Content uploaded by James J. Shelley
Author content
All content in this area was uploaded by James J. Shelley on Nov 24, 2021
Content may be subject to copyright.
Freshwater Biology. 2021;00:1–15. wileyonlinelibrary.com/journal/fwb
|
1© 2021 John Wiley & Sons Ltd.
Received: 4 March 2021
|
Revised: 6 Oc tober 2 021
|
Accepted: 18 Oc tober 2021
DOI: 10.1111/fwb.1384 4
ORIGINAL ARTICLE
Landscape context and dispersal ability as determinants of
population genetic structure in freshwater fishes
James J. Shelley1,2 | Owen J. Holland3,4 | Stephen E. Swearer1 |
Timothy Dempster1 | Matthew C. Le Feuvre1 | Craig D. H. Sherman3,4 |
Adam D. Miller3,4
James J. S helley and Owen J . Holland Co- lead auth ors
1School of BioScie nces, University of
Melbou rne, Parkville, Victoria, Australia
2Depar tment of Environment, L and, Wate r
and Planning, Arthur Rylah Institute for
Environmental Research, Heidelberg,
Victoria, Australia
3School of Life and Environmental S ciences,
Centre for Integr ative Ecology, Deakin
University, Warrn ambool, Victoria, Australia
4Deakin Genomics Cent re, Dea kin
University, Geelong, V ictor ia, Australia
Correspondence
Adam D. Mill er, Deakin U niversity, Scho ol of
Life and Environmental Sciences, Centre for
Integrative Ecology, Warrna mbool , Victoria
3280, Aus tralia.
Email: a.miller@deakin.edu.au
Funding information
Winifred Violet Scott Charit able Trust
Abstract
1. Dispersal is a critically important process that dictates population persistence,
gene flow, and evolutionary potential, and is an essential element for identify-
ing species conservation risks. This study aims to investigate the contributions
of dispersal syndromes and hydrographic barriers on patterns of population con-
nectivity and genetic structure in fishes occupying the particularly rugged and
fragmented landscape of the Kimberley Plateau, Western Australia.
2. We assessed population genetic structure between three neighbouring catch-
ments (the Mitchell, King Edward, and Drysdale rivers) in three congeneric groups
of freshwater fishes that exhibit varied dispersal syndromes within and among
groups: (1) Melanotaenia australis and M. gracilis; (2) Syncomistes trigonicus and
S. rastellus; and (3) Hephaestus jenkinsi and H. epirrhinos. Within each species we
sam p l e d th e up per, middle, an d lo wer reach e s of ea ch catch m e n t an d assess e d pa t-
terns of gene flow between and within catchments using microsatellite markers.
3. Our results suggest that contemporar y connectivity between catchments is
greatly limited or absent in all study species, regardless of their dispersal syn-
dromes. However, gene flow within catchments varied in line with predicted dis-
persal potential, with poor dispersers exhibiting limited gene flow and significant
genetic structuring.
4. We conclude that the rugged landscape and historical habitat isolation has con-
tributed to patterns of population fragmentation among fish populations from
different river catchments. However, it appears dispersal syndromes influence
connectivity and gene flow within catchments, where landscape constraints are
not as pervasive.
5. This study presents a comparative population genetic analysis of freshwater fishes
with differing dispersal syndromes and colonisation ability. Our findings pro-
vide new insights into factors shaping patterns of biodiversity on the Kimberley
Plateau, and the evolutionary uniqueness of fish communities from different river
catchments draining the plateau. More broadly, they highlight the importance of
2
|
SHELLEY Et aL.
1 | INTRODUCTION
Dispersal is a critical process that dictates the evolutionar y tra-
jector y and conservation risk of all species (Frankham et al., 2010;
Ronce, 2007). As such, understanding a species’ dispersal abilit y
and the causes and consequences of it s dispersal are central to bio-
diversit y conservation and studies of biogeography, ecology, and
evolution (Clobert et al., 2012). Dispersal and population genetic
structure are inversely related in theory, with enhanced disper-
sal typically leading to greater connectivity between populations,
and geographically larger and more genetically diverse populations
buffered from drift and inbreeding processes (Bohonak, 1999;
Luiz et al., 2013; Slatkin, 1987). Conversely, limited dispersal often
leads to geographically smaller, less connected, and genetically di-
verse populations that are more vulnerable to drift and inbreed-
ing processes, and risks of maladaptation and extinction (Lawton
et al., 2011; Templeton et al., 1990). Studies suggest that dispersal is
the reflection of syndromes emerging from trade- offs bet ween, and
covariation amongst, traits related to life history, ecological niche
width, and speed and endurance (Bradbury et al., 2008; Comte &
Olden, 2018; Stevens et al., 2014). Although the physical environ-
ment broadly defines corridors for dispersal, dispersal syndromes
provide a powerful way of inferring a species’ relative ability, and/
or propensity, to capitalise on opportunities to migrate, enhancing
patterns of population connectivity and opportunities for colonis-
ing new environment s (Coates et al., 2019; Comte & Olden, 2018;
Riginos et al., 2014). Thus, understanding the relative influence of
dispersal syndromes on patterns of population connectivit y, genetic
structure, and range size is important for informing modern conser-
vation frameworks aimed at preserving patterns of endemism and
maximising evolutionary potential (Frankham et al., 2010, 2017).
Freshwater habitats by their very nature exhort substantial dis-
tributional constraints on resident biota. Dispersal within and among
drainage basins often is limited by a la ck of physical habitat connec t-
edness resulting from geological landscape features and instream
barriers (Unmack, 2013). Opportunities for dispersal often are dic-
tated by rare flood events that influence dispersal corridors over land
or via coastal freshwater plumes, whereas drought events constrain
habitat connectivity as a consequence of reduced environmental
flows (Closs & Lake, 1996; Thacker et al., 2007). The traits that best
promote dispersal and subsequent colonisation can be broadly cat-
egorised as those facilitating movement, such as life- history traits,
and morphological trait s that influence swim speed and endurance,
and those facilitating establishment in novel habitats such as eco-
logical niche generalisation (Comte & Olden, 2018). For example,
fish species with large body sizes and caudal fins (proxies for lon-
gevity and swimming ability) often disperse further than those that
do not (Comte & Olden, 2018; Le Feuvre et al., 2016). Fur thermore,
those considered niche generalist s have a reduced likelihood of
environment– phenotype mismatches and greater capacity for colo-
nising new environments (Lawton et al., 2011; Marshall et al., 2010).
