380 Chris J. Peterson and Walter P. Carsonvegetation present in the pasture. Hypotheses
should be developed regarding which life-history
traits or functional groups are likely to be facil-
itated or inhibited by varyin gamounts of resid-
ual vegetation and then experiments conducted
that vary the relevant factors (e.g., propagule
number, resident biomass, functional group, lit-
ter mass). Overall, we suggest that a moderate
amount of extant vegetation or litter may serve to
facilitate germination and seedling survivorship
by amelioratin glocal microclimatic conditions.
As this vegetation or litter becomes very dense,
it will likely become inimical to woody species
recruitment (e.g., Carson and Peterson 1990).
This may be why microsites that are devoid of
grasses but have favorable microclimates (rotting
logs) are areas of relatively dense woody seedling
recruitment.
3 The degree that resident trees facilitate woody
species recruitment.If the nucleation model applies
broadly, then resident trees and shrubs should
typically promote the establishment of woody
vegetation in their proximity and also result in the
spread of woody vegetation around these recruit-
ment foci. Studies need to be conducted across
a range of both abiotic (e.g., soil moisture, fer-
tility) and biotic (e.g., graminoid abundance and
composition) conditions to test the robustness of
this model. These studies need to be combined
with studies that manipulate the number of trees,
structural complexity, or key microsites early in
succession to fully evaluate the degree to which
early succession in tropical systems is constrained
by variation in these putative recruitment foci.
4 Thespatialandtemporalvariabilityintheintensity
of predationonseedsandseedlings.Greater seed pre-
dation appears to occur on smaller-seeded species;
otherwise few generalizations have emerged due
to a paucity of studies. This is problematic in
terms of testin gmodels of succession. For exam-
ple, if seed predators congregate beneath trees
or shrubs (due to enhanced cover or elevated
food resources), seed predators could make areas
beneath trees and shrubs inimical to woody
species recruitment, in contrast to the predictions
of the nucleation model.
5 The correlations among life-history traits such
as seed size, growth rate, fecundity, and shade and
drought tolerance.These correlations have been
widely examined for temperate woody species
but less so for tropical species (but see Garwood
1983). Almost all of the major models assume
various correlations and trade-offs amon gplant
traits, and the validity of predictions of these mod-
els rests on the validity of these assumptions.
For example, the resource ratio model assumes
a negative correlation between competitive abil-
ity for light and competitive ability for nutrients
with relevant differences in allocation to shoots
versus roots.CONCLUSIONS
A substantial number of models or conceptual
frameworks have been developed to explain or
predict patterns or causes of species turnover in
early successional habitats in temperate regions.
We have presented a brief review of these mod-
els and found that most were limited in their
usefulness when applied to post-agricultural suc-
cession in the tropics. However, these models
have rarely been tested in tropical habitats and
at least one, the nucleation model (Yarranton
and Morrison 1974), shows real promise and
should be tested, expanded to incorporate quan-
titative models of dispersal, and revised where
appropriate. Another conceptual model, a hierar-
chical framework that provides an exhaustive list
of causes for species turnover durin gsuccession
(Pickettet al. 1987), was also considered highly
relevant for studies of succession in any habi-
tat. We suggest that future studies of succession
should be designed to test robust general models
that can then be refined and applied across larger
geographic regions.ACKNOWLEDGMENTS
We thank Brian McCarthy, Stefan Schnitzer,
Henry Stevens, and especially John Paul for com-
ments on earlier drafts of this manuscript. This
work was supported by NSF grant DEB 94-24606
to C.J.P. The NSF grant to Peterson and Haines
was named “Research toward sustainable land use
and biodiversity in a mosaic of agriculture and
tropical forest”.