Plio‐Pleistocene sea‐level changes drive speciation of freshwater fishes in north‐western Australia

Abstract

Aim
Despite the influence of sea‐level changes on biogeographic/phylogeographic patterns in freshwater ecosystems being well documented, studies that explicitly link the influence of sea‐level change with speciation are rare. We aim to test the hypothesis that sea‐level changes during the Pliocene and Pleistocene have driven speciation in north‐western Australia's (NWA’s) largest freshwater fish family, Terapontidae, building upon a body of evolutionary literature focussed on the family.

Location
North‐western Australian rivers including those draining the Kimberley Plateau.

Taxon
Grunters (Family: Terapontidae, Genera: Hannia, Hephaestus, Leiopotherapon, Syncomistes).

Methods
A GIS was used to reconstruct palaeodrainages during lowered sea levels and to delineate regions of high connectivity during low and high (current) sea‐level conditions. For seven species, the degree of phylogenetic divergence among river basins in different regions was evaluated using a maximum likelihood phylogeny and analyses of the proportion of genetic divergence expressed with 601 base pairs of the mtDNA cytochrome b (cytb) gene.

Results
A low proportion of cytb haplotypes were shared among catchments not connected by the same receiving waters (e.g. estuaries) under current (high) sea levels, indicating that contemporary dispersal is limited over fine spatial scales. Deeper phylogeographic patterns were largely congruent with reconstructed low sea‐level (LSL) drainage arrangements indicating that historic among‐catchment connectivity was far more widespread under LSL conditions.

Main Conclusions
The NWA landscape represents a geographic template that has shaped patterns of broad dispersal under low sea levels, and fine‐scale isolation under high sea levels. The weight of evidence from recent literature on species boundaries and evolutionary patterns within the terapontids suggests that most NWA species were derived rapidly and recently from a series of spatio‐temporal vicariant events caused by such sea‐level fluctuations during the late Pliocene and Pleistocene. Together, the findings provide a rare, comprehensively tested example of sea‐level change driving speciation in the tropics.

Full text

Journal of Biogeography. 2020;00:1–12. wileyonlinelibrary.com/journal/jbi
|
1© 2020 John Wiley & Sons Ltd
Received: 1 Octob er 2019
|
Revised: 20 Fe bruar y 2020
|
Accepted: 7 Ma rch 2020
DOI : 10.1111/j bi.1 385 6
RESEARCH PAPER
Plio-Pleistocene sea-level changes drive speciation of
freshwater fishes in north-western Australia
James J. Shelley1,2 | Stephen E. Swearer1 | Tim Dempster1 | Mark Adams3,4 |
Matthew C. Le Feuvre1 | Michael P. Hammer5 | Peter J. Unmack6
1School of BioScie nces, University of
Melbou rne, Me lbourne, VIC , Australia
2Depar tment of Environment, L and, Wate r
and Planning, Arthur Rylah Institute for
Environmental Research, Heid elberg, VIC ,
Australia
3Evolutionary Biology Unit, South Australian
Museum, Adelaide, SA, Australia
4School of Biological Sciences, University of
Adelaide, Adelaide, SA, Australia
5Natural Sciences, Museum and Art Gallery
of the Nor ther n Territor y, Darwin , NT,
Australia
6Instit ute for Ap plied Ecology, Uni versit y of
Canbe rra, C anberra, AC T, Australia
Correspondence
James J. Shelley, Aquatic Biodiversity
and Conservation Program, A rthu r Rylah
Instit ute for Environme ntal Research , VIC,
Australia
Email: james.shelley@delwp.vic.gov.au
Funding information
We acknowledge the f unding contributions
for this project f rom the H ermon S lade
Foundat ion, the Holsworth Wildlife
Research Endowm ent, th e Winifred Violet
Scott Ch aritable Trust an d the Nature
Conservancy (Australian Conservation
Taxonomy Award).
Abstract
Aim: Despite the influence of sea-level changes on biogeographic/phylogeographic
patterns in freshwater ecosystems being well documented, studies that explicitly link
the influence of sea-level change with speciation are rare. We aim to test the hypoth-
esis that sea-level changes during the Pliocene and Pleistocene have driven specia-
tion in north-western Australia's (NWA’s) largest freshwater fish family, Terapontidae,
building upon a body of evolutionary literature focussed on the family.
Location: N o r t h -weste r n Australia n rivers in c luding th o se drai n i ng the Ki m b e rley Pl ateau.
Tax o n: Grunters (Family: Terapontidae, Genera: Hannia, Hephaestus, Leiopotherapon,
Syncomistes).
Methods: A GIS was used to reconstruct palaeodrainages during lowered sea levels
and to delineate regions of high connectivity during low and high (current) sea-level
conditions. For seven species, the degree of phylogenetic divergence among river
basins in different regions was evaluated using a maximum likelihood phylogeny and
analyses of the proportion of genetic divergence expressed with 601 base pairs of
the mtDNA cytochrome b (cytb) gene.
Results: A low proportion of cytb haplotypes were shared among catchments not
connected by the same receiving waters (e.g. estuaries) under current (high) sea lev-
els, indicating that contemporary dispersal is limited over fine spatial scales. Deeper
phylogeographic patterns were largely congruent with reconstructed low sea-level
(LSL) drainage arrangements indicating that historic among-catchment connectivity
was far more widespread under LSL conditions.
Main Conclusions: The NWA landscap e represents a geographic template that has shaped
patterns of broad dispersal under low sea levels, and fine-scale isolation under high sea
levels. The weight of evidence from recent literature on species boundaries and evolution-
ary patterns within the terapontids suggests that most NWA species were derived rapidly
and recently from a series of spatio-temporal vicariant events caused by such sea-level
fluctuations during the late Pliocene and Pleistocene. Together, the findings provide a rare,
comprehensively tested example of sea-level change driving speciation in the tropics.
KEYWORDS
cryptic species, glacial cycles, introgression, phylogeography, Terapontidae, tropics, vicariance