Consequently, comparative population genetic studies involving
assessments of gene flow patterns across species with a variety of
dispersal syndromes provide a strong framework for investigating
the influence of dispersal syndromes on patterns of genetic varia-
tion at species, community, and landscape scales. Indeed, studies of
this nature have been vital in enhancing our knowledge of the evo-
lutionary processes shaping patterns of biodiversity and habitat use
in freshwater ecosystems (Chester et al., 2015; Hughes et al., 2013;
Prunier et al., 2018).
The Kimberley Plateau is a vast, rugged highland area in north-
western Australia, renowned for its dramatic landscapes, cultural
values, and unique and distinctive wildlife communities; it has more
endemic plant and animal species than anywhere else on the con-
tinent (Pepper & Keogh, 2014). Situated in Australia's monsoonal
tropics, the region's biotic communities are strongly influenced by
defined wet and dry seasons that drive contrasting cycles of disper-
sal and productivity across the landscape (Pepper & Keogh, 2014).
The freshwater environments on the Kimberley Plateau present an
extreme example of naturally isolated freshwater habitats. The riv-
ers flow through the sandstone plateau from their headwaters to
the ocean, meaning lowland habitats such as estuaries and flood-
plains are not common or extensive, and freshwater connections
between catchments are expected to be limited in many species
(Kennard et al., 2010; Shelley et al., 2019). Fur thermore, many of
the rivers are interrupted by high waterfalls (>10 m) that act as hard
barriers to upstream movement in many species (Shelley, Morgan,
et al., 2018a). The rivers harbour many narrow- range endemics
hypothesised to have radiated as a result of historical habitat iso-
lation (Shelley, Swearer, et al., 2018b; Shelley et al., 2020; Young
et al., 2011). Despite limited contemporar y connections, historical
opportunities for dispersal between drainages likely occurred during
glacial periods when sea levels were lower and rivers could spread
and coalesce beyond the plateau (Shelley et al., 2020). Variable spe-
cies distributions suggest the ability to capitalise on present or past
dispersal opportunities is idiosyncratic (Shelley et al., 2019, 2020).
However, our understanding of the relative contributions of intrinsic
(e.g., species- specific traits) and extrinsic (e.g., landscape/riverscape
features) factors to patterns of genetic diversity, biogeographical
structuring, and range size on the Kimberley Plateau is limited.
accounting for dispersal- related traits when planning management and conserva-
tion strategies.
KEYWORDS
dispersal syndromes, Kimberley, Melanotaeniidae, microsatellites, Terapontidae
|
3
SHELLE Y Et aL.
Freshwater fish communities exemplify the uniqueness of
the Kimberley Plateau fauna with 35 of 68 species (51%) being
endemic to the rivers draining the plateau (Shelley, Morgan,
et al., 2018a). Notably, 22 of these species are rest ric ted to one or
two river systems. These highly range- restricted communities are
thought to be amongst the most imperilled in Australia (Le Feuvre
et al., 2016). However, there are notable exceptions with several
congeneric relatives of narrow- range species being widespread
across major r iver catchments on and beyond the plateau (Shelley,
Morgan, et al., 2018a). For example, the genera Hephaestus,
Syncomistes, and Melanotaenia each have a broadly distributed
species (H. jenkinsi, S. trigonicus, and M. australis), and short- range
endemic species that are effectively restricted to the Drysdale
River (H. epirrhinos, S. rastellus, and M. gracilis) (Shelley, Swearer,
et al., 2018b; Unmack et al., 2013). These genera differ substan-
tially in dispersal syndromes related to dispers al potential (i.e., life
history and swimming abilit y) (Davis et al., 2020; Shelley, Morgan,
et al., 2018a), and these congeneric species exhibit substantial
ecological niche differences (Le Feuvre et al., 2021). Although
these factors are expected to contribute to species- level differ-
ences in dispersal and colonis ation abilit y an d, thus, dif ferences in
population connectivity, genetic structure, and range extent, this
has not be en formally tes te d. Su ch inf or mati on is n ee de d to be tter
understand the evolutionar y processes shaping biodiversity and
to inform conservation planning in the region.
This study aims to determine if differences in dispersal syn-
dromes influence cotemporary patterns of population connectivity,
genetic structure, and geogra ph ic al range size in fish species per sist-
ing in the fragmented Kimberley Plateau landscape. To achieve this,
we undertook comprehensive population genetic analyses of con-
generic species pairs from three genera (Hephaestus, Syncomistes,
and Melanotaenia), including bro adl y distri buted and short- ran ge en-
demics, across three closely situated river catchments (the Mitchell,
King Edward, and Drysdale rivers) that drain the northern Kimberley
Plateau. Each of these rivers are dissected by a large (10−40 m high)
waterfall on its main stem that provides an ideal model to test the
relative ability of the different species to capitalise on rare disper-
sal oppor tunities within as well as bet ween catchments. Specifically,
our objectives were to: (a) assess spatial patterns of gene flow and
genetic structure within and between catchments; (b) explore the
role of landscape features (i.e., catchment boundaries and waterfalls)
on popul at io n ge ne tic str uc t ur e of eac h sp ecies; an d (c) con tr ast out-
puts from genetic analyses with our current understanding of spe-
cies dispersal potential and ecological niche differences to evaluate
the relative roles of intrinsic (e.g., species- specific traits) and extrin-
sic (e.g., landscape/riverscape features) factors in shaping patterns
of genetic diversity and range size.
We tested the following hypotheses about the processes in-
fluencing the biogeographical structuring of freshwater fish com-
munities on the Kimberley Plateau: (a) patterns of gene flow and
population genetic structure will be linked to species dispersal po-
tential (i.e., species with traits associated with reduced dispersal po-
tential are expected to show limited gene flow and greater levels of
genetic structuring within and between catchments); (b) differences
in gene flow within congeneric species pairs are linked to ecologi-
cal niche width (i.e., narrow- range ecological niche specialists will
exhibit greater genetic structuring compared to wide- range, gen-
eralist congenerics in sympatr y); and (c) the complex Kimberley
Plateau landscape greatly limits contemporary gene flow within and
between catchments regardless of dispersal syndromes. We expect
that all species will exhibit significant genetic structuring between
catchments and within catchments owing to the presence of major
waterfall barriers.
2 | METHODS
2.1 | Study catchments and sites
This study focussed on the Mitchell, King Edward, and Drysdale riv-
ers that flow of f the no rth co as t of the Kimb er ley Plateau. These riv-
ers share long inland catchment boundaries, although smaller coastal
catchments lie between the river mouths. The catchments share
several wide- range species, suggesting a degree of connectivity.
