The eyespot colour pattern on butterfly wings is specified in the earl
y pupal epidermis by signals from a central ''focus.'' In Bicyclus any
nana we show that a small square of focal epidermis, grafted to a rang
e of distal wing sites, induces eyespot formation in surrounding host
tissue. Signaling is limited to the focus, and even an adjacent (paraf
ocal) graft does not maintain its normal fate (of contributing to the
eyespot) and does not influence its surroundings. Along the wing, ther
e is an abrupt change in the epidermis, as a focus grafted to a proxim
al site provokes no host response. The results of several grafting exp
eriments demonstrate that their different response properties are auto
nomous to small areas of the distal and proximal epidermis and that th
e nonresponding proximal tissue can nonetheless transmit the focal sig
nal. The Bicyclus dorsal forewing has a small anterior and a large pos
terior eyespot, and we show that this results mainly from a difference
in focal signals, not in the epidermal response. A grafted posterior
focus induces a large eyespot, whereas an anterior focus induces a sma
ll eyespot. Furthermore, the anterior and posterior eyespots differ in
proportions, and this difference also depends on the identity of the
focus, not on the responding epidermis. Eyespots are specified over ma
ny cell diameters from the focus by a mechanism which could consist of
one long-range signal, such as a morphogen gradient or of a cascade o
f short-range interactions initiated by the focus. Focal control of th
e difference in size and proportion between the anterior and posterior
eyespot is more readily compatible with a gradient mechanism. Neither
model, however, readily explains why the pattern induced by a grafted
focus is smaller, but its peripheral gold annulus is broader than in
the corresponding control eyespot. Also, there is no direct evidence f
or long-range gradients, in the butterfly wing or any other insect epi
thelium. (C) 1995 Academic Press,Inc.