2
|
SHELLEY Et aL.
1 | INTRODUCTION
Understanding the underlying mechanisms that generate biodiver-
sity is a fundamental goal of evolutionar y and conservation biol-
ogy. Vicariance, caused by geological, climatic or eustatic (sea-level
change) processes, is considered a common mechanism for gener-
ating new species via allopatric speciation (e.g. Avise, 2000, 2004;
Riginos, 2005). Comparative phylogeographic analysis provides a
strong framework for investigating whether discontinuities in spe-
cies assemblages and genetic structuring within species correspond
to past or present geographic barriers (Unmack, Hammer, Adams,
Johnson, & Dowling, 2013). Obligate freshwater taxa are particularly
suit ed to suc h ana lys es be cau se th ey are li mited to we ll- de f in ed ha b-
itat units (e.g. rivers and lakes) and any genetic signature of connec-
tivity among habitat unit s implies a freshwater connection, either
past or present (Unmack, 2013).
The glacial cycles during the Pliocene and Pleistocene have long
been viewed as strong drivers of vicariant speciation around the
globe, promoting the formation of new species through repeated
geographic isolation and connection (Avise, Walker, & Johns, 1998;
Haffer, 1969; Johnson & Cicero, 2004; Wallis, Waters, Upton, &
Craw , 20 16) . Howe ver, th i s th e o r y is th e su b j e c t of de bat e as several
studies have concluded that as many or more species arose prior to
this period (Colinvaux & De Oliveira, 2001; Knapp & Mallet, 2003;
Rull, 20 08). During glacial cycles, regions in temperate latitudes of
the world were most heavily influenced by the expansion and con-
traction of ice sheets that facilitated the isolation and connection
of aquatic communities based on movement of the boundaries of
unglaciated refuges (e.g. April, Hanner, Dion-Côté, & Bernatchez,
2013; Grif fiths, 2006; Waters & Wallis, 2001). In contrast, in tropi-
cal latitudes, wet/dry climatic cycles and new land connections as-
sociated with the exposure of the continental shelf during lowered
sea levels are considered the key factors influencing connectivity,
distribution and speciation (Hewitt, 2000; Voris, 20 00).
Despite many papers referring to the relative impor tance of
sea-level changes to biogeographic/phylogeographic patterns in
freshwater ecosystems (e.g. Dias et al., 2014; Perdices, Bermingham,
Montilla, & Doadrio, 2002; Shelley, Dempster, et al., 2019), studies
that explicitly test hypotheses relating to the influence of sea-level
change on speciation in freshwater are rare. In the tropics, a sub-
stantial body of evidence from the east coast of southern Brazil
and Uruguay provides support for sea-level change driving broad-
scale, repeated event s of vicariant speciation (Tschá, Bachmann,
Abilhoa, & Boeger, 2017), but we are not aware of further exam-
ples. Consequently, the importance of vicariant speciation, driven
by Plio-Pleistocene glacial cycles, in the diversification of tropical
freshwater biodiversit y globally remains unclear. Here, we assess the
influence of Plio-Pleistocene sea-level changes on connectivity and
speciation in tropical north-western Australia (NWA) using freshwa-
ter fishes as a model group.
NWA is a largely flat and ancient landscape (White, 2000). The
dominant geological feature in the region is the elevated Kimberley
Plateau, possibly the oldest continually exposed landform in the
world (~700 Ma; Ollier, Gaunt, & Jurkowski, 1988). The region con-
tains a dispropor tionately large number of endemic species within
many plant and animal groups (Pepper & Keogh, 2014). Freshwater
fishes conform with this broader pattern of high endemism (~35
species), and the region cont ains more than a quarter of Australia's
freshwater fish diversity (~68 species) (Shelley, Morgan, et al., 2018;
Shelley, Swearer, et al., 2018; Unmack, 2013). Within the endemic fish
communit y, 24 (67%) of species are found exclusively in streams run-
ning through the plateau (Shelley, Morgan, et al., 2018). Regardless,
due to NWA’s remoteness, the biodiversity and evolutionary history
of the region is still poorly studied (Bowman et al., 2010).
The rivers that run off the plateau are short, steep and flow
through deeply dissected gorges, often from their headwaters to
the ocean. Consequently, connectivity among catchments at current
sea levels is expected to be low (Phillips, Storey, & Johnson, 2009;
Shelley, Dempster, et al., 2019). Furthermore, due to the long-term
geological stability of the plateau, historical river rearrangement s
or capture events are unlikely (Unmack, 2013). Alternatively, low-
ered sea levels exposed one of the widest continent al shelves in the
world (Yokoyama, Purcell, Lambeck, & Johnston, 2001), providing
greater chances for rivers to coalesce together before they reached
the ocean and, consequently, for aquatic organisms to disperse.
Therefore, predictable phylogenetic signatures of Plio-Pleistocene
connectivity among rivers that drain the plateau, repeated across
multiple taxa, would provide strong evidence that range expansion
was facilitated by the exposure of the continental shelf under low
sea levels rather than dispersal among catchments during chance
events (e.g. river capture, flooding of low divides or coastal dispersal)
(Burridge, Craw, & Waters, 2006; Waters et al., 2007).
Geographically widespread molecular studies have been lack-
ing in NWA. A recent region-wide molecular analysis of the region's
most species rich freshwater fish family, Terapontidae, provides a
unique oppor tunit y to investigate the potential influence of sea-
level change on phylogenetic affinities and ultimately speciation in
this freshwater group (Shelley, Swearer, et al., 2018). The previous
study focussed on cryptic species discovery, employing bi-paren-
tally inherited, multi-locus nuclear markers (allozyme markers and
Recombination Activation Gene one [RAG1] sequences) to explore
species boundaries, suppor ted by sequence data from the mtDNA
cytochrome b gene (cytb). The results indic ated that 22 described
and candidate terapontid species are present in the AMT.
Species relationships, the relative timing of speciation events,
and diversification rates have also previously been estimated using
the combined sequence data, providing insight into the evolutionary
history of the family (Shelley, Swearer, et al., 2018; Shelley, Unmack,
Dempster, Le Feuvre, & Swearer, 2019). Several findings from these
studies lend support to the hypothesis that sea-level change has
driven speciation in NWA terapontids. These include (a) the con-
generic species in sympatr y (12 NWA terapontid species exhibit
sympatry of congeners) are not sister species (with one exception)
suggesting that they arose in allopatry and have subsequently come
into secondary contact, (b) 75% of speciation events in NWA en-
demic species occurred during the late-Pliocene and Pleistocene

|
3
SHELLE Y Et aL.
when sea-level fluctuations were at peak intensity, and diversifica-
tion rates increased significantly at this time, and (c) reproductive
boundaries of sympatric congeneric species appear to be incom-
plete in most cases, perhaps reflecting the consequences of inde-
pendently derived species reconnecting early in their evolutionary
history. However, without an understanding of whether patterns of
dispersal in freshwater taxa have been influenced by changes in sea-
level heights, the link between sea-level fluctuations and speciation
remains tenuous.
In this study we compare the genetic structure of seven ter-
apontid fishes to investigate the role of sea-level changes in driv-
ing phylogenetic patterns, vicariant speciation and freshwater fish
diversification in NWA. We hypothesize that phylogenetic related-
ness among populations of widespread species will be high among
rivers that coalesced or had closely situated river mouths during
lowered sea levels, while evidence of recent dispersal will be rare
or absent among rivers that are disconnected under current sea lev-
els. Together, with evidence from existing evolutionar y studies of
terapontids in the region, we propose that the weight of evidence
favours the hypothesis that sea-level fluc tuations during the Plio-
Pleistocene have driven speciation in NWA terapontids.
2 | MATERIALS AND METHODS
2.1 | Regional setting
We define NWA using the boundaries of the Kimberley freshwa-
ter fish biogeographic province (Shelley, Dempster, et al., 2019).
The area is bounded by the Great Sandy Desert to the south, the
Indian Ocean to the west and north, and the eastern boundar y of the
Fitzmaurice River catchment in the east; the land area is roughly the
size of Spain (see Figure 1). Dramatic shif t s in sea level and pre ci pi ta-
tion occurred during Quaternary glacial cycles that alternated every
100,000–150,00 0 years (reviewed in Byrne et al., 2008). Compared
with today, sea levels ranged from −120 to −140 m during cool
and hyper-arid periods of glacial maxima to between +5 and +8 m
during the warm and wet interglacial periods (Hope et al., 2004).
FIGURE 1 North-western Australia showing the hypothesized low sea-level (a) and high sea-level (b) regions, and the distribution of
sample sites (numbered dots). In both panels rivers are denoted by blue lines and major catchment boundaries are outlined in black. In
Figure 1a, low sea-level drainage patterns are shown in the blue region, which shows bathymetry down to 135 m below present-day sea
level. The dashed lines mark the low sea-level boundaries to dispersal and river catchments in our study area are coloured according to their
region. In Figure 1b, river catchments are coloured according to their high (current) sea-level region. Inset is a map of Australia and the state
boundaries with a square highlighting the study region. Rivers are denoted by capital letters and are as follows: (A) Fitzroy, (B) Isdell, (C)
Charnley, (D) Calder, (E) Sale, (F) Glenelg, (G) Prince Regent, (H) Roe, (I) Mitchell, (J) King Edward, (K) Drysdale, (L) King George, (M) Berkley,
(N) Durack, (O) Pentecost, (P) Ord, (Q) Victoria and (R) Daly. Sample site numbers refer to the full sample locality details supplied in Table S1