Each of the rivers has a large water fall on the main stem that
prevent upstream migration in many species (Shelley, Morgan,
et al., 2018a). On the Mitchell River, Mitchell Falls consists of a series
of four falls with a total height of c. 40 m, with the largest single drop
being c. 15 m (Figure 1). King Edward Falls on the King Edward River,
and Solea Falls on the Drysdale River stand c. 13 m and c. 10 m high,
respectively (Figure 1).
Sampling sites were chosen below (lower catchment popula-
tions) and above (middle and upper catchment populations) each
major waterfall. River access was the major determinant of site se-
lection in the remote region which has little road access. Samples
were collected from 20 sites within the upper, middle, and lower
river locations of each of the three catchments (Figure 1). At each
site, between six and 30 individuals (average 20) were euthanised
in 40 0 mg/L clove oil in water, then muscle tissue was biopsied and
stored in 100% ethanol (Table 2).
2.2 | Species and traits
We investigated three congeneric species pairs with variable eco-
logical traits and range sizes to determine whether these differing
characteristics correspond with pat terns of gene flow and genetic
structure. These included the Western and longnose sooty grunt-
ers (H. jenkinsi and H. epirrhinos) and longnose and Drysdale grunters
(S. trigonicus and S. rastellus) from the family Terapontidae, and the
Western and slender rainbowfishes (M. australis and M. gracilis) from
the family Melanotaeniidae.
A dataset of quantitative and qualitative observations of life-
history trait variation (fecundity, egg size, body size, size at mat-
uration, and spawning habits), swimming ability (caudal fin aspect
ratio), and ecological specialisation (habitat and diet) was compiled
4
|
SHELLEY Et aL.
for each species from published literature (Table 1). Quantitative
variables included: (a) maximum body size recorded for species; (b)
caudal fin aspect ratio (aspect ratio = h2/s; h = height of caudal
fin; s = surface area of fin); (c) female standard length at matura-
tion (calculated as mean or, if not repor ted, median or minimum
length at maturation); (d) the mean diameter of mature intraovar-
ian oocytes; (e) mean fecundity; and (f) maximum recorded fecun-
dity. Habitat and diet dat a were taken from a recent investigation
of specialisation in each of the study species and the Kimberley
fish community more broadly (Le Feuvre et al., 2021). The study
qualitatively assessed habitat use for each species and included
landscape characteristics (n = 5; e.g., watercourse type, catchment
position, permanence) at the site level and structural traits at the
microhabitat level (n = 6; e.g., physical habitat, substrate, vegeta-
tion, slope) across all or most of each species’ range. Diet content
was attributed to 23 categories (e.g., Ephemeroptera lar vae, fishes,
ter restr ia l, or aquatic veget ation) and cha ra cterise d at dif ferent on-
togenetic stages.
2.3 | Colonisation potential predictions
We used scientific literature (discussed in Section 2.1) and expert
opinion to conduct a semi- quantitative assessment of attributes
considered to influence colonisation potential (i.e., potential to dis-
perse and establish in new, distant habitats). Our predictions for
each study species are visually presented in Figure 2. The position
of each species along each axis is based upon the relative suitabil-
ity of life- history and morphological characters (dispersal potential
axis), and habit at and dietary niche width (niche generalisation axis)
for colonisation of new and distant habitats (Comte & Olden, 2018).
As the relationship between life- history and morphological traits
and dispersal ability is linear (Comte & Olden, 2018), we weighted
species along that axis accordingly. For instance, species with the
smallest body size, lowest fecundity, largest egg size, and so on were
placed to the left, whereas those with increasingly large body size
and fecundity, decreasingly small egg size, and so on were place fur-
ther to the right.
FIGURE 1 The Kimberley Plateau study region in northwestern Australia. Study catchment s and sites are depicted, distinguished by
colour and shape: Mitchell River (blue, diamonds), King Edward River (purple, circles), and Drysdale River (yellow, squares). River locations
(upper, middle, and lower) are indicated by different shading of each catchment's colour and infill of the site markers. The major waterfall
barrier present on the mainstream of each catchment is pictured. Inset is a broader map of Australia, with a red square highlighting the study
region
|
5
SHELLE Y Et aL.
Each species’ weighting on the niche generalisation axis
was based predominantly on analyses presented in Le Feuvre
et al. (2021). Specifically, we used their calculations of dietary niche
breadth, on a scale of zero (specialist) to one (generalist), and their
principal component analysis showing the relative degree of habi-
tat generalisation between each species on to guide us. However,
we also considered the influence that prey and habitat type would
be expected to have on dispersal ability. For instance, species that
specialised on specific, but abundant food resources (e.g., fish, in-
ver teb rates, and algae) were score d only slight ly lower than general-
ists as resource availability would not be a strong limiting factor, but
some spatial and temporal variation would be expected. Regarding
habitat niche generalisation, species that exhibited preferences for a
certain substrate and/or macrohabitat type were weighted towards
the specialist end of the axis. However, those that prefer pool, per-
manent, and upper catchment habitat were weighted more heavily
as aversion to fast- flowing, temporary or lower catchment reaches is
expected to negatively influence ability and propensity to disperse.
The Hephaestus species pair are both moderately large species
(42 − 45 cm standard length; SL) of grunters that are endemic to
the Kimberley freshwater fish bioregion (herein referred to as the
Kimberley) (Shelley, Morgan, et al., 2018a). Hephaestus jenkinsi is
widespread throughout the Kimberley, whereas H. epirrhinos is
restricted to the Drysdale River and a tributary of the lower King
Edward River (Shelley, Morgan, et al., 2018a). They mature late and
lay large numbers (hundreds of thousands) of small (1.50– 1.65 mm),
non- adhesive demersal eggs during a single wet season spawning
event (Davis et al., 2020; Shelley, Morgan, et al., 2018a). Based on
these characters, both species are predicted to have high dispersal
potential. On the one hand, H. jenkinsi also is an abundant species
that exhibits a generalist diet and habitat preferences and as such, is
predicted to have high colonisation potential (Le Feuvre et al., 2021).
The two species have similar caudal fin aspect ratios (1.64) suggest-
ing similar swimming ability. On the other hand, H. epirrhinos is found
in low abundance compared to H. jenkinsi and exhibits an affinity
towards deeper permanent habitat and has a specialist carnivorous
diet, although this is an abundant resource (Le Feuvre et al., 2021;
Shelley, Morgan, et al., 2018a). Therefore, we hypothesise that it has
a moderate colonisation potential.
