We investigate which hypothesis, the "better mate hypothesis" or the "bette
r territory hypothesis" best explains the unusually high divorce rate (59%)
in a population of blue tits (Parus caeruleus) living in a sclerophyllous
habitat characterised by severe environmental constraints (trophic, parasit
ic, climatic) on the island of Corsica, France. Using data from the breedin
g seasons 1985-1998 and from a brood size experiment (1990-1993) we examine
d the causes of divorce and their consequences on breeding performance, mat
e assortment and territory choice. Breeding performance had no significant
effect on whether birds re-united or divorced in the next breeding season.
Re-uniting pairs did better than divorced females and the latter improved t
heir breeding performance compared to prior to divorce, but this was mainly
due to age and territory effects. There were no differences in male perfor
mance depending on whether they re-united or divorced. The age combination
of pairs did not differ between re-uniting and divorcing pairs, but mate as
sortment changed after divorce with males re-mating more often with older p
artners than females. Manipulation of brood size showed a trend for birds w
ith enlarged broods to divorce more. Pairs responded significantly to terri
tory quality by divorcing more often in poor than in good breeding sites. B
oth faithful pairs and male divorcees had shorter breeding dispersal distan
ces than female divorcees. Divorce rates were determined by the large diffe
rences in quality among breeding sites. Males, whatever their status, usual
ly retained their previous territory whereas divorced females moved signifi
cantly longer distances and improved their breeding site. Moving to a bette
r territory after divorce benefits only females which appear to be the choo
sing sex in the decision to divorce. This study strongly supports the "habi
tat mediated hypothesis" and we suggest that the large observed intraspecif
ic variation in the magnitude of divorce rates in many species of birds is
mostly determined by habitat characteristics.