4
|
SHELLEY Et aL.
Bathymetric analysis by Yokoyama et al. (2001) suggested that all
catchments surrounding the Joseph Bonaparte Gulf coalesced into
one brackish coastal lake, which we herein refer to as Lake Bonaparte
(Figure 1a). The region therefore provides an ideal setting for testing
hypotheses regarding the phylogeographic affinities of freshwater
communities under low sea-level (LSL) conditions.
2.2 | Study taxa and sympatry of species
The family Terapontidae (~67 species in total) includes ~40 small- to
medium-sized, stout, obligate freshwater species, 32 of which are
distributed broadly across northern Australia (Shelley, Swearer, et al.,
2018; Vari, 1978). Shelley, Swearer, et al. (2018) presented evidence for
13 new candidate species in NWA, seven of which were formally de-
scribed by Shelley, Delaval, and Le Feuvre (2017). We adopt both can-
didate and described species into the taxonomic framework used here.
Of the 22 terapontids found in NWA , half are narrow-range endemics
restricted to one or two river systems (Shelley, Morgan, et al., 2018).
Sympatry among congeners is relatively rare in Australia and
when found usually involves a widespread and a narrow range spe-
cies (Unmack, 2013). However, it is a common phenomenon across
NWA (11 genera, 33 species) and, on the Kimberley Plateau, sym-
patry is observed among narrow-range endemics (Shelley, Morgan,
et al., 2018). Within the NWA terapontids, Syncomistes (eight spe-
cies), Hephaestus (two species) and Leiopotherapon (two species)
exhibit sympatry of congeners. The remaining seven species have
discrete ranges that immediately neighbour a congeneric. The sym-
patric species either arose in sympatry (e.g. Pigeon, Chouinard, &
Bernatchez, 1997) or in isolation followed by range expansion that
led to secondary contact.
The clustering of many secondary contact zones is typical of re-
gions influenced by glacial expansion and contraction in temperate
latitudes (Campton & Ut ter, 1987). Such regions have been described
as ‘suture zones’ (Remington, 1968). Suture zones represent a unique
opportunity for the study of speciation, as each region of secondar y
contact represents a test of reproductive isolation in which the per-
meabilit y of gene exchange among the previously allopatric genetic
lineages can be assessed (Barton & Hewitt, 1985). As such, each case
of sympatry in the terapontids provides insight into (a) likely instances
of secondary contact, and (b) the strength of species boundaries,
which provides supportive information regarding the relative timing of
seco nd ary con tac t (i.e. recen t or not recen t) . Th e secon d po in t is bas ed
on the assumption that weaker species boundaries, evidenced by
greater instances of hybridization, will exist between recently derived
species and stronger reproductive barriers will exist between species
that diverged deeper in evolutionary time (Coyne & Orr, 2004).
2.3 | Sampling
Fishes were sampled across NWA by J.J.S., M.C.L., T.D. and S.E.S.
between 2012 and 2015, covering 41 sites and an area of some
512,000 km2 (Figure 1). Samples from outside the study area were
sou rced by M.P.H. and P.J.U. and from sever al separate tissue col le c-
tions (see acknowledgements). Most described or candidate species
from NWA were sampled from all known major river systems, with
minor gaps in coverage including L. unicolor, that was not sourced
from the Isdell, Charnley, Drysdale and Durack rivers. Phylogenetic
analysis included 54 of the 67 described and candidate terapontid
species, including all Australian species. However, as we were ulti-
mately only interested in the relationships recovered within each of
our target species lineages, we only present those results here. The
details of the complete cytb dataset are discussed in the following
section and the complete cytb terapontid phylogeny is presented in
Shelley, Swearer, et al. (2018).
2.4 | Sequence analysis
We used the cytb sequence dataset (601 bp) presented in Shelley,
Swearer, et al. (2018). Full details of the development of the dataset
and analyses are found there. Briefly, tree building was performed
with a likelihood approach in GARLI 2.0 (Zwickl, 2006). We identi-
fied the best-fitting model of molecular evolution for the analysis
using the Akaike Information Criterion (AIC) in jModelTest 2.1.4
(Darriba, Tab oada, Doallo, & Posada, 2012). jModelTest identified
TIM1+I+G as the best model. GARLI was run using 10 search repli-
cates with the following default settings changed: attachmentsper-
taxon = 294, genthreshfortopoterm = 10,000. For bootstrapping,
we ran 1,000 replicates with the same settings except treerejec-
tionthreshold was reduced to 20 as recommended in the GARLI
manual.
2.5 | Interpreting introgression
The mtDNA (cytb) and nuclear (RAG1) gene trees derived from
the datasets in Shelley, Swearer, et al. (2018) contained several
cases of discordance in the recovered species relationships, par-
ticularly within sympatric species from the genus Syncomistes. In
these cases, sympatric species shared a single cytb haplotype,
while the RAG1 dataset clearly distinguished among them and
recovered their relationships. The RAG1 data were further sup-
ported by the allozyme and morphological datasets employed.
Discordance between mitochondrial and nuclear gene trees may
arise for several reasons, including sporadic introgression, a lack
of lineage sorting and/or ancestral polymorphism, and paral-
ogy, that is, the inclusion of nuclear copies (Wallis et al., 2017).
However, given that nearly all cases of heterospecific mtDNA
haplotypes involved sympatric, nonsister species, and further-
more, allozyme analysis detected hybrid crosses among a vari-
ety of sympatric species pairs, sporadic mtDNA introgression
was identified as the most likely reason for this pattern (Shelley,
Swearer, et al., 2018). Therefore, we interpret these results as
likely cases of introgression.