The Syncomistes pair are a small- to medium- sized (15– 17 cm
SL) species of grunters that are endemic to the northern Kimberley
Plateau within the Kimberley (Shelley, Morgan, et al., 2018a). The
moderately widespread S. trigonicus is found in the Drysdale, King
Edward, and Roe rivers, whereas S. rastellus is restricted to the
Drysdale River (Shelley, Morgan, et al., 2018a). They mature moder-
ate ly early (60– 70 mm) and lay moder at e nu mb ers (ten s of tho us an ds)
of medium- sized (2– 3 mm) non- adhesive demersal eggs during a sin-
gle wet season spawning event (Davis et al., 2020; Shelley, Morgan,
et al., 2018a). Based on these characters, both species are predicted
to have moderate dispersal potential. The caudal fin aspect ratio of
S. rastellus is 2.50 compared to 1.78 in S. trigonicus suggesting that
S. rastellus has a greater swimming ability. Syncomistes trigonicus is an
abun d a n t sp e c ies , wi t h ge n e r alis t ha b i t at requi r e m e n t s an d a pr i m a r i ly
algivorous diet, although it displays a higher degre e of omnivor y than
S. rastellus. Conversely, S. rastellus occurs in low abundance, is a spe-
cialist algivore, and prefers deep pool habitat over rocky substrate,
particularly in the upper catchment (Le Feuvre et al., 2021; Shelley,
Morgan, et al., 2018a). We hypothesise that S. trigonicus has high and
S. rastellus moderate colonisation potential overall.
The Melanotaenia pair are small (6– 8 cm SL) species of rainbow-
fish. Melanotaenia australis is widespread throughout the Kimberley
and extends into the neighbouring Pilbara and Northern freshwater
fish biogeographical provinces, whereas M. gracilis is found only in
the Dr ysdale River and one small creek within the King Edward River
catchment (Shelley, Morgan, et al., 2018a). The caudal fin aspect ratio
of M. australis (1.63) is greater than that of M. gracillis (1.44) suggest-
ing that M. australis has greater swimming ability. They are both highly
abundant, mature early (23– 27 mm) and lay small numbers (tens) of
small (0.92– 0.94 mm), adhesive eg gs onto vegetation throughout the
year (Davis et al., 2020; Shelley, Morgan, et al., 2018a). As such they
both exhibit low dispersal potential. However, M. australis exhibits
generalist diet and habitat requirements and as such we expec t it has
moderate colonisation potential. Conversely, M. gracilis is a specialist
insectivore throughout its adult life and has a strong affinity towards
sa n dy su bstr ates in pe rma n ent st rea m secti ons (Le Fe u vre et al. , 202 1) .
As such, we hypothesise that it has a low colonisation potential.
2.4 | DNA extraction and microsatellite genotyping
We adopted widely published laboratory protocols for DNA extrac-
tion and genot yping and, given limited journal space and that we are
presenting a large multi- species dataset, we present only key details
here and provide the full methods in the Supporting Information as
Methods S1. In brief, total genomic DNA was extracted using a modi-
fied Chelex protocol (Walsh et al., 1991). DNA samples were geno-
typed at 5– 16 microsatellite loci, including novel markers developed
for Hephaestus and Syncomistes by next- generation DNA sequencing
following the approach of Miller et al. (2013), and markers previously
developed for Melanotaenia by Mondol et al. (2014). POWSIM v4.0
(Ryman and Palm 2006) was used for evaluation of the α error and
statistical power of the microsatellite loci for accurately detecting dif-
ferent levels of FST. The statistical powe r of the microsatellite markers
to detect various levels of true FST values between populations was
tested taking into account the sample sizes, number of loci, and aver-
age allele frequencies of each dataset. All analyses showed that the
microsatellite markers for each dataset will detect a true FST of 0.01 or
larger with a probability of ≥99%, and an FST as low as 0. 005 with 95%
confiden ce interval (CI). The alpha er ro r (i.e., the pro babil it y of obtain-
ing false significances when the true FST = 0) in each case was zero.
2.5 | Population genetic analyses
Frequency variation of nuclear microsatellite alleles among pop-
ulations was used to assess patterns of gene flow and genetic
6
|
SHELLEY Et aL.
structuring among populations for each of the six focus species.
Genodive v3.01 (Meirmans, 20 09) was used to calculate deviations
from Hardy– Weinberg equilibrium (HWE), Weir and Cockerham's
measure of FIS, global estimates of population differentiation FST
with 95% confidence limits (Weir & Cockerham, 1984), population
pairwise measures of FST and Jost's Dest (Jost, 2008), observed and
expected heterozygosity, allelic richness, and average numbers of
alleles per locus, all with significance determined using 1,000 per-
mutations (Weir & Cockerham, 1984). Genodive was also used to
perform an analysis of molecular variance (AMOVA) using pairwise
FST as th e di s tan ce me asu re, with 1, 0 0 0 per mut a ti o ns . The mo del for
analysis partitioned variation within and between sampling locations,
and between rivers for broadly distributed species. The sequential
Bonferroni procedure (Rice, 1989) was used to adjust significance
levels when performing multiple simultaneous comparisons.
Estimates of genetic structure for each species group was also
inferred using the Bayesian software package STRUCTURE v2.3.2
(Hubisz et al., 2009). STRUCTURE identifies the number of most
likely distinct genetic clusters, assigning individuals to clusters,
and identifies admixture between them. To determine the number
of genetic clusters (K), five independent simulations were run for
K = 1– 9 (depending on the total number of river reaches sampled
for each species), with a 50,000 iteration burn- in and 500,000
Markov chain Monte Carlo (MCMC) iterations. The analysis was
TABLE 1 Ecological traits of each study species from which their colonisation potential was hypothesised. Sources: 1Shelley et al. (2018a),
2herein, 3Davis et al. (2020), 4Le Feuvre et al. (2021)
Species
Relative range size and
abundance1
Max.
