3. Origin of Euteleostean Biodiversity
47
However, there is no direct evidence for the actual occurrence of the special
tectonic settings, without which it would have been virtually impossible for even
the euryhaline cichlids to travel between well-separated African and South American landmasses. Since cichlids are primarily fishes of lowland tropics, it is also
unlikely that they dispersed through freshwater systems in high-latitude areas. Because the Cretaceous earth had a relatively warm climate and high sea-levels throughout the period (Briggs 1995; refs. therein), it seems unlikely that African and South
American landmasses became temporarily connected through a landbridge after
their complete disconnection at 100 MYA. We therefore deduced that dispersal, if
any, could have happened only when the separated landmasses were relatively close
to each other (the dispersal model in Fig. 4B).
Among the three possible models for the intercontinental distribution of cichlids
(Fig. 4), the vicariance model seems to explain most naturally the phylogenetic
relationship among continental cichlid groups, but this model as well as the dispersal model have to assume that old lineages were replaced by new ones by remarkable adaptive radiations (see the legend of Fig. 4). Cichlids in the African rift
lakes are known for their explosive adaptive radiation and this may have occurred
frequently in the Cretaceous, too. The seawater-origin model remains possible but
..
Fig. 4. Models for the evolutionary history of cichlids that may reconcile the molecular and
geological data. (A) the vicariance model, (B) the dispersal model, and (C) the seawaterorigin model. In A, multiple freshwater cichlid lineages were already widespread on
Gondwanaland before its breakup. The Indian-Malagasy landmass was first separated from
Gondwanaland and old lineages now inhabiting India and Madagascar were geographically
isolated. Mrican and South American landmasses were then separated from each other and
cichlids could not disperse between them. Remarkable adaptive radiation of new lineages
and their replacement of old ones shaped extant Mrican and neotropical cichlids to be mono. phyletic to each other. The vicariance model may predict that extant Mrican and neotropical
ckhlid lineages separated from each other at 100-120 MYA. 100 MYA is the age of complete separation of Mrican and South American landmasses (Smith et al. 1994; The Plates
Project 1998). However, the lineage separation may have somewhat preceded the complete
separation of the two landmasses. Fossil evidence on taxa such as ostracods, foraminiferans,
and nannoplanktons strongly favored a marine north-south connection between the two landmasses as early as the Lower Albian (approximately 110 MY A) (Briggs 1995; refs. therein).
Geological evidence shows that the two landmasses were connected only through narrow
equatorial regions at 120 MYA (Smith et al. 1994; The Plates Project 1998). In B, after the
main landmasses became disconnected, some freshwater cichlids readapted to brackish water
could disperse in short distances between them. They replaced preexisting lineages (if any)
on each landmass and gave rise to extant lineages. The dispersal model may predict that
extant Mrican and neotropical lineages separated from each other at 80-100 MY A. Even
euryhaline cichlids readapted to brackish water may not have been able to disperse between
well-separated African and South American landmasses after 80 MY A. In C, after
Gondwanaland became well-fragmented, freshwater-adapted cichlids leading to extant lineages arose from labroid ancestor(s) that could travel across paleo-oceans. The seawaterorigin model does not, by itself, constrain the lineage separation time between landmasses.
However, Fig. 3C supports the premise that the lineage separation time between African
and neotropical cichlids based on this model cannot have been much younger than 80 MY A
47
However, there is no direct evidence for the actual occurrence of the special
tectonic settings, without which it would have been virtually impossible for even
the euryhaline cichlids to travel between well-separated African and South American landmasses. Since cichlids are primarily fishes of lowland tropics, it is also
unlikely that they dispersed through freshwater systems in high-latitude areas. Because the Cretaceous earth had a relatively warm climate and high sea-levels throughout the period (Briggs 1995; refs. therein), it seems unlikely that African and South
American landmasses became temporarily connected through a landbridge after
their complete disconnection at 100 MYA. We therefore deduced that dispersal, if
any, could have happened only when the separated landmasses were relatively close
to each other (the dispersal model in Fig. 4B).
Among the three possible models for the intercontinental distribution of cichlids
(Fig. 4), the vicariance model seems to explain most naturally the phylogenetic
relationship among continental cichlid groups, but this model as well as the dispersal model have to assume that old lineages were replaced by new ones by remarkable adaptive radiations (see the legend of Fig. 4). Cichlids in the African rift
lakes are known for their explosive adaptive radiation and this may have occurred
frequently in the Cretaceous, too. The seawater-origin model remains possible but
..
Fig. 4. Models for the evolutionary history of cichlids that may reconcile the molecular and
geological data. (A) the vicariance model, (B) the dispersal model, and (C) the seawaterorigin model. In A, multiple freshwater cichlid lineages were already widespread on
Gondwanaland before its breakup. The Indian-Malagasy landmass was first separated from
Gondwanaland and old lineages now inhabiting India and Madagascar were geographically
isolated. Mrican and South American landmasses were then separated from each other and
cichlids could not disperse between them. Remarkable adaptive radiation of new lineages
and their replacement of old ones shaped extant Mrican and neotropical cichlids to be mono. phyletic to each other. The vicariance model may predict that extant Mrican and neotropical
ckhlid lineages separated from each other at 100-120 MYA. 100 MYA is the age of complete separation of Mrican and South American landmasses (Smith et al. 1994; The Plates
Project 1998). However, the lineage separation may have somewhat preceded the complete
separation of the two landmasses. Fossil evidence on taxa such as ostracods, foraminiferans,
and nannoplanktons strongly favored a marine north-south connection between the two landmasses as early as the Lower Albian (approximately 110 MY A) (Briggs 1995; refs. therein).
Geological evidence shows that the two landmasses were connected only through narrow
equatorial regions at 120 MYA (Smith et al. 1994; The Plates Project 1998). In B, after the
main landmasses became disconnected, some freshwater cichlids readapted to brackish water
could disperse in short distances between them. They replaced preexisting lineages (if any)
on each landmass and gave rise to extant lineages. The dispersal model may predict that
extant Mrican and neotropical lineages separated from each other at 80-100 MY A. Even
euryhaline cichlids readapted to brackish water may not have been able to disperse between
well-separated African and South American landmasses after 80 MY A. In C, after
Gondwanaland became well-fragmented, freshwater-adapted cichlids leading to extant lineages arose from labroid ancestor(s) that could travel across paleo-oceans. The seawaterorigin model does not, by itself, constrain the lineage separation time between landmasses.
However, Fig. 3C supports the premise that the lineage separation time between African
and neotropical cichlids based on this model cannot have been much younger than 80 MY A