|
5
SHELLE Y Et aL.
2.6 | GIS bathymetry manipulation and defining
low and high sea-level regions
To test the hypothesis that phylogenetic relatedness among populations
of widespread species will be high among rivers that coalesced or were
closely situated during lowered sea levels , but not ably lower among riv-
ers that are disconnected under current sea levels, we first developed
a geographic framework of low and high sea-level (HSL) regions. These
regions were based on historic and current patterns of catchment con-
nectivit y and clear geological divides interpreted from bathymetry data
from the continental shelf surrounding the Kimberley region.
Drainage patterns during low sea levels were modelled using
Spatial Analyst 1.1 and ArcView 3.1, based on a bathymetric 30
arc-second grid produced by the Australian Geological Survey
Organization. Palaeo-drainages are displayed from the current coast-
line to the sea floor depth of −135 m (the maximum estimated fall in
sea level during the last glacial maximum; Clark & Mix, 2002). Based
on th es e data , we identified fou r broad regio ns that we hy pothesize d
would have experienced high intra-regional connectivity and inter-re-
gional isolation during lowered levels (Figure 1a). Furthermore, we
defined 12 HSL sub-regions that we hypothesized would experience
freshwater connection under current sea levels based on whether
they flowed into the same receiving water body (e.g. estuary or bay)
or were seasonally connected across a floodplain (Figure 1b). Given
that his to ri cal sea -l ev el hei ght s we re onl y +5 to +8 m highe r th an cur-
rent levels (Hope et al., 2004), and heightened sea levels would only
exacerbate isolation in catchments, we considered the conditions to
be compa rable and so we ref er to cu rrent sea level s as ‘hi gh’ se a- le ve l
conditions. Within this framework, we refer to ‘widespread’ species
as those that: (a) occur across LSL boundaries, or (b) are distributed
across ≥3 HSL sub-regions or beyond the study region. ‘Narrow-
range’ species do not meet these criteria.
2.7 | Phylogeographic analysis
To assess genetic similarity within and among our hypothesized LSL re-
gions, we calculated the proportion of genetic differences (p-distance)
from the cytb data using MEGA 6.06. Given the last glacial maximum
(sea-level minimum) occurred only 18 ka, our hypothesized HSL sub-
regions would have been formed more recently (current sea levels were
reached around 6 ka) and the lineages would not have diverged enough
to be analysed in the same fashion as the LSL regions (Yokoyama et al.,
2001). Instead, we calculated the proportion of haplotypes in a sub-re-
gion that are shared across sub-regions, as an indic ator of connectivit y.
3 | RESULTS
3.1 | Phylogeographic analysis of LSL regions
Based on the geographic framework derived from our GIS analysis,
we identified four ‘widespread’ species or sibling pairs with either
region-wide (L. unicolor, He. jenkinsi), or inter-regional (Ha. greenwayi
I/Ha. greenwayi II, Western and Nor ther n regio ns; S. trigonicus/S. wu-
nambal, Northern, Mitchell, and Eastern regions) distributions that
were suitable for our phylogeographic analysis of LSL boundaries.
We also included S. bonapartensis in the analysis as it is distributed
across the large Eastern region and could provide further insight into
changed levels of connectivity around the catchments that drain
into Lake Bonapar te. The final phylogeographic dataset included
cytb sequences from 1 to 5 individuals from each catchment (154
samples in total) across the range of each target widespread species
(Figure 1; Table S1). A summar y cytb sample sizes for each species
used in our phylogeographic analysis, listed by catchment and low
and HSL region is presented in Table 1. The mean cytb p-distances
in and among LSL region lineages for these species are presented in
Table 2 and summarized in Table 3.
All of these species except L. unicolor, Australia's most wide-
spread species and an extreme disperser (Bostock, Adams,
Laurenson, & Austin, 2006), displayed phylogeographic structure
broadly consistent with biogeographic regionalization shaped by
low sea levels. The phylogenetic relationships of samples within all
species, except L. unicolor, are presented in Figure 2. Hephaestus jen-
kinsi displayed three well-supported lineages (88 and 99% of boot-
strap replicates) that mirrored a Western, Northern and Mitchell/
Eastern distribution (Figure 2a). Although nested in the Eastern
region, the Mitchell was the most divergent population within that
lineage (mean p-distance from Eastern lineage 1.0%). Within-region
genetic variance was negligible in the Western (mean p-distance
0.18%), Northern (mean p-distance 0.09%) and Mitchell lineages
(mean p-distance 0.00%), although was considerably higher in the
Eastern lineage, mainly driven by divergences in the King Edward
and Drysdale river populations on the northern Kimberley Plateau
(mean p-distance 0.47%). Syncomistes trigonicus formed two deeper
lin ea ges with moder ate suppor t at the base no de (72% of boot str ap
replicates) that cluster according to the Northern and Eastern re-
gions (mean p-distance 1.1%), while the sister taxa S. wunambal fro m
the Mitchell formed a well-supported lineage that was sister to the
neighbouring Northern lineage (97% of bootstrap replicates, mean
p-distance 1.1%) (Figure 2b). In the Eastern lineage, the Drysdale
and King Edward rivers formed two shallower lineages, the lat ter
being poorly supported, although the populations share haplotypes
that implies some degree of mixing. Hannia greenwayi I showed lit-
tle with in -r eg io n di ve rgence amon g th e We stern pop ul at io ns (me an
p-distance 0.36%), while Ha. greenwayi II formed a moderately sup-
ported (8 4% of bo ot strap rep li cates) sister lin eage in the neighb ou r-
ing Northern region (mean p-distance 1.0%) (Figure 2c). Syncomistes
bonapartensis displayed little within-region genetic divergence over
most of the reg io n, be tween the Daly River (i mm ediatel y nor th-ea st
of the study region; Figure 1) and Drysdale populations (mean p-dis-
tance 0.13%) (Figure 2d). However, the King Edward population at
the western boarder of the region formed a moderately well-sup-
ported (78% of bootstrap replicates) and deeply divergent lineage
(mean p-distance 2.4%) that increased the overall within-region
variance to 1.04%.

6
|
SHELLEY Et aL.
Considering all species analysed, the average genetic variation
among LSL regions was at least 2.9x (and up to 7x) the within-re-
gion variance (Table 3). In addition to the intra-specific phyloge-
netic patterns among the widespread groups, 19 of the 22 NWA
te r apo nti d sp eci e s wer e en dem i c to ju s t one of th e de fin e d regi ons
(exceptions are He. jenkinsi, S. trigonicus and L. unicolor) (Table 3).
3.2 | Phylogeographic analysis of HSL regions
We defined 12 HSL sub-regions that we hypothesized would
be largely isolated under current sea levels (Figure 1b) and as-
sessed the propor tion of endemic haplotypes found in each
sub-region versus the proportion shared among sub-regions.
As predicted, there was little sharing of haplotypes across sub-
regions (Figure 2a–d, Table 3). In the Western region, only 13%
of haplotypes were shared among neighbouring HSL sub-regions,
specifically He. jenkinsi from Walcott Inlet and Doubtful Bay, and
Ha. greenwayi from King Sound and Walcott Inlet. In the Northern
region onl y on e ha plotype (7%) wa s shared betwee n th e St Geo rge
Basin and Prince Fredric Harbour HSL sub-regions (Ha. greenwayi
II). The Mitchell region included only one major catchment and all
haplotypes were unique to the region. Finally, only three haplo-
types (10%) in the Eastern region were shared among HSL sub-
regions; S. trigonicus from Deep Bay and Napier Broome Bay, S.
bonapartensis from Napier Broome Bay and Cambridge Gulf, and
He. jenkinsi from Cambridge Gulf and Queens Channel. Of these
six shared hapl otypes, all but one (S. bonapartensis) was on ly foun d
in adjacent catchments.
4 | DISCUSSION
Our study suggest that the NWA landscape represents a predict-
able geographic template for a range of species that has been and is
like ly still influencing patter ns of spe ciation. These results supp or t a
growing body of evidence from genetic and morphological analysis
of the broader Terapontidae family that suggest s a large proportion
of ter a pon tid s in NWA we re de riv ed re cen tly from a se rie s of sp at i o-
temporal vicariant events caused by late Pliocene and Pleistocene
sea-level fluctuations. The combined evidence provides a rare, well-
supported example of sea-level change driving speciation in the
tropics. It lends important empirical support to the hypotheses that
eustatic processes in the tropics worked concurrently with glacial
expansion and contraction in temperate regions to generate large
numbers of freshwater species over small temporal and geographic
scales during the Plio-Pleistocene ice ages (Hewitt, 2000).
4.1 | Phylogeography of north-western Australian
terapontids
Our multi-species, region-wide mitochondrial dataset provides ro-
bust evidence that sea-level change has had a strong influence on
TABLE 1 Cytochrome b sample sizes each species used in our phylogeographic analysis, listed by catchment and low and high sea-level
region. Letters in parentheses refer to river catchment codes in Figure 1. Dashes refer to catchment s where the species occurred but was
not sampled
Low
sea-level
region
High sea-level
region Catchment
Leiopotherapon
unicolor
Hannia
greenwayi I
Hannia
greenwayi
II
Hephaestus
jenkinsi
Syncomistes
bonapartensis
Syncomistes
trigonicus
Syncomistes
wunambal
Weste rn King Sound Fitzroy (A) 2 4 5
Walcott Inlet Isdell (B) — 2 2
Charnl ey (C) — 2 3
Calder (D) 2 2 4
Doubtful Bay Sale (E) 2 4
Glenelg (F) 1 5 4
Northern St George Basin Prince Regent (G) 5 5
Prince Fredrick
H.
Roe (H) 2 4 5 3
Mitchell Walmesly Bay Mitchell (I) 2 3 5
Eastern Deep Bay King Edward (J) 2 7512
Napier Broome
Bay
Drysdale (K) 2 5 4 8
Koolama Bay K ing George (L) 3
Cambr idge Gul f D urack (N) — 1 5
Pentecos t (O) 2 6 2
Ord (P) 2 5 4
Queens
Channel
Victoria (Q) — 3 —