length
(mm)1
Caudal fin aspect
ratio2
Size at maturity –
female (mm)1,3 Spawning period / surface1,3
Egg
Size
(mm)1,3
Fecundit y (average /
maximum)1,3
Diet (presented by ontogenetic stage,
given as size in mm)4Habitat preference4
H. jenkinsi
Widespread and
abundant
450 1.64 120 Wet season / Demersal 1.65 185,000/400,000 Aquatic invertivore (≤40),
Omnivore (>40)
Generalist
H. epirrhinos
Range- restricted and
low abundance
420 1.64 120 Wet season /
Demersal
1.50 100,000/200,000 Aquatic invertivore (≤120),
Macrophagous carnivore (>120)
Deep pools, per manent habitat
S. trigonicus
Moderately
widespread and
abundant
170 1.78 60 Wet season /
Demersal
2.00 20,000/30,000 Meiophagous omnivore (≤40),
Algivore- detritivore / Omnivore (>40)
Generalist habitat t ype, but preference for
permanent habitat
S. rastellus
Range- restricted and
low abundance
150 2.50 70 Wet season /
Demersal
3.00 25,000/32,000 Algivore- detritivore / Omnivore (≤90),
Algivore- detritivore (>90)
Deep pools, rocky subs trate, upper catchment
M. australis
Widespread and very
abundant
80 1.63 23 Year- round /
Vegetation
0.94 15/47 Meiophagous omnivore (all sizes) Generalist
M. gracilis
Range- restricted and
very abundant
60 1.44 27 Year- round /
Vegetation
0.92 6/15 Meiophagous omnivore (≤40)
Insectivore (>40)
Permanent habit at, sandy substrate
FIGURE 2 Predicted colonisation potential of each study
species based on hypothesised dispersal potential (x- axis, low to
high) and degree of niche generalisation (y- axis, low to high). The
position of each species indicates its relative overall dispersal
potential from “low” (red quadrant) to “high” (green quadrant)
|
7
SHELLE Y Et aL.
performed using the admixture model of population structure
(each individual drawing some of their genome from each of the K
populations) and allele frequencies were set as independent among
populations. The most likely K was estimated using Evanno's ΔK
method (Evanno et al., 2005) in STRUC TURE HARVESTER (Earl &
Vonholt, 2012).
Additional to the STRUCTURE analysis, discriminant analysis
of principal components (DAPC) was implemented in the adegenet
2.0.1 package for R (Jombart, 2008; Jombart & Ahmed, 2011) to
compare between the methods and to obtain a graphical depiction
of patterns of genetic structure for each species. The results were
like those of STRUCTURE and did not change our interpretation of
the results. So, for the sake of brevity, we present the details of this
analysis in Methods S1 and Results S1.
Rates of recent migration were estimated between each pair
of sampling locations using a Bayesian algorithm implemented in
BayesAss v3.0.3 (Wilson & Rannala, 2003). This is a commonly
used approach for estimating the strength and directionality of
gene flow in animal and plant systems, including panmictic spe-
cies (Booth Jones et al., 2017; Ferreira et al., 2017). BayesAss esti-
mates migration among locations within the last three generations.
To identify movements among populations, ten independent runs
of 107 MCMC iterations were used following a burn- in period of
107 and a sampling interval of 500 steps. Chains were compared
to a stationar y posterior distribution for convergence by perform-
ing multiple runs with dispersed starting values. The proportion of
individuals that were assigned as migrants (migration rates) and as-
sociated 95% CIs were estimated among each of the sampling loca-
tions. Estimates <5% typically have CIs overlapping zero and were
therefore not reported.
3 | RESULTS
3.1 | Hephaestus
3.1.1 | Wide- range species – – H. jenkinsi
A total of 152 individual H. jenkinsi samples from eight locations
spanning the Drysdale, King Edward, and Mitchell rivers were
genotyped at 14 microsatellite loci (Table 2). Significant deviations
from HWE (p < 0.01) were observed at three H. jenkinsi sample
locations (middle Drysdale River, middle and lower Mitchell River),
which were accompanied by significant inbreeding coefficients
(FIS) indicating a homozygote excess (Table 2). These patterns ap-
pear to be driven by a small number of different loci (1 − 2 loci
for middle and lower Mitchell River), whereas the middle Drysdale
River location showed significant deviations at many (six) loci po-
tentially indicating non- random mating. Estimates for total num-
ber of alleles, allelic richness, and expected heterozygosit y varied
across H. jenkinsi sample locations, with permutation tests indicat-
ing Drysdale River estimates (mean a = 4.00; r = 2.18; HE =0.45) to
TABLE 1 Ecological traits of each study species from which their colonisation potential was hypothesised. Sources: 1Shelley et al. (2018a),
2herein, 3Davis et al. (2020), 4Le Feuvre et al. (2021)
Species
Relative range size and
abundance1
Max.
length
(mm)1
Caudal fin aspect
ratio2
Size at maturity –
female (mm)1,3 Spawning period / surface1,3
Egg
Size
(mm)1,3
Fecundit y (average /
maximum)1,3
Diet (presented by ontogenetic stage,
given as size in mm)4Habitat preference4
H. jenkinsi
Widespread and
abundant
450 1.64 120 Wet season / Demersal 1.65 185,000/400,000 Aquatic invertivore (≤40),
Omnivore (>40)
Generalist
H. epirrhinos
Range- restricted and
low abundance
420 1.64 120 Wet season /
Demersal
1.50 100,000/200,000 Aquatic invertivore (≤120),
Macrophagous carnivore (>120)
Deep pools, per manent habitat
S. trigonicus
Moderately
widespread and
abundant
170 1.78 60 Wet season /
Demersal
2.00 20,000/30,000 Meiophagous omnivore (≤40),
Algivore- detritivore / Omnivore (>40)
Generalist habitat t ype, but preference for
permanent habitat
S. rastellus
Range- restricted and
low abundance
150 2.50 70 Wet season /
Demersal
3.00 25,000/32,000 Algivore- detritivore / Omnivore (≤90),
Algivore- detritivore (>90)
Deep pools, rocky subs trate, upper catchment
M. australis
Widespread and very
abundant
80 1.63 23 Year- round /
Vegetation
0.94 15/47 Meiophagous omnivore (all sizes) Generalist
M. gracilis
Range- restricted and
very abundant
60 1.44 27 Year- round /
Vegetation
0.92 6/15 Meiophagous omnivore (≤40)
Insectivore (>40)
Permanent habit at, sandy substrate
8
|
SHELLEY Et aL.
be significantly higher ( p < 0.05) than estimates from the Mitchell
River (mean a = 1.85; r = 2.00; HE = 0.14) and upper and middle
King Edward River loc ations (mean a = 1. 57; r = 1.19; HE =0. 22).
Estimates from the lower King Edward River (a = 4.14; r = 2.50;
HE = 0.32) did not differ significantly from the Dr ysdale River
estimates.
Global estimates of FST and DEST across all loci indicate significant
genetic structuring among locations from the different river systems
(FST = 0.48, 95% CIs = 0.38– 0.56; DEST = 0.42, 95% CIs = 0.32– 0.53).
