Page 14 - iCC044-201501_March
P. 14
gin of the group, but our time-calibrat- ing phylogenies (e.g. bootstrap resam- lution reject Gondwanan vicariance as a
ed phylogeny permits investigation of the pling scores or Bayesian posterior prob- viable mechanism for the modern distri-
timing of deep divergences within the clade abilities). We also note that some phyloge- bution of the group, but they demand what
(table 1). We estimate the divergence of netic hypotheses derive from successive can only be considered a series of highly
South American and African cichlids as reweighting exercises [39], while others unlikely transoceanic dispersal events. Like
Eocene, with the origin of the African assume restricted placement of fossil the fossil record, the salinity tolerance of
cichlid crown within the same interval. species prior to analysis [13]. There is cichlids has been subjected to contrasting
This is consistent with the placement no doubt that Lumbrera cichlids are sig- interpretations; it has been cited as both
of the middle Eocene (approx. 46 Ma) nificant on account of their antiquity and consistent [12] and inconsistent [10] with
†Mahengechromis as an early crown pseu- geographical provenance. However, in the marine dispersal. Experimental evidence
docrenilabrine [12]. By contrast, our esti- absence of demonstrably robust phyloge- points to high salinity tolerance in some
mated Eocene-Oligocene age for the South netic placements of these fossil lineages cichlids [50,51], but the fact that no cichlid
American cichlid crown contradicts pub- within a group well known for convergent inhabits the open ocean indicates that long-
lished interpretations of the fossil cichlids morphological evolution [42], their exact distance marine migration is improbable.
from the ‘Faja Verde’ level of the Lumbrera implications for the timing of major events Dispersal across the south Atlantic would
Formation of Argentina. These fossils are in cichlid evolution are likely to remain appear to be especially unlikely, given
often cited as early-middle Eocene in age ambiguous. that it measured roughly 1000 km [52]
[13,38,39], leading to calibration minima of in width by the time of the inferred diver-
49 Ma in recent molecular clock studies [6]. (c) Comparison with other putative examples of gence between South American and
However, the hard minimum for the age Gondwanan vicariance African cichlids in the Eocene (figure 1).
of these fossils is 33.9 Ma, which derives Despite the presence of a substantial
from radiometric dating of overlying tuff Among vertebrates assumed to have lim- marine barrier, it is clear that at least
layers [40]. This more appropriate minimum ited dispersal ability across marine barriers, two groups of terrestrial mammals — pri-
age estimate only partially reconciles our cichlids are not unique in showing a broad mates and hystricognath rodents — dis-
time scale with previous phylogenetic inter- distribution across southern landmasses persed from Africa to South America at
pretations of the Lumbrera cichlids, each combined with a fossil record that com- approximately this time [53]. More gener-
of which has been placed within the South mences long after the tectonic break-up of ally, there is strong evidence from other
American crown in association with specific Gondwana. Several groups of freshwater animal groups and plants for surprisingly
cichline tribes (†Protocara as either a geo- fishes, reptiles, mammals and plants show high levels of biotic interchange between
phagine or a stem member of an unnamed disjunct distributions, with members pre- South America and Africa throughout the
clade comprising Chaetobranchini, Geo- sent in South America and Africa, but only Late Cretaceous and Palaeogene [54,55].
phagini, Cichlasomatini and Heroini [39,41]; a few instances seem definitively attribut- Geological evidence indicates the pres-
†Gymnogeophagus eocenicus as phylo- able to driftbased vicariance [14,43,44]. ence of a chain of now-submerged islands
genetically nested within a living genus Instead, molecular clock analyses for a across the south Atlantic during the
[38]; and †Plesioheros as a crown heroine range of groups with apparent vicariant Palaeogene [52]. These islands coincided
[13]). It is difficult to evaluate confidence distributions across southern continents with strong east-to-west palaeocurrents
in the evolutionary relationships proposed [45-48] paint a picture of widespread across the south Atlantic and both have
for these fossils because published anal- ‘pseudo-congruence’, where similar bio- been invoked as key elements of a selec-
yses using morphological data do not geographic patterns originate at different tive dispersal route from Africa to South
provide support for nodes in accompany- times that may be disjunct with the age America during the Eocene [12,52]. It is
of specific palaeogeographic events [49].
Our consistent time scales for cichlid evo-
12 facebook.cichlids.org.au
