304 Tad C. Theimer and Catherine A. Gehring
a site will successfully establish there. Factors
that increase dispersal limitation could potentially
maintain diversity by allowing inferior competi-
tors to survive in some areas where superior
competitors never successfully establish (Tilman
1994, Hurt and Pacala 1995, Hubbell 1997,
Tilman 1999). However, if seedling densities are
consistentlybelowlevelsnecessaryforstrongcom-
petitive effects, dispersal limitation similar to that
we documented could simply reduce local species
richness. Our results support Wright’s (2002)
hypothesis that in many cases vertebrates main-
tain seedling densities at levels that preclude
strong competitive effects. For example, even in
the exclosure plot with highest seedling density,
we detected no reduction in recruitment or sur-
vivalbythefourthyearafterexclosure,aswouldbe
expected if strong competitive interactions were at
play. Our results also underscore that even in sys-
tems in which seed arrival may not limit dispersal
(e.g., Webb and Peart 2001), terrestrial verte-
brates could act to increase dispersal limitation
through their effects on seeds and seedlings.
Fourth,indirecteffectsof terrestrialvertebrates,
like that of increasing AMF spore species rich-
ness, could potentially alter seedling community
dynamics, but these effects may take considerable
timetobeexpressed.Theimportanceof AMFspore
dispersal to rainforest seedling establishment and
survival remains poorly understood, but there is
growing evidence from other types of plant com-
munities that some species are of greater benefit
to a given plant species than others (e.g., van der
Heijdenet al. 1998a, Bever 2002, Klironomos
2003) and that AMF diversity may contribute
to plant diversity (van der Heijdenet al. 1998b).
Limited distributions of some fungal species in
the absence of vertebrates thus may alter plant
communities. Similarly, Manganet al. (2004)
hypothesized that differences in the AMF com-
munities of island and mainland sites in tropical
Central America could contribute to the differ-
encesinplantcommunitiesobservedatthosesites.
Lower AMF inoculum potential in the absence of
vertebrate dispersers, as we observed, also may
favor seedling species that do not form associa-
tions with AMF or that do not depend on them for
growth or nutrient acquisition, leading to changes
in plant communities and potentially further
reducing the populations of mycorrhizal fungi.
For example, we observed that one third of the
common seedling species on our rainforest study
site were never or rarely observed to form asso-
ciations with AMF or other types of mycorrhizal
fungi (Gehring and Connell 2006). These non-
mycorrhizal seedling species would be expected to
becomemorecommonintheabsenceof terrestrial
vertebrates, potentially leading to further reduc-
tions in fungal diversity and inoculum potential.
Although these seedling species showed similar
abundance in exclosure and open plots when AMF
spore data were collected in 2001, a prelimi-
nary analysis of these seedling species in 2003
showed that the percentage of seedlings of non-
mycorrhizal plant species was marginally higher
in exclosure plots than in open plots (mean for
open plots=34.8±4.5 SE [standard error] and
for exclosure plots=41.8±5.8 SE,t=1.418,
P=0.08). The diversity of microsites available
for plant establishment has been recognized as
an important factor maintaining tree diversity
(e.g., Grubb 1977, Denslow 1980). We suggest
that the effects of terrestrial vertebrates on the
presence, diversity, and species composition of
mycorrhizal fungal inoculum could be a relatively
unexplored, but potentially important, aspect of
this microsite diversity.SUGGESTIONS FOR FUTURE
RESEARCH
We suggest three avenues of research that would
be fruitful in testing the ideas presented here
and in expanding our knowledge of the poten-
tial interaction of vertebrates, seedlings, and AMF
spores. First, comparison of seedling and spore
communities that differ both in the proportion
of species dependent on vertebrates for dispersing
propagules and in the susceptibility of propag-
ules to vertebrate mortality would elucidate how
different communities may respond to vertebrate
loss or addition. In the case of tree seedlings,
such studies should distinguish between verte-
brates that act primarily as seed dispersal agents
that potentially reduce dispersal limitation, versus
those that act as agents of seed and seedling mor-
tality and thereby increase it. Determining how