Pairwise estimates of FST and DEST (Table S1) indicate a lack of ge-
netic structuring among locations within the Drysdale River, with
evidence of gene flow between the upper and lower locations. The
TABLE 2 Population genetic statistics for Hephaestus, Syncomistes, and Melanotaenia species populations screened for microsatellite
loci. Number of individuals genotyped (n), mean values over loci are presented for number of alleles (a), allelic richness (r), expected (HE) and
observed (HO) heterozygosities, Hardy– Weinberg equilibrium (HWE) p- values and inbreeding coefficients (FIS) (significance after correc tions
for multiple comparisons are indicated by bold tex t). The number of loci typed reflects the total number of loci used for population genetic
analyses, excluding those found to be monomporphic or influenced by null alleles
Species / Population nLoci typed ar HEHOHWE FIS
H. jenkinsi
Upper Drysdale R. 26 14 4.07 2.35 0.46 0.44 0.12 0.05
Middle Drysdale R. 17 14 4.43 2.04 0.43 0. 29 0.00 0.34
Lower Dr ysdale R. 17 14 3.50 2.15 0.46 0.46 0.49 0.00
Middle Mitchell R . 20 14 2.00 1.26 0.19 0.11 0.00 0.42
Lower Mitchell R. 20 14 1.69 1.14 0.09 0.06 0.00 0.32
Upper King Edward R. 18 14 1.64 1.21 0 .13 0.13 0.32 −0.06
Middle King Edward R. 10 14 1.50 1.17 0.11 0.11 0.57 0.00
Lower King Edward R . 24 14 4.14 2.46 0.42 0.39 0.10 0.05
H. epirrhinos
Upper Drysdale R. 30 54.40 2.42 0.57 0.48 0.01 0.15
Middle Drysdale R. 17 53.80 2.48 0. 59 0.43 0.00 0.27
Lower Dr ysdale R. 18 54.60 2.58 0. 59 0.50 0.04 0.14
S. trigonicus
Upper Drysdale R. 30 14 4. 07 2.13 0.49 0.44 0.01 0.09
Middle Drysdale R. 20 14 3.71 2.14 0 .47 0.44 0 .12 0.06
Lower Dr ysdale R. 13 14 2.93 2.09 0.45 0.42 0.13 0.07
Upper King Edward R. 26 14 1.50 1.06 0.04 0.05 0.30 −0.11
Middle King Edward R. 23 14 1.64 1.18 0.10 0 .11 0.33 −0.06
Lower King Edward R . 30 14 2.86 1.42 0.25 0. 24 0. 24 0.03
S. rastellus
Upper Drysdale R. 30 14 4.00 2.30 0.50 0.46 0.02 0.08
Middle Drysdale R. 10 14 3.36 2.26 0 . 51 0.50 0.30 0.03
Lower Dr ysdale R. 6 14 2.85 2.17 0.54 0.54 0.44 0.01
M. australis
Upper Drysdale R. 9 7 6.00 4. 51 0.73 0.70 0.19 0.05
Middle Drysdale R. 7 7 5.86 4.21 0.67 0.65 0.34 0.03
Lower Dr ysdale R. 28 710 .29 5.37 0.72 0.65 0.01 0.10
Upper Mitchell R . 20 77.71 4.92 0. 74 0.63 0.00 0 .14
Middle Mitchell R . 20 77.4 3 4.85 0. 74 0.66 0.02 0.10
Lower Mitchell R. 20 76.14 3.54 0.69 0.66 0.15 0.05
Upper King Edward R. 20 78.86 5.86 0.75 0.71 0.06 0.06
Middle King Edward R. 20 78.14 5.73 0.70 0. 61 0.00 0 .14
Lower King Edward R . 20 710.29 6.79 0. 76 0.73 0.12 0.04
M. gracilis
Upper Drysdale R. 20 56.83 3.38 0.57 0.56 0. 25 0.03
Middle Drysdale R. 20 56.40 4.02 0.56 0.42 0.00 0.25
Lower Dr ysdale R. 20 57. 8 0 4 .71 0.63 0.44 0.00 0.31
|
9
SHELLE Y Et aL.
middle Drysdale River population differed significantly (FST = 0.26,
p < 0.001; Table S1), but its uniqueness is expected to be overstated
due to evidence of non- random mating (multiple loci exhibiting sig-
nificant deviations from HWE and significant FIS; Table 2). A lack of
genetic structuring also was evident among the locations from the
Mitchell River and among the upper and middle locations from the
King Edward Rivers. The lower King Edward River was an exception in
differing significantly from the upper and middle locations (FST =0.64
and 0.54 respectively, p < 0.001; Table S1). The findings are sup-
ported by AMOVA analyses that suggest a high level of microsatellite
variation between rivers (32%, p < 0.01), whereas between- location
variation within rivers explained 22% (p < 0.01) and within- location
variation explained 44% (p < 0.01) of the tot al variation.
Bayesian STRUCTURE analyses also identified significant ge-
netic structuring among the three river systems. STRUCTURE analy-
ses identifie d five distinc t popul ation clus ters (K = 5; Fi gure 3), three
of which indicate distinct ancestry between the Drysdale, Mitchell,
and King Edward rivers, and two additional clusters comprised of
individuals from the divergent middle Drysdale River and lower King
Edward River locations.
BayesAss analyses provided evidence of bi- directional migra-
tion between the upper and lower Drysdale River locations at rates
of 19% and 7%, respectively (Figure 3). Within the Mitchell River,
there was evidence of upstream migration between the lower and
middle loc ation at a rate of 25%, but there was no evidence of re-
cent downstream migration between them. Within the King Edward
River there was evidence of downstream migration only between
the upper and middle locations at a rate of 20%.
3.1.2 | Narrow- range species – – H. epirrhinos
A total of 65 individual H. epirrhinos samples from three locations
spanning the Dr ysdale River were genot yped at five microsatellite
loci (Table 2). A significant deviation from HWE (p < 0.01) was ob-
served at the middle Drysdale River, which was accompanied by a
significant FIS, indicating a homozygote excess, although this appears
to be driven by a single microsatellite locus. Estimates of genetic
diversit y were largely consistent among all H. epirrhinos locations
from the Drysdale River (mean a = 4. 27; r = 2.50; HE = 0.58). Global
estimates of FST and DEST across all loci indicate a lack of genetic
structuring among locations from the Drysdale River (FST =0.01,
95% CIs = 0– 0.02; DEST = 0.02, 95% CIs = 0– 0.05), with no pair-
wise pop ulation estimates differ ing signif ic antly from zero (p > 0.05;
Table S1). AMOVA analyses were consistent suggesting that among-
location variation accounted for little of the total microsatellite vari-
ation (2%, p > 0.05), whereas with in - lo cation variation accounted for
the majority of the variation (98%, p < 0.01).
