Leaf phenology was monitored for 49 woody species (trees and tall shru
bs) each month over a 2.5-year period in a humid, wet-dry tropical euc
alypt savanna at Solar Village, near Darwin, Australia. In the 10 most
common species, which spanned the range of phenological types, phenol
ogy was monitored every two weeks. To investigate the relationships be
tween leaf phenology and plant water status, pre-dawn leaf water poten
tial was monitored in eight common species every 4-6 weeks, Four main
phenological types were described: (1) evergreen species, which retain
ed full canopy throughout the year; (2) brevi- or partly deciduous spe
cies, in which the amount of canopy fell significantly, but briefly, d
uring at least one dry season during the study period, but to levels n
ot below 50% of full canopy; (3) semideciduous species in which canopy
fell to below 50% of full canopy in each of the dry seasons; and (4)
fully deciduous species, which lost all leaves during the early-mid dr
y season, and remained leafless for at least one month. Of these 49 sp
ecies, 24% were evergreen, 20% were brevideciduous, 29% were semidecid
uous, and 27% were fully deciduous. Leaf fall occurred 1-2 months earl
ier in the dry season for the fully deciduous species than for the sem
ideciduous species. Leaf fall in all species was coincident with the a
ttainment of seasonal minima in leaf water potential, which were, on a
verage, about -1.5 to -2.0 MPa in the evergreen and semideciduous spec
ies, compared with about -0.5 to -1.0 MPa in the fully deciduous speci
es. Leaf flushing occurred throughout the dry season in the two evergr
een species, with a major peak in the late dry season. In five semidec
iduous species and one of the fully deciduous species, leaf flushing c
ommenced in the late dry season prior to the occurrence of any rain. I
n two fully deciduous species, leaf flushing occurred only after the f
irst storms of the early wet season. There was variation in the timing
of flushing, both between species within years and between years for
some species. However, all species commenced leaf flushing after water
potentials rose, following the attainment of seasonal minima in pre-d
awn leaf water potential. Soil moisture at 1 m did not fall below perm
anent wilting point during the dry season; hence, reserves of soil wat
er at the end of the dry season were sufficient to support the whole-p
lant rehydration that preceded leaf flushing in the absence of rain. T
hese results are consistent with hypotheses, developed in the neotropi
cs, that leaf phenology in trees from the wet-dry tropics is largely c
ontrolled by endogenous mechanisms.