Eucalyptus improvement was initiated 20 years ago in Madagascar and ai
med at selecting species and provenances adapted to the very diverse c
limatic conditions of the island. The present paper reviews the result
s obtained by interspecific, provenance, and progeny tests. The good a
daptation of Eucalyptus grandis and Eucalyptus robusta was confirmed b
y various experiments. Malagasy provenances of both species exhibited
poorer performance than those recently introduced from Australia. This
is thought to be the result of a narrow genetic base and inbreeding.
The results obtained from tests conducted in diverse environments indi
cate that E. grandis provenances are already well performing but exhib
it limited variation. Provenances and progenies performances appear to
be quite irregular and unpredictable on the basis of site as shown by
the results presented here. Significant genetic differences exist amo
ng provenances of E robusta tested for growth parameters. The analysis
of variance revealed that genetic gain in volume production is possib
le through provenance selection. However, it is suggested to introduce
a wider range of provenances of E. grandis and E. robusta for further
evaluation and to select plus-trees within provenance rather than sea
rch for superior provenance. Tremendous variation in growth rate among
trees offers great opportunity for further genetic improvement. It ap
pears that there are remarkable provenance x site and progeny x site i
nteractions. Future reafforested areas need to be planted with genetic
ally improved material. Smallholder's ideotypes require that priority
should be given to a few major selection criteria: growth rate and vol
ume production, juvenile performance, coppicing ability and fire resis
tance. The priorities of the Madagascar genetic improvement programme
are to pursue provenance introduction and progeny evaluation and to pr
oduce genetically improved material to be distributed on a large scale
to village communities. It is suggested that these priorities be comb
ined for cost efficiency. Evaluation of different provenances and prog
enies in diverse ecological areas, individual selection, and large-sca
le production of genotypes with a broad base is presently conducted in
breeding seedling orchards to satisfy national needs for Eucalyptus s
pp. seeds using synthetic varieties.