For close to a century, recruitment of larvae to a local population has bee
n widely accepted as a primary determinant of marine population dynamics(1,
2). However, progress in elucidating the causes of recruitment variability
has been greatly impeded by our ignorance of the sources of recruits. Altho
ugh it is often assumed that recruitment is independent of local reproducti
on(3-6), there is increasing circumstantial evidence that physical(7,8) and
behavioural(9,10) mechanisms could facilitate larval retention near source
populations. To develop a direct method for reconstructing the dispersal h
istory of recruiting larvae, we put forward the hypothesis that differences
in nutrient and trace-element concentrations between coastal and open ocea
ns could result in quantifiable differences in growth rate and elemental co
mposition between larvae developing in coastal waters (locally retained) an
d larvae developing in open ocean waters (produced in distant locations). U
sing this method, we show that recruitment to an island population of a wid
ely distributed coral-reef fish may often result from local retention on le
eward reefs. This result has implications for fisheries management and mari
ne reserve design, because rates of dispersal between marine populations-an
d thus recruitment to exploited populations-could be much lower than curren
tly assumed.