Pollinators,
such
as bees, exploit patchily distributed food resources that replenish
over time. Finding
an efficient route to visit multiple flowers and return to the nest is
a
complex optimisation task analogous to the well-known Traveling
Salesman
Problem in graph theory. Here I will present experiments on bumblebees
foraging
in meadows of computer-controlled flowers equipped with automated
tracking
systems to discuss how foragers locate flowers and often develop
near-optimal circuits
to link them as they gain experience. Comparative analyses of our
behavioural
data across experimental conditions reveal simple search patterns and
learning
heuristics that bees, and virtually all nectar foraging animals, may
use to efficiently
exploit their complex foraging environments with brains of limited
computational
power.