and Busby 1987, Feinsinger et al. 1988b). In a series
of elegant indoor experiments that controlled the se-
quence in which a bird visited flowers of a single
species (no pollinator-sharing) or of two species (pol-
linator-sharing), pollinator-sharing caused a decline
in pollen dispersal and receipt. The mechanism is
pollen loss; when a bird visits two species of flowers
in a single foraging bout, pollen of one species can be
lost to flowers of the other species. The same local
effect of neighborhood can occur in the field but dif-
fers substantially in degree from one species to
another (Feinsinger et al. 1991). Even though some
guildmates can negatively affect others' pollination
success, they may not exert strong selective pressure
(Feinsinger et al. 1986, 1992). Monteverde's bird-
pollinated plants do not consistently demonstrate
the predicted links among competitive mechanisms,
population effects, and community organization (Fein-
singer and Tiebout 1991).
Character displacement: predictions and tests. Plants
that share pollinators have been thought to benefit
from traits that reduce sharing. For example, plants
might be under strong selective pressure to flower at
different times, or to place pollen on different parts
of their pollinators, especially if counterselection
pressures are not strong. Their flowering periods or
pollen would then be "displaced" from one another.
The idea has come under lively debate (e.g., Poole and
Rathcke 1979, Wheelwright 1985a), and research in
Monteverde has contributed substantially to it.
Murray et al. (1987) showed that there is potential
for competition via pollinator-sharing in the two
guilds of shrubs and epiphytes that are pollinated by
hummingbirds, because birds move among plant spe-
cies as they forage, carry more than one species of
pollen at a time, and can transfer pollen among
species (essentially wasting pollen; Feinsinger et al.
1991). Despite the potential for competition, however,
phenological displacement of flowering did not occur
in either guild at Monteverde. Using long-term quanti-
tative records of flowering periods and rigorous tests
for phenological differences (sensu Poole and Ratchke
1979, Fleming and Partridge 1984), Murray et al.
(1987) found no evidence of temporal organization.
They also found no evidence of the alternative evo-
lutionary divergence in pollen placement: fine-scale
measurements of the length and positions of the flow-
ers' female and male reproductive structures did not
reveal greater morphological displacement among
plant species with similar flowering seasons than
among those with very different ones (Murray et al.
1987).
Why don't flowering phenologies and floral mor-
phologies diverge? The reasons do not appear to in-
clude seasonal, fruiting, or phylogenetic constraints
within these guilds. Although competition for pollina-
tion may occur (at the level of the local flower neigh-
borhood), it is limited and sporadic (Feinsinger et al.
1986, 1992). Similar conclusions were drawn for 23
species of lauraceous trees that use the same birds as
seed dispersers and may share some of their small
insect pollinators (Wheelwright 1985b; see Wheel-
wright, "A Hypothesis," pp. 281-282). Flowering
times are not displaced, and fruiting periods are ag-
gregated rather than separated. These studies illus-
trate that sharing the services of dispersers does not
necessarily result in phenological or morphological
character displacement (Feinsinger 1987).
Phenological measurements on a fine scale can also
be used to assess the potential for pollinator- sharing.
Koptur (1983) documented temporal flowering pat-
terns among individuals, populations, species, and of
individual flowers in seven Inga (Fabaceae) tree spe-
cies. Observations suggested significant pollinator-
sharing, but fine-scale phenological measurements
revealed that it was low. All three studies suggest that
for guildmates, plant-plant competition via pollina-
tor sharing cannot be assumed. Neither interactions
nor phenological patterns can be studied as isolated
phenomena, particularly in species-rich guilds char-
acteristic of Monteverde.8.1.5. Pollination Patterns
A plant's reproductive success depends in part on
how many compatible pollen grains (per ovule) its
stigmas accumulate. Inferring the size and purity of
pollen loads from information such as frequency of
visits by potential pollinators can be misleading. Fur-
thermore, the notion that reproductive success is
highly correlated with the size and purity of the pol-
len load received may be simplistic. Large numbers
of compatible pollen grains may not always be opti-
mal for pollination (Young and Young 1992), and
some Monteverde species are unaffected by hetero-
specific grains mixed with compatible grains (S. Kins-
man, unpubl. data). Documenting pollination limi-
tation can be difficult and requires experimental
techniques (Dafni 1992, Young and Young 1992,
Kearns and Inouye 1993). The role of pollination limi-
tation in a plant's reproductive success must be de-
termined to pinpoint whether pollinator-sharing fos-
ters competition or facilitation among plant species
(Rathcke 1983, Feinsinger 1987).
Many plants' seed-set or fruit-set may be limited
by pollination. If their pollinators delivered more (or
more diverse) pollen, the plants could develop more
(or genetically superior) seeds (Burd 1994). Pollina-
tion limitation of plant reproductive success may be254 Plant-Animal Interactions