Phylogenetic Data for Rainbowfishes (Melanotaeniidae)
General blurb and disclaimer
The following data are being generated as part of my research on the biogeography of various Australian fish groups.
Please do not use or copy these data to anywhere else as they are only
preliminary, and hence very likely to change as more data is accumulated.
Additionally, most of the DNA sequences have not been thoroughly checked
for reading errors. They are largely being placed here so others who have
contributed via sample collection and in other ways can see the
progress.
Background on rainbowfish taxonomy/research
Prior to the work of G. Allen, rainbowfish taxonomy was confused to say
the least! At the end of his revisionary studies published with N. Cross
in 1980, rainbowfishes allied with Melanotaenia splendida were distributed as show in the linked figure. The
only other minor change was the 1986 separation of M. splendida
fluviatilis into M. fluviatilis and M. duboulayi,
the former being restricted to the Murray-Darling Basin, the latter to the
east coast drainages.
The results presented here follow on from earlier work by students D. Zhu
and K. McGuigan in C. Moritz's lab at the University of Queensland. A
couple of their main discoveries were the separation of fishes related to
Melanotaenia splendida into two distinct lineages (as shown geographically in this linked
figure--except for M. eachamensis which is in the dub-fluv
clade), and the identification of additional populations of M.
eachamensis and an additional lineage, separate, but similar to M.
eachamensis. Furthermore, they identified likely hybridisation
between M. nigrans and M. s. australis in the Blackmore
River and M. exquisita and M. s. inornata in the
South Alligator River.
The present study
The following provides an extension of those results, with a more intensive
investigation into the fine scale population genetics of the M.
splendida complex in Australia. After a major field trip in 1997, a larger number of collections had been
assembled (red dots are my own collections, green dots are those at
the South Australian Museum's Evolutionary Biology Unit) The
splendida complex refers to the species shown here. Data presented here include the first
600 bases of the mitochondrial cytochrome b gene and allozyme
electrophoresis. Species names are typically based on geography and colour
patterns.
Mitochondrial results (cytochrome b)
Results gathered for cyt. b so far by and large
confirm Zhu and McGuigan's results. Several points are noteworthy though.
Several populations identified based on morphology and geography don't
"fit" in their logical location on the tree. These include the positions
of both M. exquisita populations, M. s. australis from
Charlotte and Blackmore rivers, M. duboulayi from Mullet Creek,
and M. s. splendida from Comet and Dawson rivers. Another point
worthy of mention is the position of two presumed F1 hybrids, one from the
Drysdale River (rfhh), the other from Cairns (rf97gg). The female in the
former hybrid appears to be M. gracilis, the latter female
appears to be M. trifasciata, which is not native to the Cairns
area.
Allozyme electrophoresis
In order to investigate what influence of hybridisation on the history of
these populations noted above, I undertook an allozyme electrophoretic
analysis to provide a survey of nuclear genes. Present results are based
on 5 enzymes and 6 loci.
Northern results
Results are presented in two sections (as shown
here).
The populations in the northern part of the study are
shown here. For allele frequences click on north UPGMA trees are used to show how these
populations appear to be related to one another. The UPGMA tree for the northern study site has
several contrasts relative to mitochondrial DNA data. Firstly, M.
exquisita from Bindoola Creek shows no close relationship to any
other populations. While I don't have fish from sympatric, or nearby
M. s. australis populations for direct comparison, it seems
likely this population has exchanged mitochondria, but not exchanged
nuclear genes with M. s. australis! The Douglas River population
of M. s. australis clusters with M. s. inornata, rather
than its own type, largely due to similarities in allele frequencies at
PGM and MDH-1. Although it is not shown on the mitchondrial DNA tree, it
falls within the other M. s. australis samples. It remains a
little enigmatic without more data. The other M. exquisita
population from Mary River clusters within M. s. inornata
populations. This M. exquisita population is fixed at each loci
which is usually also the dominant one in the M. s. inornata
populations examined. Without more M. exquisita populations this
is difficult to interpret, but it seems likely they have exchanged both
mitochondrial and nuclear genes. Lastly, M. nigrans is somewhat
distinctive with one fixed, and one nearly fixed difference (PGDH and ICDH
respectively). However, this population is from Cape York and there may
be differences between these and Darwin populations. It seems somewhat
clear though that both the Blackmore and Charlotte river populations of
M. s. australis have exchanged mitochondrial genes, but have
probably not swapped nuclear genes.
Southeastern results
Populations examined for the southeastern coast are shown here (except M. fluviatilis). Allele
frequencies are shown here. The UPGMA tree for the southeastern study site
shows the Mullet Creek population of M. duboulayi clustering with
M. s. splendida as it did with cyt. b data. There are no fixed
differences between these species, most differences are due to allele
frequency differences, with most loci being fixed in M. duboulayi
(except for GPI and Mullet Creek). Hence, while the differences are
minor, they appear to be consistent across several loci (esp. PGDH, and to
a lesser extent ICDH and GPI). Hence, despite having similar colour
patterns and morphology to surrounding M. duboulayi populations,
genetically Mullet Creek fish are most similar to M. s.
splendida. The other enigmatic populations are those from the Comet
and Dawson rivers. They have dub-fluv mitochondrial DNA, they look
similar colour wise to M. fluviatilis (but also with a hint of
M. s. splendida) and allozymically they are both quite different
to other species examined (especially MDH-2, as well as major frequency
differences with ICDH and PGDH) and also somewhat different to each other
(MDH-2 and PGDH). The most likely conclusion at this stage is they are a
previously unrecognised taxa or maybe two taxa. It is notable though that
the Comet River population shares a number of alleles with M. s.
splendida and M. fluviatilis.
S7 intron results
Rainbowfish S7 neighbour joining tree based on
562 bases (with all gaps and ambiguous characters removed) of the first
intron in the S7 gene. Distances are Jukes-Cantor. Note that M.
exquisita is somewhat monophyletic (relative to cyt. B data) and that
the two clades containing the spendidas identified in cyt. B have
collapsed and are largely indistinuguishable. Note that M. s.
australis (Charlotte River and Blackmore River) are not grouping with
M. nigrans as they do with cyt. B. Also, M. s.
australis (King Edward River) is not clustering with M.
gracilis as it does in cyt. B.
Rainbowfish S7 fast bootstrap tree (1000 reps)
based on 562 bases of the first intron in the S7 gene.
Present conclusions
Rainbowfish genetics (and hence the history of the populations) appears
far more complicated than previously recognised. Clearly hybridisation
has not been uncommon in the past between M. exquisita and sympatric
M. splendida species, and between M. nigrans and M.
s. australis. However, there is no evidence (at least not at this
stage) that hybridisation is occurring today between these species, it all
appears to be historical (although how long ago is yet to be determined),
although more data are necessary to test that hypothesis. Unfortunately
for now I am unable to continue examining populations around Darwin until
I can collect more material. However, I have quite a number of additional
populations from the east coast that should allow a clearer picture to
emerge. I also have a few additional enzyme systems to incorporate into
this study which may improve the resolution of the patterns that exist.
I am also working on obtaining the other half of cyt. b (which is
typically more phylogenetically informative) and sequencing nuclear genes.
Please remember that most of what is on this page is only preliminary
data, and as more comes in, the present interpretations will change.
If you have any questions please feel free to email Peter Unmack at
.