|
7
SHELLE Y Et aL.
patterns of connectivity among NWA terapontids. The high degree
of genetic similarity among our hypothesized LSL regions, relative
to among-region variabilit y, is most easily explained by predicted
connections during lowered sea levels rather than by chance dis-
persal which would be expected to exhibit less frequent patterns of
connectivity and only among closely situated rivers. Furthermore,
the low proportion of shared haplotypes among catchments not
connected by the same receiving waters (e.g. estuaries or bays) or
floodplains indicates that dispersal is limited over far smaller spatial
scales under HSL conditions. A high level of current genetic subdivi-
sion among NWA rivers, particularly those that drain the northern
Kimberley Plateau, has been further corroborated for a number of
freshwater fish taxa (Huey, Cook, Unmack, & Hughes, 2014; Phillips
et al., 20 09). Herein, we describe the generalized pat terns observed
within and between the four identified ‘widespread’ species/cryptic
species pairs with either region-wide or inter-regional distributions.
The comparatively high degree of genetic similarity in the large
Eastern region (10 major catchments) obser ved in widespread He.
jenkinsi and S. bonapartensis matches the predicted connections
during lowered sea levels when all of the region's catchments drained
into a shallow coastal lake (Lake Bonaparte) (Figure 1a; Yokoyama
et al., 2001). Under HSL conditions, the King Edward, Drysdale
and King George rivers drain north off the Kimberley Plateau into
the Timor Sea, rather than east into what is now the Cambridge
Gulf, severely limiting the opportunity for connectivity with other
Eastern region catchments. This would explain the higher degree
of genetic divergence observed among populations in these rivers
and the presence of narrow-range endemics that have sister species
in catchments to the east (S. rastellus, Shelley, Swearer, et al., 2018;
Melanotaenia gracillis, Unmack, Allen, & Johnson, 2013). In particular,
the King Edward River would be the last river to connect to the Lake
Bonapar te palaeo-drainage basin as sea levels lowered and would be
the first to disconnect as they rose again. The population of S. bona-
partensis there exhibited unusually high cytb divergence relative to
the other Eastern region catchments but fell just short of the cryp-
tic species criterion set out in Shelley, Swearer, et al. (2018) based
on nDNA markers. Regardless, the King Edward population appears
to be progressing towards speciation. Additional to the genetic evi-
dence, the north coast catchments contain fragmented populations
TABLE 2 Mean cytochrome b p-distances (as percentages)
within and among low sea-level region lineages for widespread
terapontid species
Low sea-level
region
Eastern
(J–Q)
Mitchell
(I)
Northern
(G–H)
Western
(A–F)
Hephaestus jenkinsi
Eastern (0.47 )
Mitchell 1.0 (0.00)
Northern 2.0 2.4 (0.09)
Western 1.7 2.0 2.7 (0.18)
Leiopotherapon unicolor
Eastern (0.40)
Mitchell 0.3 (0.00)
Northern 0.4 0.3 (0.00)
Western 0.4 0.5 0.5 (0.30)
Syncomistes trigonicus + Syncomistes wunambal*
Eastern (0.45)
Mitchell 1.1* (0.07)
Northern 1.1 1 .1* (0.11)
Western
Hannia greenwayi I + Hannia gre enwayi II*
Eastern
Mitchell
Northern (0.00)
Western 1.0* (0.36)
Syncomistes bonapartensis
Eastern (1.04)/
(0.13**)
Mitchell
Northern
Western
Note: Mean within-lineage p-distance is shown in brackets on the
diagonal. Comparisons among closely related sister species, rather than
within species, are marked with an (*). The results for S. bonapartensis
marked with (**) refer to within-region variation calculations with the
outlying King Edward River lineage removed. The capital letters in
bracket s following low sea-level region names refer to specific rivers
depicted in Figure 1
TABLE 3 Summary of low sea-level and high sea-level region comparisons with measures of among- and within-region genetic variability
for widespread terapontid species, based on the cytochrome b dataset. Results for average within and among low sea-level (LSL) region
variance are presented as the average variance across all widespread species analysed, followed by the number of widespread species
analysed and the range of values in brackets. The outlying population of S. bonapartensis from King Edward River was removed from the
genetic variance analysis of the Eastern region as it obscured the dominant pattern. No. haps. shared refers to number of haplotypes that are
shared among the hypothesized high sea-level (HSL) sub-regions (Figure 1b) that are nested within the corresponding LSL region
Low sea-level
region No. rivers
Avg. % within LSL
region variance
Avg. % variance among
LSL regions
Endemic taxa in LSL
regions
No. haps. shared among
HSL sub-regions (%)
Eastern 60.36 (3; 0.13–0.47) 1.05 (2; 1–1.1) 9/1 2 3/30 (10)
Mitchell 10.02 (2; 0.00–0.07) 1.40 (2; 1–2.4) 1/3 N/A
Northern 20.05 (3; 0.00–0.11) 1.80 (3; 1–2.7) 5/8 1/14 (7)
Western 60.28 (2; 0.18–0.36) 1. 85 (2; 1–2.7) 4/6 2/1 5 (13)

8
|
SHELLEY Et aL.