Likewise, Bayesian STRUCTURE analyses identified a single ge-
netic cluster (K = 1; Figure 3 indicating connectivity among locations
within the Drysdale River. BayesAss analyses provide evidence of
recent downstream and upstream migration between all locations
(range 7%– 16%), wit h downs tr eam mar ginally stronger than ups tream
migration (mean downstream =13%, mean upstream =8%; Figure 3).
3.2 | Syncomistes
3.2.1 | Wide- range species – – S. trigonicus
A total of 142 individual S. trigonicus samples from six locations
spanning the Drysdale and King Edward Rivers were genotyped
at 14 microsatellite loci (Table 2). Significant deviation from HWE
(p < 0.01) as observed in the upper Drysdale River, but was driven
by devi a tio ns at two lo ci onl y. Es t ima tes fo r tota l nu mbe r of a ll ele s,
alle li c ric hne ss , and ex pec ted heterozygosit y varied betwe en lo ca-
tions across rivers with estimates from the Dr ysdale River (mean
a = 3. 57; r = 2.12; HE = 0.47) being significantly greater (p < 0.05)
than the King Edward River (mean a = 2.00; r = 1.22; HE = 0 .13)
following permutation tests. Global estimates of FST and DEST
across all loci indicate significant genetic structuring among loca-
tio ns from t he dif ferent ri ver sy stem s (FST = 0.43 , 95% CIs = 0.34 –
0. 51; DEST = 0.31, 95% CIs = 0.18– 0.46). Pairwise estimates of FST
and DEST (Table S1) indicate a lack of genetic structuring among
all locations within the Drysdale River (upper, middle, and lower).
Significant structuring was obser ved among all locations from
King Edward River, with the lower King Edward River being highly
divergent, differing significantly from the upper and middle loca-
tions (FST = 0.43 and 0.41, respectively, p < 0.001; Table S1). The
findings are suppor ted by AMOVA analyses that suggest a high
level of microsatellite variation between rivers (39%, p < 0.01),
whereas between- location variation within rivers explained 10%
(p < 0.01), and within- location variation explained 51% (p < 0.01)
of the total variation.
Likewise, Bayesian STRUCTURE analyses identified three popu-
lation clusters (K = 3; Figure 3), two representing the Drysdale and
King Edward Rivers, and a third representing the genetically differ-
entiated lower King Edward River.
BayesAss analyses indicated the strength of migration among
locations within the Drysdale River ranged from 7% to 17%, with
evidence of bi- directional migration between the upper and middle
locations (Figure 3). By contrast, the lower Drysdale appears to be
a recipient of migrants from both the upper and middle locations,
but evidence for reciprocal migration was absent. Likewise, analy-
ses indicated the strength of migration among locations within the
King Edward River ranged from 7% to 22%, with evidence of bi-
directional migration between the upper and middle locations. There
was no evidence of recent migration between the lower location and
the upper and middle locations on the King Edward River, which was
expected given the divergent nature of this population.
3.2.2 | Narrow- range species – – S. rastellus
A total of 46 individual S. rastellus samples from three locations span-
ning the Drysdale River were genotyped at 14 microsatellite loci
( Ta b l e 2). No signif ic a n t de v i a t i o n s fr o m HWE (p < 0 .01 ) we re obser v e d
across locations and estimates of genetic diversity were largely con-
sistent among locations (mean a = 3.4 0; r = 2.24; HE = 0.52). Global
estimates of FST and DEST across all loci indicate a lack of genetic
10
|
SHELLEY Et aL.
structuring among locations from the Drysdale River (FST = 0.00;
DEST = 0.00), with no pairwise estimates differing significantly from
zero ( p > 0.05; Table S1). AMOVA analyses were consistent suggest-
ing that among- location variation accounted for little of the total mi-
crosatellite variation (0%, p > 0.05), whereas within- location variation
accounted for the majority of the variation (100%, p < 0.01).
Likewise, Bayesian STRUCTURE analyses identified a single ge-
netic cluster (K = 1; Fig u re 3) in dic ati ng co n necti v ity amo n g loca t ion s
within the Drysdale River. BayesAss analyses provided evidence of
downstream only migration between locations, ranging from 5% to
27% (Figure 3). The upper Drysdale River was a strong source of mi-
grants for the middle and lower catchment, at rates of 27% and 25%,
respectively. Analyses also provided evidence of recent downstream
migration between the middle and lower Drysdale River locations at
a rate of 5%.
3.3 | Melanotaenia
3.3.1 | Wide- range species – – M. australis
A total of 164 individual M. australis samples from nine locations
spannin g the Dry sd al e, Mit ch el l, and Kin g Ed wa rd River s we re geno-
typed at seven microsatellite loci (Tables 2). Significant deviations
from HWE (p < 0.01) were observed at four M. australis sample lo-
cations (lower Drysdale River, upper and middle Mitchell River, and
middle King Edward River), but these patterns appear to be driven by
a small number of loci (1 − 2 loci) which differed among populations.
Estimates of total number of alleles, allelic richness, and expected
heterozygosity were largely consistent among locations within and
between rivers (Drysdale River, mean a = 7. 38; r = 4.70; HE =0.71;
Mitchell River, mean a = 7.10; r = 4.44; HE =0.72; King Edward River,
FIGURE 3 Graphical depiction of the
degree of population genetic structuring
estimated for each fish species from
STRUCTURE analysis. Summary plots
indicate genetically divergent population
clusters (K) represented by dif ferent
colours, and the estimated membership
coefficient (y- axis) for individuals from
each sampling location in each population
cluster. Migration estimates derived from
BaysAss analyses are represented by
arrows and indicate the relative strength
and direc tionality of migration between
sampling locations. Dashed arrows
indicate migration across a major waterfall
barrier
|
11
SHELLE Y Et aL.
mean a = 9.10 ; r = 6.12; HE =0.74), with permutation test s indicating
no significant dif ference (p > 0.05).
