The montane butterflies of the eastern Afrotropics - Biogeography and Ecology of the Rain Forests of Eastern Africa - page 145

the whole area from East Africa to South Africa.
No clue eists as to what could have been the
original habitat preference of the genus.

tradicts the hypothesised South African origin:
there are no recurrent patterns of sister-group
relationships between EAT and South Africa,
there are no taxa at the same ime widespread in
and endemic to EAT, and there is no concenra-
ion of montane grassland taxa in the southen
areas in EAT. The few South African connec-
ions found can indicate only limited jump disper-
sal, possibly by species with a slightly wider
ecological tolerance than most grassland species.
Interesingly, the pattem(s) found rather indicate
a geographic connection in the opposite direcion
- from the north..Even though only a limited
number oftaxa are involved we find here the pat-
ten predicted for a South Arican connecion
under the jump dispersal model: same species as
or sister-group relaionships with taxa outside
Africa, and the highest concenraion in the north
(Ethiopia).
Almost three quarters of the species do not
show close relaionships with taxa outside EAT.
Even though their sister-group relaionships are
badly known, we can safely assume that in most
cases the sister group is to be found in EAT.
These species agree with, or at least do not con-
tradict, the predicions of the intenal differeni-
aion model in which it is supposed that species
originated on the spot from species ith a dif-
ferent habitat preference, and later epanded
their range according to their dispersal powers,
leading to a mosaic of small, isolated ranges and
differently overlapping larger ranges. Under this
model coherence of areas of endemism is a func-
ion of dispersal rather than of former connec-
ions, since the latter would have led to a much
less 'chaoic' patten. The model will be further
discussed after the forest species have been dealt
with.
As stated above, the connecion with the
Cameroun highlands is apparently attributable to
a few species dispersing to the west. Since so fe w
species are involved and these species, because of
their also otherwise wide disribuions, possibly
have a larger ecological tolerance than the other
montane grassland species, they cannot be indica-
ive ofa former coninuous belt ofmontane grass-
land

Eicochysops. In view ofthe range in habitat prefer-
ences, the occurrence of six more species in the
study area and the possibility that the only two
a that are confined to montane grassland are
conspecific with each other as well as with taxa
with a wider ecological tolerance, the preference
for montane grassland is most likely a derived
character in the genus.

Euchysops. Since preference for montane grass-
land occurs in only three of the 23 species, and
nine more species of the genus occur in EAT, a
shift in habitat preference towards montane
grassland is not unlikely.

lepiochysops. As for Euchysops, apart for the
numbers being higher.

Antanartia. Since abyssinica is the only montane
grassland species and three other species of the
genus occur in the area it seems likely that the
habitat preference is a derived condiion.

Neocoenyra. The montane grassland species are
resricted to the Easten Arc mountains. The
eight other species of the genus occur in a variey
of other habitats at lower elevaions. The related
genus Pseuonympha, which is restricted to
southen Africa, has a number of montane grass-
land species apart from species in other open
habitats. Even if the two genera were sister groups
there is no reason for considering the preference
for montane grassland a primiive condition l
Neocoenyra.

sussion

The two pattens above were predicted based on
the altenaive hypotheses that the montane grass-
land fauna ofEAT would either be a derivative of
the South African grassland fauna, or be a local
development from a fauna or faunas with different
ecological requirements. The patten found con-

condiions sretching from EAT to

Cameroun.