INSIGHTS | PERSPECTIVES
28 2 JANUARY 2015 • VOL 347 ISSUE 6217 sciencemag.org SCIENCE
tion on whether crosses between species can
produce fertile male progeny.
Fontaine et al. explored the consistency
of an Anopheles species tree by partitioning
the genome alignment of the six gambiae
complex species into 50-kb windows and
computationally inferring a separate phylogeny for each window. If the species split
in an orderly fashion, then essentially every window ought to reflect the same tree.
There are 85 possible topologies to the trees
with six gambiae complex species and one
outgroup, and in fact all 85 tree topologies
were seen in at least one genomic window
(see the first figure). Evidently the gambiae
complex of Anopheles has not been respecting species boundaries.
There are two possible reasons that the
overall pattern of divergence of genomic segments may differ from the overall species
tree, which has one of the more rarely supported topologies genome-wide. First, there
may be introgression when crosses between
species produce fertile female hybrids that
result in gene flow between species. Second,
there may be incomplete lineage sorting,
whereby an ancestral population splits into
two daughter species and each becomes fixed
for a different allele that was segregating at
a previously polymorphic locus in the ancestral species. The local phylogeny at such a locus inferred in the two daughter species may
then not necessarily be consistent with the
branching pattern of the species tree.
Fontaine et al. come down firmly on the
side of introgression (see the first figure).
This process seems plausible, but the level
of gene flow needed to essentially shuffle
the autosomal (non–sex chromosome) varia-
tion is quite high. This much interspecific
hybridization is surprising without more in-
trogression on the X chromosome, especially
in light of Neafsey’s result that the X-to-au-
tosome rate of transfer is unusually high in
Furthermore, the timing of the gambiae-
arabiensis introgression remains unclear.
In particular, it is not clear whether intro-
gression is still happening (in which case
arabiensis must still be undergoing hybrid-
ization with both gambiae and coluzzi) or
whether hybridization ceased some time ago.
Fontaine et al. have done a marvelous job in
highlighting the truly odd character of these
genomes, and their explanation is consistent
with the data, but it raises many additional
questions that warrant deeper study.
The breakdown of tidy bifurcating trees
with distinct species at the tips has been
seen in many systems ( 10). When there is
extensive exchange across species, the phy-
logeny is no longer treelike but rather has a
web of crossing lineages in the form of a net-
work (see the second figure). This so-called
reticulate evolution is especially
evident in bacteria, where ge-
netic exchange can be so perva-
sive that the concept of species
becomes quite slippery ( 10). Re-
ticulate evolution has been seen
in many other species groups,
but the pattern in the gambiae
complex of mosquitoes is so ex-
treme that it, too, challenges any
clear definition of species in this
group. Fontaine et al. adhere to
a classical view that there is a
“true species tree,” presumably
the phylogeny that is shown by
the genes that mediate male
and female fertility. But given
that the bulk of the genome has
a network of relationships that
is different from this true spe-
cies tree, perhaps we should
dispense with the tree and ac-
knowledge that these genomes are best de-
scribed by a network, and that they undergo
rampant reticulate evolution.
Beyond these two papers, additional tests
of whether Anopheles mosquitoes have an
accelerated rate of evolution with extensive
introgression between species may come
from contrasts of observed and expected
patterns of polymorphism within species,
requiring the sampling and sequencing of
multiple individuals from within each species ( 11, 12). Such a population genetic approach may be the simplest way to resolve
the lingering puzzles about this system. In
particular, the sizes and sequence diversity
of introgressed segments could be used to
model the past timing and extent of hybridization events. The ability to detect positive
selection for genomic features that might
confer human host adaptation would also
be greatly improved with polymorphism
data. Additional sequencing to characterize
polymorphism in An. gambiae would answer some questions, but would undoubtedly also raise new ones. For now, the two
papers succeed in dramatically advancing
Anopheles genomics and providing baseline
resources to answer many questions beyond those pursued here. ■
1. D. E. Neafsey et al ., Science 347, 1258522 (2015).
2. M. C. Fontaine et al , Science 347, 1258524 (2015).
3. A. G. Clark et al ., Nature 450, 203 (2007).
4. S. H. Martin et al ., Genome Res. 23, 1817 (2013).
5. K. Lindblad-Toh et al ., Nature 478, 476 (2011).
6. J. Rogers, R. A. Gibbs, Nat. Rev. Genet. 15, 347 (2014).
7. C. S. Mc Bride et al ., Nature 515, 222 (2014).
8. R. A. Holt et al ., Science 298, 129 (2002).
9. M. Coetzee et al., Zootaxa 3619, 246 (2013).
10. W. F. Doolittle, Science 284, 2124 (1999).
11. J. E. Pool et al., PLOS Genet. 8, e1003080 (2012).
12. T. F. Mackay et al ., Nature 482, 173 (2012).
col gam ara qua mel mer
More web than tree. The pattern of evolution seen in
the An. gambiae species complex resembles a network
more than a tree. This type of evolutionary network is
referred to as reticulate evolution ( 10).
True species phylogeny Inferred phylogenies on chromosomes
col gam ara qua mel mer
((col,gam),ara) ((col,gam),ara) (qua,mer)
Jumbled mosquito genomes. (A) The “true species tree” and major introgression events (red arrows) in the An. gambiae complex
inferred from the X chromosomal sequences by Fontaine et al.: An. coluzzii ( col), An. gambiae ( gam), An. arabiensis ( ara), An.
quadriannulatus ( qua), An. melas ( mel), and An. merus ( mer). (B) Locally inferred trees. The bars show the proportions of 50-kb
windows on each chromosome that yield phylogenies consistent with one of the three topologies shown above. Large portions of
the genome indicate that ara and col+gam are sister groups (dark blue); other regions of the genome also group together qua and
mer (light blue), compatible with the two major introgression events. Only windows on the X chromosome predominantly recover
the species topology, grouping together ara and qua (red). Altogether there are 85 tree topologies, and the gray areas correspond to
other topologies distinct from the three depicted ones.