|
9
SHELLE Y Et aL.
of M. nigrans and M. exquisita that are found in the Victoria and Daly
rivers to the far east of the Lake Bonapar te palaeo-catchment (Allen,
Midgley, & Allen, 2002). Given these findings, it seems likely that
the Lake Bonaparte drainage basin provided a periodic freshwater
connection for aquatic communities across the entire Eastern region
and potentially as far as the Daly River outside our study region.
Overall, the Eastern region contains nine endemic terapontids (of
12) and fifteen endemic fish species (of 50).
The Mitchell and Northern regions are isolated in the nor th of
the Kimberley Plateau. Although they are closely situated and lie
between the Eastern and Western regions, their genetic distinctive-
ness is high, matching our predictions. The LSL inter-regional bar-
riers that define these regions were each attributed to the rugged
topology and heterogeneous nature of the continental shelf edge
along the nor thern Kimberley coastline. These regions also have a
rich endemic fauna relative to the small number of streams they en-
compass. The Nor thern region (t wo major catchments) has five en-
demic terapontids (of 8) and nine endemic fish species (of 22), while
the Mitchell region (one major catchment) has at least one endemic
terapontid (of 3) and two endemic obligate freshwater fish species
(of 10). An isolated population of A . percoides occurs in the lower
Mitchell River, but its taxonomic status has not been assessed using
genetics.
As predicted, a distinct genetic break was observed between the
neighbouring Western and Northern regions. Although the Western
region is comparatively large (eight major catchments), genetic sim-
ilarity in the region is high. The region also contains a rich endemic
communit y, including four endemic grunters (of 6) and nine endemic
obligate freshwater species (of 29).
4.2 | Allopatric origins and secondary contact
Our overarching hypothesis was that geographic isolation of tera-
pontid lineages during the Plio-Pleistocene sea-level heights trig-
gered the onset of reproductive isolation and consequent speciation
within the family. Previous molecular clock analysis indicated that
75% of speciation events within Kimberley endemic terapontid spe-
cies occurred since the late-Pliocene (<3 Ma; Shelley, Swearer, et al.,
2018), with this period marking a shift to a signif icantly higher dive r-
sification rate in the terapontids (Shelley, Unmack, et al., 2019). Our
phylogeographic analysis provides additional insight into the process
behind this divergence pattern. The results suggest that this period
would have been charac terized by cycles of fine-scale isolation,
which would have promoted allopatric speciation, followed by in-
creased connec tivit y among catchments that would have promoted
broader dispersal and led to secondar y contact among allopatrically
derived congeneric s. In each case of sympatry, the contac t zones
included representatives of independently derived lineages rather
tha n sister taxa, thus providin g st ro ng evid en ce for al lo patric sp ec ia-
tion followed by secondary contact, rather than sympatric specia-
tion (Shelley, Swearer, et al., 2018).
The identification of fixed allozyme differences between spe-
cies in each of the three sympatric species groups tested, and/
or the presence of conserved and highly distinct morphological
features among all sympatric groups (Shelley et al., 2017; Shelley,
Swearer, et al., 2018; Vari, 1978), indicates that they have largely
remained reproductively isolated in their zone of secondary con-
tact (Adams, Raadik, Burridge, & Georges, 2014). Nevertheless,
introgressive hybridization was detected among all sympat-
ric species, except S. bonapartensis and He. epirrhinos (Shelley,
Swearer, et al., 2018). Both S. bonapartensis and He. epirrhinos
represent early-branching lineages of the NWA Syncomistes and
Hephaestus groups and have had longer to develop stronger re-
productive barriers, mediated by either prezygotic (genetic) or
post-zygotic (behavioural) processes, than their more recently
derived congeners (Scribner, Page, & Bartron, 2000). Overall,
naturally occurring introgression was found to be fairly common
within NWA terapontids, and further observations on the wider
Hephaestus genus suggest that it is prevalent in the family (Pusey
et al., 2016). Regardless, it was observed to occur at far higher
frequencies within sympatric members of the recently evolved
Syncomistes complex (Shelley, Swearer, et al., 2018). This intro-
gression appeared to be recent as each pairing mostly shared one
individual haplotype. The high frequency of mtDNA introgression,
contrasting with highly conserved morphologies and fixed allo-
zyme differences, indicates that significant, although incomplete,
reproductive isolating barriers have developed in these species
(Gay, Crochet, Bell, & Lenormand, 2008).
4.3 | Vicariant speciation and biodiversity in NWA
The combined evidence suggests that NWA terapontids are at dif-
ferent st ages of allopatric divergence and speciation, a mosaic likely
caused by the same vicariant processes. As such, this study sup-
ports the hypothesis that changing sea levels during late Pliocene
and Pleistocene glacial cycles are a key driver of speciation and
distributional patterns in the region. Alternative explanations for
the patterns of speciation observed here, such as chance disper-
sal events following river capture or flooding followed by founder
event speciation, seem far less likely given the geological stability
of the region (White, 200 0). Also, the patterns of speciation were
high ly pr edi c tab le ba sed on our hyp oth e si ze d low and HS L div isi on s,
FIGURE 2 Maximum likelihood (ML) tree branches, inferred from the full mitochondrial cytochrome b gene dataset, for: (a) Hephaestus
jenkinsi, (b) Syncomistes trigonicus and S. wunambal, (c) Hannia greenwayi I and Hannia greenwayi II, and (d) Syncomistes bonapartensis. The
hypothesized low sea-level region that each lineage was found in is indicated by coloured lines. Samples are labelled by the catchment they
are from (see Figure 1, Table S1). Nodal suppor t is indic ated by bootstrap values (estimated from 1,000 replications). Only bootstrap values
>50 are presented

10
|
SHELLEY Et aL.
whereas random dispersal would not be expected to follow these
patterns.
Comparative phylogeographic analysis matched to molecular
clock estimates showed that a large proportion of NWA terapon-
tids likely arose via vicariance during Plio-Pleistocene glacial cycles,
when fluctuating sea levels caused periodic connection and isolation
among catchments over relatively small spatial scales. These taxa
display varying degrees of genetic and morphological differentiation,
although in each instance of secondary contact, moderate-to-strong
reproductive barriers are apparent, sufficient to maintain species in-
tegrity despite habitat sharing and sporadic (or historic) introgression
(Shelley, Swearer, et al., 2018). Therefore, we can generalize that iso-
lation during interglacial HSL periods occurred at time scales large
enough to induce spe ciation in fishes , as has be en suggested in oth er
taxonomic groups (Avise et al., 1998; Haffer, 1969; Johnson & Cicero,
2004).
The impact of fluctuating sea levels on phylogenetic and com-
munity structure was predictable based on continental shelf geog-
raphy and modelled LSL drainage patterns. This process is likely to
influence speciation and biogeography across all obligate freshwater
communities in NWA and deep mtDNA structuring across the region
has been identified in a number of other fishes (Huey et al., 2014;
Shelley, Swearer, et al., 2018) and crustaceans (De Bruyn, Wilson, &
Mather, 2004). A productive line of future research would include
a comprehensive biodiversity assessment of remaining groups to
compare phylogeographic patterns and detect additional cryptic
species. This would allow generalizations as to the influence of vi-
cariance, driven by fluctuating sea levels, on speciation and distri-
bution of aquatic fauna in the region and, as similar studies mount,
more broadly across the tropics.
ACKNOWLEDGEMENTS
We acknowledge the funding contributions for this project from
the Hermon Slade Foundation, the Holsworth Wildlife Research
Endowment and the Winifred Violet Scott Charitable Trust. We also
thank Aaron Davis and Martin Gomon for their advice regarding data
analysis and interpretation. We thank Gerry Allen, Jon Armbruster,
Michael Baltzly, Joshua Brown, Chris Burridge, Stephen Caldwell,
Adam Fletcher, David Galeotti, Mark Kennard, Adam Kerezsy, Alfred
Ko'ou, Andrew McDougall, Masaki Miya, David Morgan, Tim Page,
Colton Perna, Ikising Petasi, Michael Pusey, and Ross Smith and the
Hydrobiology team for their efforts in helping to collect and/or pro-
vide specimens.
CONFLICT OF INTERESTS
The authors declare that they have no competing interests.
CONSENT FOR PUBLICATION
Not applicable.
ETHICS APPROVAL AND CONSENT TO PARTICIPATE
All work had animal ethics approval from the Research Ethics and
Integrit y office at the University of Melbourne, ID 1,212,470.1. All
collections by the authors were made under Government of Western
Australia, Depar tment of Fisheries permit, Ref. 220/12. Collections
by the authors in National Parks were made under Department of
Parks and Wildlife Permit SF008685 (2012–2013) and SF009877
(2014–2015).
DATA AVAILAB ILITY STATE MEN T
All unique gene sequences used in this analysis are available from
Genbank and the full details of the accession numbers presented in
Table S2.
ORCID
James J. Shelley https://orcid.org/0000-0002-2181-5888
Stephen E. Swearer https://orcid.org/0000-0001-6381-9943
Tim Dempster https://orcid.org/0000-0001-8041-426X
Mark Adams https://orcid.org/0000-0002-6010-7382
Matthew C. Le Feuvre https://orcid.org/0000-0001-9592-5927
Michael P. Hammer https://orcid.org/0000-0002-0981-4647
Peter J. Unmack https://orcid.org/0000-0003-1175-1152
REFERENCES
Adams, M., Raadik, T. A., Burridge, C. P., & Georges, A. (2014). Global
biodiversity assessment and hyper-cryptic species complexes: More
than one species of elephant in the room? Systematic Biology, 63(4),
518–53 3. ht tps://doi .o rg /10 .10 93/s ys bi o/sy u017
Allen, G. R., Midgley, S. H., & Allen, M. G. (2002). Field guid e to the freshwa-
ter fishes of Australia. Perth, Australia: Wes tern Australian Museum.
Apr il, J., Hanne r, R. H., Dion-Côté, A.-M., & Ber nat che z, L . (2013). Glacial
cycles as an allopatric speciation pump in north-eastern American
freshwater fishes. Molecular Ecology, 22(2), 409–422. https://doi.
org /10.1111/me c.12116
Avise, J. C. (2000). Phylogeography: The history and formation of species.
Cambridge, MA: Harvard University Press.
Avise, J. C. (2004). Molecular markers, natural history and evolution (2nd
ed.). Sunderland, MA: Sinauer A ssociates.
Avise, J. C., Walker, D., & Johns, G. C . (1998). Speciation durations and
Pleistocene effects on vertebrate phylogeography. Proceedings of
the Royal Society of London B: Biological Sciences, 265(14 07), 1707–
1712. https://doi.org/10.1098/rspb.1998.0492. Retrieved from
h t t p : / / r s p b . r o y a l s o c i e t y p u b l i s h i n g . o r g / c o n t e n t / 2 6 5 / 1 4 0 7 / 1 7 0 7 .
abstract
Barton, N. H., & Hewitt, G. M. (1985). Analysis of hybrid zones. Annual
Review of Ecology and Systematics, 16, 113–148.
Bo st ock , B . M ., Ad ams, M. , L aure nson , L. J. B., & Au sti n, C. M. (2 0 06).
The molecular systematics of Leiopotherapon unicolor (G ün the r,
1859): Testing for cr yptic speciation in Australia’s most wide-
spread freshwater fish. Biological Journal of the Linnean Society,
87(4), 537–552. http s://doi.o rg /10.1111/j .1095-8 312 .2006.
0 05 8 7. x
Bowman, D. M. J. S., Brown, G. K., Braby, M. F., Brown, J. R., Cook, L. G.,
Crisp, M. D., … Ladiges, P. Y. (2010). Biogeography of the Australian
monsoon tropics. Journal of Biogeography, 37(2), 201–216. https://
doi.org/10.1111/j.1365-2699.2009.02210.x
Burridge, C. P., Craw, D., & Waters, J. M. (2006). River capture, range
expansion, and cladogenesis: The genetic signature of fresh-
water vicariance. Evolution, 60(5), 1038–1049. https://doi.
org /10.1111/j .0 014 -3820 .2 00 6. tb 011 81.x
Byrne, M., Yeates, D. K., Joseph, L., Kearney, M., Bowler, J., Williams, M. A.
J., … Wyrwoll, K.-H. (2008). Birth of a biome: Insights into the assembly
and maintenance of the Australian arid zone biota. Molecular Ecology,

