Constraints on Succession 373examinin gthe models of Pickett (1976), Noble
and Slatyer (1980),Tilman (1985, 1988), Huston
and Smith (1987), or Pickettet al. (1987). Three
studies evaluated Egler’s IFC model (Harcombe
1977, Myster 2003a, Pena-Claros 2003) and
three studied the role of remnant trees in pas-
tures and used the nucleation model as their
conceptual foundation (Hollet al. 2000, Slocum
2001, Carriereet al. 2002). In an approach sim-
ilar to that of Egler (1954), Finegan (1996) and
Gómez-Pompa and Vásquez-Yanes (1981) pre-
sented models where most of the species from all
successional stages establish very early in succes-
sion (for a test of these models see Pena-Claros
2003).Nearlyadozenpapersmentionedtheinter-
action categories of Connell and Slatyer (1977)
though these studies were not designed a priori to
distinguish among these categories. Furthermore,
a search of citations in tropical post-agricultural
succession papers on Web of Science showed only
one citation of Tilman’s work (Ganade and Brown
2002), one of Huston and Smith (Nepstadet al.
1996), and none of Pickett, or Noble and Slatyer.
Overall, few studies were undertaken to test gen-
eral conceptual models and most studies were
not theory driven. Most researchers report that
their primary motivation was either to identify
the processes limitin gwoody species establish-
ment early in succession (e.g., seed dispersal,
germination, predation) or to add to the existing
case studies of succession in the tropics and to
aid restoration efforts. As a result, several broad
generalizations have emerged regarding the con-
straints on woody species establishment in tropical
old fields and pastures (e.g., Hollet al. 2000).
Nonetheless, it was clear that no widely adopted
theory is currently central to studies of tropical
post-agriculturalsuccessionregardlessof whether
these theories were developed in temperate or
tropical regions.THE TWO MAJOR CONSTRAINTS ON
TROPICAL POST-AGRICULTURAL
SUCCESSION
Two major constraints appear to strongly influ-
ence the pattern of early succession in tropical
habitats: recruitment limitation and inhibition
by resident vegetation, particularly perennial
graminoids that are present at the time of
abandonment.Recruitment limitationRecruitmentlimitationcausedbyasparse
seedbankVery few seeds of forest species are present in the
soil seed bank of old fields and abandoned pas-
tures (Uhlet al. 1982, Zahawi and Augspurger
1999, Wijdeven and Kuzee 2000, Zimmerman
etal. 2000, Cubina and Aide 2001, Slocum 2001,
Myster 2004). For example, the seed bank density
of woody species was nearly five times greater in
intact forest than in abandoned pastures in east-
ern Amazonia (679 m−^2 versus 144 m−^2 ), and
viable seeds of primary forest tree species were
essentially absent (Nepstadet al. 1996). Similarly,
in southern Mexico, Guevaraet al. (2004) found
that primary forest tree species made up only
0.3% of the soil seed bank around remnant figs
in pastures.Thus it seems unlikely that succession
typically starts with woody species establishing,
in even small numbers, from a seed bank.Recruitmentlimitationcausedbylow
seedrain
Seed rain into pastures is typically low, concen-
trated near forest edges, and enhanced when fields
have structural complexity. Although there are
few studies on the behavior of animal disper-
sal vectors (predominantly birds and bats), avian
movements appear to be more common in pas-
tures that have scattered shrubs and trees versus
thosedevoidof structuralcomplexity.Avianmove-
ments typically occur within 80 m of the forest
edge (Sisk 1991, Da Silvaet al. 1996, but see
Puyravaud 2003). Seed rain into abandoned agri-
cultural lands (independent of remnant trees)
declines rapidly across short distances from the
forest edge (e.g., Martinez-Garza and Gonzalez-
Montagut 2002). Doschet al. (2007) found that
45 m from a forest edge seed rain in five pastures
averaged less than 1% of that at the forest–pasture
interface. No species with seeds larger than 1 mm
arrived more than 10 m from the forest during