dispersal limitation hypothesis" attributes this restricted habitat distribution to the ineffective seed dispersal of large-seeded species into newly formed habitats or gaps (Foster et al. 1986, Fleming 1988, Losos 1993, but see Heithaus 1982). Under this hypothesis, if these seeds were to arrive in early-seral zones, they would survive at least as well as small-seeded species (Foster et al. 1986). By contrast, the "postdispersal limitation hypothesis" postulates that other factors associated with seed size restrict these species to late-seral habitats. These factors fall into two groups. First, life history traits of large-seeded species, such as low seed set or short seed dormancy, may allow pioneers with high numbers of seeds or long seed dormancy to establish more successfully in newly created habitats (Harper et al. 1970, Foster 1986). Second, physical and biotic factors that vary across successional zones may create unfavorable conditions for large seeds in early-seral forests. These factors may include soil moisture, atmospheric humidity, seed predation, herbivory, or pathogens (Connell and Slatyer 1977, Shugart 1984, Foster 1986, Walker and Chapin 1986, Walker et al. 1986, Pickett et al. 1987, Losos 1993). Even if large seeds were to arrive in an early-seral zone, postdispersal factors could inhibit them from dominating until later in the successional process (Egler 1954).

Several studies have attempted to identify which of the many confounding factors most significantly influence seed size and habitat preference (Foster and Janson 1985, Mazer 1989). For the most part, they have employed statistical analyses to distinguish among associated traits (Foster and Janson 1985, Mazer 1989, Rydin and Borgegard 1991). Typically, these studies divide a community of plants into early- and late-seral species or pioneer and shade-tolerant species, and then partition the variation among species that is attributable to different characters. For example, Mazer (1989) showed that variation in seed mass among Indiana dune angiosperms could be explained by habitat, life history, taxonomic relationships, water and light, onset of flowering, and duration of flowering. Foster and Janson (1985) found that the seed mass of mature forest species in an Amazonian forest was higher than that of pioneers, even after statistically controlling for the effects of tree height, dispersal mode, and growth form.

These cross-species analyses comprise an important first step toward understanding interspecific variation in seed size, but, by themselves, can only suggest which factors may hamper large-seeded species from occurring in early-successional habitats. To further explore this issue, experimental manipulation is extremely helpful because it can test the performance of a particular species under a variety of conditions (Walker and Chapin 1986). One could experimentally examine the survivorship of late-successional species and early-successional species to determine in which seral zone the species survive best. After a range of species is analyzed, the patterns underlying the seed size/seral zone relationship should emerge. Transplant experiments between different successional zones are still relatively rare (Collins and Quinn 1982, Walker and Chapin 1986, Walker et al. 1986, Tilman 1988, Winn 1988). Rabinowitz (1978) carried out one of the few transplant experiments that explicitly tested whether large propagules were dispersal limited. Because mangrove species with large propagules survived better in habitats that were naturally dominated by other mangrove species with smaller propagules, she concluded that differential dispersal contributed heavily to mangrove zonation.

In this study, I use experimental manipulations to investigate the degree to which postdispersal limitation restricts the distribution of late-seral species. Two large-seeded palm species, Iriartea deltoidea and Astrocaryum murumuru var. javarense, both are largely absent from early- and mid-seral forest but are common in late-seral forest and abundant in mature forest (J. Terborgh, unpublished data). However, Iriartea is found occasionally in early- and mid-seral forest, whereas Astrocaryum is almost never found there (Losos 1993). It is possible that restricted seed dispersal has created this distribution pattern. In a related palm seed bank study, a low density of Iriartea seeds was found in early- and mid-seral forest compared to a much higher density in late-seral and mature forests where the species is abundant (Losos 1993). No Astrocaryum seeds were found in early- or mid-seral forests.

Given positive evidence for the dispersal limitation hypothesis, this study focuses on two questions dealing with the influence of postdispersal factors: 1) Is seedling survivorship throughout succession consistent with a species' natural habitat distribution? 2) If postdispersal limitation does play an important role in structuring the distribution of both Iriartea and Astrocaryum, what accounts for the presence of a low natural density of Iriartea seedlings and the absence of Astrocaryum seedlings in early- and mid-seral forest? To address these questions, I transplanted Iriartea and Astrocaryum seeds in four successional zones of an Amazonian floodplain forest to compare seedling survivorship across habitats. Equal or better survivorship in the early- or mid-seral forest would strongly suggest that dispersal plays a key role in limiting these species to late-seral and mature habitats. Alternatively, lower survivorship in early- and mid-seral forests would point to the importance of postdispersal factors in conjunction with dispersal limitation. The slight variation in natural distribution of the two species allows the opportunity to distinguish which postdispersal factors, if any, affect each palm.