374 Chris J. Peterson and Walter P. Carson
13 months of monitoring. Holl (1999) found that
seeds of animal-dispersed woody species arrived
in open pasture at a density of only 3 m−^2 year−^1
and only one genus (Solanum) had seed input into
seed traps more than 5 m from forest. Cubina
and Aide (2001) found that only three species
and 0.3% of seeds dispersed more than 4 m
from the forest edge. Those seeds that do arrive
in pastures are generally small-seeded pioneer
species that are dispersed by wind (e.g.,Trema
micrantha)orbats(e.g.,Cecropiaspp.) (Martinez-
Garza and Gonzalez-Montagut 1999, 2002). Ingle
(2003) studied the seed rain into successional
land adjacent to lower montane rainforest in the
Philippines. Wind-dispersed seeds were 15 times
more common than vertebrate-dispersed seeds
and more than 95% of the total seed rain was
within 40 m of the forest edge. Seed input appears
to be high only in very small pastures surrounded
by forest (e.g., Zahawi and Augspurger 1999,
WijdevenandKuzee2000,Myster2004)orwhere
seeds originate from numerous remnant trees
growing within the pasture (Guevara and Laborde
1993, Slocum and Horvitz 2000). Overall, woody
colonist density and diversity decays rapidly with
distance from forest edge and this likely reflects
very low seed input (Peterson and Haines 2000,
Chinea 2002, Myster 2003a), although excep-
tions exist (Aideet al. 1996, Duncan and Duncan
2000).
Recruitmentlimitationcausedbyseed
predationRates of seed predation as measured via seed
removal studies are typically high in early suc-
cessional communities. For example, in Costa
Rican pastures Holl and Lulow (1997) found
that 66% of all seeds were removed within the
first 30 days for 10 different woody species. In
the eastern Amazon, Nepstadet al. (1996) doc-
umented more than 80% seed removal within
20 days for 6 of 11 woody species. Duncan and
Duncan (2000) estimated that nearly 50% of
seeds of six species were removed within 11 days
at a site near Kibale National Park in Uganda.
Removal rates in these early successional com-
munities are typically greater than those found in
nearby intact forest or in treefall gaps (Nepstad
et al. 1996). It is especially important to note
that removal rates typically decrease as seed size
increases (Figure 22.1; Osunkoya 1994, Nepstad
etal. 1996, Sarmiento 1997, Duncan and Duncan
2000, Joneset al. 2003, Myster 2003b). When
Zimmermanet al. (2000) placed the very small
seeds ofCecropia schreberianain Puerto Rican pas-
ture, 100% of seeds were removed within 8 hours.
Thismaysuggestthatlargeseedshaveasizerefuge
from the smaller seed predators that are typically
present in pastures (Figure 22.1), and that in the
rare instances when large seeds arrive in pastures,
they may escape predation. Overall, most studiesSeed mass (g)1 e− 41 e− 31 e− 21 e− 11 e+ 01 e+ 1% Remaining020406080100
Holl (2002)
Myster – PR (2004)
Myster – EC (2004)
Jones et al. (2003)
Nepstad et al. (1996)
Duncan and Duncan (2000)
y = 35.81 + 5.66x Figure 22.1 Percent of original
number of seeds remainin gversus
seed mass (g), for 21 species, pooled
across five studies (r^2 =0.41,
P<0.001). From Nepstadet al.
(1996), Duncan and Duncan (2000),
Holl (2002), Joneset al.(2003), and
Myster (2004; PR, Puerto Rico; EC,
Ecuador). Four of the five within-study
regressions of seeds remaining versus
seed mass were also significant.