|
11
SHELLE Y Et aL.
17(2 0), 4398–4 417. https: //doi.org/10 .1111/j.1365-294X.200 8.
03 8 99.x
Campton, D. E., & Ut ter, F. M. (1987). Genetic str ucture of anadromous
cutthroat trout (Salmo clarki clarki) populations in the Puget Sound
area: Evidence for restricted gene flow. Canadian Journal of Fisheries
and Aquatic Sciences, 44(3), 573–582. https://doi.org/10.1139/
f87-070
Clark, P. U., & Mix, A. C. (2002). Ice sheets and sea level of the Last
Glacial Maximum. Quaternary Science Reviews, 21(1– 3) , 1–7. htt ps: //
doi.org/10.1016/S0277 -3791(01)0 0118 -4
Colinvau x, P. A ., & De Oliveir a, P. E. (2001). Amazon plant diversity and
climate through the Cenozoic. Palaeogeography, Palaeoclimatology,
Palaeoecology, 16 6(1–2), 51–63. https://doi.org/10.1016/S0031
-0182(00)00201 -7
Coyne, J. A., & Orr, H. A. (2004). Speciation. Sunderland, MA: Sinauer
Associates.
Darriba, D., Taboada, G. L., Doallo, R., & Posada, D. (2012). jMod-
elTest 2: More models, new heuristics and parallel computing.
Nature Methods, 9(8), 772–772. https://doi.org /10.1038/nmeth.
2109
De Br uyn, M., Wilson, J. C., & Mather, P. B. (20 04). Reconciling geogra-
phy and genealog y: Phylogeography of giant freshwater prawns from
the Lake Carpentaria region. Molecular Ecology, 13(11), 3515–3526.
https://doi.org/10.1111/j.1365-294X.2004.02348.x
Dias, M. S., Oberdorff, T., Hugueny, B., Leprieur, F., Jézéquel, C., Cornu,
J.-F., … Tedesco, P. A. (2014). Global imprint of his torical connectiv-
ity on freshwater fish biodiversit y. Ecology Letters, 17(9), 113 0–1140 .
https ://doi.or g/10.1111/e le.12319
Gay, L., Crochet, P.-A., Bell, D. A., & Lenormand, T. (2008). Comparing
clines on molecular and phenot ypic traits in hybrid zones: A window
on tension zone models. Evolution, 62(11), 2789–2806. https://doi.
org /10.1111/j .155 8-564 6. 20 08.00 491 .x
Griffiths, D. (20 06). Pattern and process in the ecological biogeog-
raphy of European freshwater fish. Journal of Animal Ecology,
75(199 0), 734 –751. h tt ps ://doi.or g/10.1111/ j.13 65-2656 .2 00 6.
01094.x
Haffer, J. (1969). Speciation in Amazonian forest birds. Science, 165(3889),
131–137. https://doi.org/10.1126/scien ce.165.3889.131
Hewitt, G. (2000). The genetic legacy of the Quaternary ice ages. Nature,
405(6789), 907–913. https://doi.org /10.1038/35016000
Hope, G., Kershaw, A. P., Kaars, S. V. D., Xiangjun, S., Liew, P.-M., Heusser,
L. E., … Moss, P. T. (2004). Histor y of vegetation and habitat change
in the Austral-Asian region. Quaternary International, 118, 103–126.
htt ps://doi.org/10.1016/S104 0 -6182(03)0 0133 -2
Huey, J. A., Cook, B. D., Unmack, P. J., & Hughes, J. M. (2014). Broadscale
phylogeographic structure of five freshwater fishes across the
Australian Monsoonal Tropics. BioOne, 33(1), 273–287. ht tps://doi.
or g/10 .10 8 6/674 98 4
Johnson, N. K., & Cicero, C . (20 04). New mitochondrial DNA dat a
affirm the import ance of Pleistocene speciation in North
American birds. Evolution, 58(5), 11 22–1130. h tt ps ://doi.
org /10.1111/j .0 014 -3820 .2 00 4.tb0 04 45. x
Knapp, S., & Mallet, J. (2003). Refuting refugia? Science, 300(5616), 71.
Ollier, C. D., Gaunt, G. F. M., & Jurkowski, I. (198 8). The Kimberley Plateau,
Western Australia - A Precambrian erosion surface. Zeitschrift Fur
Geomorphologie, 32(2), 239–246.
Pepper, M., & Keogh, S. J. (2014). Biogeography of the Kimberley, Western
Australia: A review of landscape evolution and biotic response in an
ancient refugium. Journal of Biogeography, 41(8), 1443–1455. https://
doi .org/10.1111/jb i.1 2324
Perdices, A., Bermingham, E., Montilla, A., & Doadrio, I. (2002).
Evolutionary history of the genus Rhamdia (Teleostei: Pimelodidae)
in Central America. Molecular Phylogenetics and Evolution, 25(1), 172–
189. htt ps://doi.or g/10.1016/S1055 -7903(02)002 24 -5
Phillips , R. D., Storey, A. W., & Johnson, M. S. (2009). Genetic struc-
ture of Melanotaenia australis at local and regional scales in the east
Kimberley, Western Australia. Journal of Fish Biology, 74 (2), 437–451.
https ://doi.or g/10.1111/ j.10 95-86 49.2008 .02099.x
Pigeon, D., Chouinard, A., & Bernatchez, L. (1997). Multiple modes of
speciation involved in the parallel evolution of sympatric mor-
photypes of Lake whitefish (Coregonus clupeaformis, Salmonidae).
Evolution, 51(1), 196–205. https://doi.org/10.2307/2410973
Pusey, B. J., Bentley, A ., Burrows, D., Perna , C., Davis, A., & Hughes, J.
(2016). Contrasting and complex evolutionar y histories within the
terapontid grunter genus Hephaestus revealed by nuclear and mito-
chondrial genes. Marine and Freshwater Research, 67(12), 1813–1825.
htt ps://doi.org/10.1071/MF15198
Remington, C. L. (1968). Suture-zones of hybrid interaction between re-
cently joined biotas. In Evolutionary biology (pp. 321–428). New York,
NY: Plenum Press.
Riginos , C. (20 05). Cr yptic vicariance in Gulf of California f ishes
parallels vicariant patterns found in Baja California mam-
mals and reptiles. Evolution, 59(12), 2678–2690. https://doi.
org /10.1111/j .0 014 -3820 .2 00 5.tb00 9 79.x
Rull, V. (2008). Speciation timing and neotropical biodiversity: The
Tertiary-Quaternary debate in the light of molecular phyloge-
netic evidence. Molecular Ecology, 17(11), 2722–2729. https://doi.
org /10.1111/j .136 5-294X. 20 0 8. 03789.x
Sc ri bn er, K. T., Pa ge , K. S. , & Ba rtron , M. L. (20 00) . Hy bri di za tio n in fre sh-
water fishes: A review of case studies and cytonuclear methods of
biological inference. Reviews in Fish Biology and Fisheries, 10 (3), 293–
323. https://doi.org/10.1023/A:10166 42723238
Shelley, J. J., Delaval, A., & Le Feuvre, M. C. (2017). A revision of the
grunter genus Syncom istes (Teleostei, Terapontidae, Syn comistes)
with descriptions of seven new species from the Kimberley re-
gion, northwestern Australia . Zootaxa, 4367(1), 1–103. https://doi.
org /10.11646 /zoota xa.4367.1.1
Shelley, J. J., Dempster, T., Le Feuvre, M. C., Unmack, P. J., Laffan, S.
W., & Swearer, S. E. (2019). A revision of the bioregionalisation of
freshwater fish communities in the Australian Monsoonal Tropics.
Ecology and Evolution, 9(8), 4568–4588. https://doi.org/10.1002/
ece3.5059
Shelley, J. J., Morgan, D. L., Hammer, M. P., Le Feuvre, M . C., Moore, G.
I., Gomon, M. F., … Saunders, T. (2018). A field guide to the freshwa-
ter fishes of the Kimberley. Perth, Australia: Murdoch University Print
Produc tion Team.
Shelley, J. J., Swearer, S. E., Adams, M ., Dempster, T., Le Feuvre, M. C.,
Hammer, M. P., & Unmack, P. J. (2018). Cry ptic biodiversity in the
freshwater fishes of the K imber ley endemism hotspot, northwest-
ern Australia. Molecular Phylogenetics and Evol ution, 127, 843–858.
https://doi.org/10.1016/J.YMPEV.2018.06.032
Shelley, J. J., Unmack, P. J., Dempster, T., Le Feuvre, M. C., & Swearer,
S. E. (2019). The Kimberley, north-western Australia, as a cradle
of evolution and endemic biodiversit y: An example using grunters
(Terapontidae). Journal of Biogeography, 46(11), 2420–2432. https://
doi .org/10.1111/jb i.1368 2
Tschá, M. K., Bachmann, L., Abilhoa, V., & Boeger, W. A . (2017). Past
connection and isolation of catchments: The sea-level changes af-
fect the distribution and genetic variability of coastal freshwater
fishes. Estuarine, Coastal and Shelf Scie nce, 190, 31–39. https://doi.
org/10.1016/J.ECSS.2017.02.030
Unmack, P. J. (2013). Biogeogr aphy. In P. Humphries, & K . Walker (Eds.),
Ecology of Australian freshwater fishes. Melbourne, Australia: CSIRO
Publishing.
Unmack, P. J., Allen, G . R., & Johnson, J. B. (2013). Phylogeny and bioge-
ography of rainbowfishes (Melanot aeniidae) from Australia and New
Guinea. Molecular Phylogenetics and Evolution, 67(1), 15–27. https://
doi.org/10.1016/j.ympev.2012.12.019

