Page 12 - iCC044-201501_March
P. 12
molecular-clock analyses. Furthermore, we fossil record, but the upper bounds of crown acanthomorph lineages are likely
have not included any internal calibra- the CIs do increase by more than 10 Ma. to be sufficiently ancient to have vicariant
tions within Cichlidae, so that our relaxed- Depending on the scope of geographical Gondwanan distributions, as we estimate
molecular-clock estimate of the evolution- analysis, we estimate the time of cichlid the age of the group as between 106.4Ma
ary time scale for the group is truly inde- origin based only on articulated remains as (95% CI: 98.5-132.2Ma) and 109.2Ma (95%
pendent of its fossil record, which contrib- ranging from 59.8 Ma (95% CI: 56.1-75.1 CI: 98.5-136.0Ma). The most restrictive
utes to our palaeontological estimates of Ma; landmasses inhabited by modern cich- group containing cichlids that we can date
divergence times (see §2). lids) to 60.2 Ma (95% CI: 56.1-77.8 Ma; all with this method and which is of sufficient
Gondwanan landmasses). The Gondwanan apparent antiquity to have been affected by
3. Results vicariance hypothesis requires a pre- the initial rifting of Gondwana is Eurypterygii,
Our three approaches to estimating a Eocene record of cichlids that is roughly the radiation containing Acanthomorpha,
10-30 times worse than their recorded Myctophiformes and Aulopiformes [25].
time scale of cichlid origin and diversifi- fossil history, with rescaled recovery poten- Our estimates for the time of origin for
cation yield overlapping CIs that diverge tials conditioned on point estimates for this major teleost clade range between
significantly from the predictions made by the origin of the group at 135 Ma ranging 131.1Ma (95% CI: 104.9-163.2Ma) and
the Gondwanan vicariance biogeographic from 2.8-3.3% (all fossils) to 6.6-6.9% 142.1 Ma (95% CI: 126.2-166.2 Ma).
hypothesis, and are discussed in turn in (articulated fossils only) of their original
§3a,b (figure 1). values. Classical confidence intervals deliver (b) A molecular time scale for cichlid evolution
similar results to the Bayesian estimates (see The phylogeny of Ovalentaria and the
(a) Palaeontological time scales for cichlid evolution electronic supplementary material, table S2).
The distribution of cichlid-bearing fos- Analysis of outgroup ages provides broadly major cichlid lineages inferred from the 10
similar estimates for the timing of cichlid nuclear genes is similar to previous molec-
sil horizons, combined with an empiri- origin to those derived from the distribu- ular and morphological analyses [8,10,19],
cally informed function describing fossil tion of cichlid fossil horizons, in terms of with Etroplinae (India, Madagascar)
recovery potential, indicates an age of both the magnitude of point estimates resolved as the earliest-diverging clade
origin for cichlids in the Late Cretaceous and the degree of uncertainty surround- and Ptychochrominae (Madagascar) as the
or Palaeocene. If only the records of land- ing them. We find a mean age of 60.7 Ma sister lineage to the unnamed clade that
masses that are currently inhabited by (95% CI: 46.8, 90.1 Ma) using the old- contains the African (Pseudocrenilabrinae)
cichlids are considered, the time of origin est possible fossil ages for outgroups. and Neotropical (Cichlinae) cichlid line-
of the clade is estimated as 59.2 Ma (95% The time scale for cichlid origin is predict- ages (figure 1; electronic supplementary
CI: 56.1-67.6 Ma). By contrast, a slightly ably more recent using the youngest pos- material, figure S1). The 10 nuclear gene
younger age estimate of 57.8 Ma (95% CI: sible fossil ages for outgroups, but only phylogeny preserves the parallels between
56.1-62.4 Ma) is obtained if the record of slightly so, with a mean age of 57.0 Ma patterns of cichlid interrelationships and
all Gondwanan landmasses is considered. (95% CI: 46.8-81.2 Ma). Using this same the fragmentation history of Gondwana
Restricting the scope of analysis to con- approach, it is also possible to determine that has led to the prominence of vicari-
sider articulated remains alone provides probable times of origin for a series of more ance biogeographic scenarios for this lin-
a more conservative means of estimating inclusive clades that contain Cichlidae: eage [9]. However, the Bayesian random
the time of origin for cichlids, because Ovalentaria, Percomorpha, Acanthopterygii, local molecular-clock analyses yield age
early members of this group might not be Acanthomorpha, Eurypterygii, Euteleostei estimates for the origin of cichlids consist-
recognized on the basis of less diagnos- andTeleostei. This exercise implies that no ent with those derived from analysis of
tic skeletal debris. Point estimates for the fossils alone (figure 1 and table 1; elec-
timing of cichlid origin under this approach tronic supplementary material, figure S1).
do not change drastically from those
obtained using the entirety of the cichlid
10 facebook.cichlids.org.au