In this paper, we examine how ecological costs of resistance might be manif
ested through plant relationships with pollinators. If defensive compounds
are incorporated into floral structures or if they are sufficiently costly
that fewer rewards are offered to pollinators, pollinators may discriminate
against more defended plants. Here we consider whether directional selecti
on for increased resistance to herbivores could be constrained by opposing
selection through pollinator discrimination against more defended plants. W
e used artificial selection to create two populations of Brassica rapa plan
ts that had high and low myrosinase concentrations and, consequently, high
and low resistance to flea beetle herbivores. We measured changes in floral
characters of plants in both damaged and undamaged states from these popul
ations with different resistances to flea beetle attack. We also measured p
ollinator visitation to plants, including numbers of pollinators and measur
es of visit quality (numbers of flowers visited and time spent per flower).
Damage from herbivores resulted in reduced petal size, as did selection fo
r high resistance to herbivores later in the plant lifetime. In addition, f
loral display (number of open flowers) was also altered by an interaction b
etween these two effects. Changes in floral traits translated into overall
greater use of low-resistance, undamaged plants based on total amount of ti
me pollinators spent foraging on plants. Total numbers of pollinators attra
cted to plants did not differ among treatments; however, pollinators spent
significantly more time per flower on plants from the low-resistance popula
tion and tended to visit more flowers on these plants as well. Previous wor
k by other investigators on the same pollinator taxa has shown that longer
visit times are associated with greater male and female plant fitness. Beca
use initial numbers of pollinators did not differ between selection regimes
, palatability and/or amount of rewards offered by high- and low-resistance
populations are likely to be responsible for these patterns. During period
s of pollinator limitation, less defended plants may have a selective advan
tage and pollinator preferences may mediate directional selection imposed b
y herbivores. In addition, if pollinator preferences limit seed set in high
ly defended plants, then lower seed set previously attributed to allocation
costs of defense may also reflect greater pollinator limitation in these p
lants relative to less defended plants.