I tested three hypotheses that could explain variation in the timing o
f breeding in populations of House Wrens (Troglodytes aedon) at four s
ites in Costa Rica. The sites were located at 200-1,500 m elevation on
both sides and on top of the central mountain range, and had climates
differing in temperature, the severity of the dry season, and total r
ainfall. For the first hypothesis, that breeding is timed to coincide
with peaks in food availability, I monitored the monthly abundance of
arthropod prey and wren clutch initiations at the four sites. Cross-co
rrelation analysis showed that at three sites, wrens initiated clutche
s several months prior to when prey levels were high. Indeed, breeding
began when prey levels were at their annual low. At the fourth site,
prey levels varied little throughout the year and House Wrens nested n
early year round. These results indicate that sufficient food to produ
ce eggs or feed nestlings may have been available throughout the year.
Breeding appeared to be timed so that juvenile dispersal and molt occ
urred when food was most plentiful. The second hypothesis, that breedi
ng is timed to avoid seasons when nest predation is high, was not supp
orted because the rate of nest predation did not vary temporally. The
third hypothesis, that breeding is timed to avoid climatic events that
can increase the physiological costs of reproduction, was not support
ed at the three lower elevation sites. Clutch initiation at the highes
t site, however, did not commence until the early dry season wind and
mist subsided. The termination of breeding was not correlated with cli
matic changes at any of the sites. Thus reproduction in tropical House
Wrens seems generally to be timed to facilitate post-breeding activit
ies, not activities associated with nesting itself.