12
|
SHELLEY Et aL.
Unmack, P. J., Hammer, M. P., Adams, M., Johnson, J. B., & Dowling, T.
E. (2013). The role of continental shelf width in determining fresh-
water phylogeographic patterns in south-eastern Australian pygmy
perches (Teleostei: Percichthyidae). Molecular Ecology, 22(6), 1683–
1699. ht tps://doi.or g/10.1111/mec .12 20 4
Vari, R. P. (1978). The terapon perches (Percoidei: Teraponidae): A cladis-
tic analysis and taxonomic revision. Bulletin of the A merica n Museum
of Natural History, 159, 175–340.
Voris, H. K . (2000). Maps of Pleistocene sea levels in
Southeast Asia: Shorelines, river systems and time dura-
tions. Journal of Biogeography, 27(5), 1153–1167. h tt ps://doi.
org/10.1046/j.1365-2699.2000.00489.x
Wallis, G. P., Cameron-Christie, S. R., Kennedy, H. L., Palmer, G., Sanders,
T. R., & Winter, D. J. (2017). Inter specific hybridization causes long-
term phylogenetic discordance between nuclear and mitochondrial
genomes in freshwater fishes. Molecular Ecology, 26(12), 3116–3127.
https ://doi.or g/10.1111/m ec .14096
Wallis, G. P., Waters, J. M., Upton, P., & Craw, D. (2016). Transverse alpine
speciation driven by glaciation. Trends in Ecology & Evolution, 31(12),
916–926. https://doi.org/10.1016/j.tree.2016.08.009
Waters, J. M., Rowe, D. L., Apte, S., King, T. M., Wallis, G. P., Anderson,
L., … Burridge, C. P. (2007). Geological dates and molecular rates:
Rapid divergence of rivers and their biotas. Systematic Biology, 56(2),
271–282. https://doi.org/10 .108 0/10635 15070 1313855
Waters, J. M., & Wallis, G. P. (2001). Cladogenesis and loss of
the marine life-history phase in freshwater galaxiid fishes (
Osmeriformes : Galaxiidae ). Evolution, 55, 587–597. https://doi.
org/10.1554/0014-3820(2001)055
White, M. E. (2000). Running down: Water in a changing la nd. Sydney,
NSW, Australia: Kangaroo Press.
Yokoyama, Y., Purcell, A ., Lambeck, K., & Johnston, P. (2001). Shore-line
reconstruction around Australia during the Last Glacial Maximum
and Late Glacial Stage. Quaternary International, 85, 9–18. https://
doi.org/10.1016/S1040 -6182(01)00028 -3
Zwickl, D. J. (2006). Genetic algorithm approaches for the phylogenetic
analysis of large biological sequence datasets under the maximum likeli-
hood criterion. Austin, T X: University of Texas. PhD Thesis.
BIOSKETCH
James Shelley conducted this research as a PhD student in the
School of Biosciences at the Universit y of Melbourne. His thesis
investigated biodiversity, evolution and conservation of freshwa-
ter fishes in the Kimberley region of Western Australia. He is now
based at the Arthur Rylah Institute for Environmental Research in
Melbourne, Australia.
Authors’ contributions: J.J.S., S.E.S, T.D. and M.C.L. conceived
the ideas; J.J.S., M.C.L, P. J.U., S.E.S., T.D. and M.P.H collec ted
the data; J.J.S. conducted the analysis under the supervision of
P.J.U. and M.A.; J.J.S. led the writing and all authors contributed
through editing.
SUPPORTING INFORMATION
Additional supporting information may be found online in the
Supporting Information section.
How to cite this article: Shelley JJ, Swearer SE, Dempster T,
et al. Plio-Pleistocene sea-level changes drive speciation of
freshwater fishes in north-western Australia. J Biogeogr.
2020;00:1–12. ht tps: //doi. or g/10.1111/ jb i.1385 6