Global estimates of FST and DEST across all loci indicate significant
genetic structuring among populations from the different river sys-
tems (FST = 0.14, 95% CIs = 0.07– 0.25; DEST = 0.47, 95% CIs = 0.34–
0.57). Pairwise estimates of FST and DEST (Table S1) indicate some
genetic structuring among locations within the Drysdale River, with
each pair wise estimate involving the upper catchment location
being significantly different from zero (mean FST = 0.11 , p < 0.001;
Table S1), whereas pairwise estimates bet ween the middle and lower
locations did not differ significantly. These findings should be inter-
preted cautiously as a consequence of the small sample sizes at the
upper and middle locations that could contribute to inflated FST val-
ues and unreliable estimates of genetic structure. Likewise, pairwise
estimates indicate some genetic structuring among locations within
the Mitchell River, with evidence of connectivity among the upper
and middle catchment, whereas each pairwise estimate involving
the lower location differed significantly from zero (mean FST = 0.22,
p < 0.001; Table S1). By contrast, all pairwise estimates among loca-
tion s fro m th e Kin g Ed war d River dif fere d sig ni fic ant ly fr om zer o ind i-
cating genetic structuring among all populations. Although significant
differentiation was observed between the middle and lower catch-
ments, FST was moderate (0.05) suggesting potential for low levels of
gene flow between these locations. These findings are supported by
AM OVA an a lys e s tha t sug ges t sig nif i c ant vari atio n bet wee n loc ation s
within (12%, p < 0.01) and between (5%, p < 0.05) rivers , wit h wit hin-
location variation explaining 83% (p < 0.01) of the total variation.
Likewise, Bayesian STRUCTURE analysis was largely consistent
in identifying distinctive population clusters for each of the river sys-
tems (Figure 3). Although STRUCTURE analyses identified two pop-
ulation clusters within the Mitchell River which is consistent with FST,
only a single population cluster was resolved for the Drysdale River
despite evidence of genetic differentiation of upper catchment based
on FST (Figure 3; Table 2). STRUCTURE analyses also identified two
population clusters for the King Edward River despite all pairwise es-
timates of FST indicating structure among all locations. However, a
common shared ancestry between the middle and lower catchment
locations is evident and is consistent with the low but significant es-
timates of FST between these locations (Table S1). Given that some
pairwise FST values between populations within the Drysdale and
Mitche ll Rive rs are low, it is likely the ST RUCT URE simply lac ked sen-
sitivity for detecting these low levels of genetic structuring between
locations, a known limitation of the package (Evanno et al., 2005).
BaysAss analyses provided evidence of upstream only migra-
tion patterns within the Drysdale River, with evidence of migration
from the lower catchment to the middle and upper catchment at
rates of 15% and 8%, respectively (Figure 3). There was evidence of
bi- directional migration in the Mitchell River between the upper and
middle catchment, at rates of 15% (downstream) and 10% (upstream)
(Figure 3). There was no evidence of migration to or from the lower
catchment, which is consistent with FST (Figure 3; Table S1). Finally,
BaysAss analyses provided evidence of recent downstream migration
between the middle and lower King Edward River, at a rate of 20%.
This is largely consistent with the low but significant estimates of FST
between these locations (Table S1), and the shared common ancestry
indicated by STRUCTURE analyses (Figure 3), suggesting some level
of connectivity between these locations within the King Edward River.
3.3.2 | Narrow- range species – – M. gracilis
A total of 60 individual M. gracilis samples from three locations span-
ning the Drysdale River were genotyped at five microsatellite loci
(Tables 1 and S1). Si gn if icant deviations from HWE (p < 0.01 ) were ob-
served at the middle and lower Drysdale River locations, which were
accompanied by significant inbreeding coefficients (FIS) indicating a
homozygote excess. These estimates appear to be driven by a single
microsatellite locus which differed between locations. Estimates of
genetic diversity did not differ significantly (p > 0.05) based on per-
mutation tests (mean a = 7.01; r = 4.04; HE =0.59). Global estimates
of FST and DEST across all loci provide evidence of genetic structuring
among locations from the Drysdale River (FST = 0.21, 95% CIs = 0 .11–
0.33; DEST = 0.67, 95% CIs = 0.55– 0.81). Pairwise estimates (Table S1)
indicate that this is driven by significant genetic differentiation the
upper catchment (mean FST = 0.25 p < 0.01), whereas pairwise es-
timates between the middle and lower Drysdale River did not differ
significantly from zero (p < 0.05). AMOVA analyses were consistent
suggesting significant among- location variation accounted for 26% of
the total microsatellite variation (p > 0.05), and within- location varia-
tion accounted for 74% (p < 0.01) of the total variation.
Likewise, Bayesian STRUCTURE analyses identified two genetic
clusters (K = 2; Figure 3) with the upper Dr ysdale River forming one
cluster and the middle and lower catchment constituting a second
cluster. BaysAss analyses provide evidence of recent bi- directional
migration between the middle and lower catchment, at rates of 21%
and 11%, respectively (Figure 3).
4 | DISCUSSION
We show that contemporar y dispersal between catchment s on the
Kimberley Plateau is greatly limited in all six study species, regard-
less of their respective dispersal syndromes. This result is consistent
with our third hypothesis that the rugged plateau landscape plays a
significant role in shaping patterns of contemporary gene flow and
genetic structuring between catchments. By contrast, estimates of
within- catchment gene flow and genetic structure were variable and
in line with the predicted dispersal potential of our study species.
These findings suggest that dispersal syndromes influence patterns
of connectivit y and gene flow when geographical constraints are
not as pervasive, supporting our first hypothesis. However, disper-
sal ability does not appear to be responsible for the disparate range
sizes of congeneric pairs, given similar pat terns of gene flow and ge-
netic structuring observed within the Drysdale River, and evidence
of historical dispersal between catchments in some narrow- range
species. As such, it appears that additional biological factors, such
as ecological niche differences, are probably responsible for differ-
ences in species range extent.
12
|
SHELLEY Et aL.
4.1 | Contemporary versus historical landscape
influences on genetic structure between catchments
Given that contemporary gene flow between the northern plateau
catchments is effectively absent in our study species, it appears that
over land dispers al across catchment divides duri ng floo d events is ex-
tremely rare under current environmental conditions, probably due
to the lack of flat lowland habitats. Therefore, range expansion of
wide- range species most likely resulted from historical dispersal op-
portunities. Such opportunities likely arose during the Pliocene and
Pleistocene when global fluctuations in sea level, caused by the ex-
pansion and contraction of Antarctic ice sheets, periodically exposed
a broad cont inenta l shelf aro und northern Austr alia. This allowed cur-
rently isolated catchments to coalesce or come closer together, and
freshwater fauna to disperse more broadly (Cook et al., 2014; Shelley
et al., 2020). Sea levels reached their lowest point during the Last
Gl ac ial Max imu m c. 18 kyr ago and curr en t (h igh) se a le ve ls retu rne d c.
6 kyr ago (Yokoyama et al., 2001), which marks a likely upper bound-
ary for the timescale over which between- catchment dispersal was
possible. Indeed, patterns of historical connectivity (associated with
lower sea levels) and contemporary isolation between Kimberley river
catchments have been inferred from phylogenetic studies in H. jen-
kinsi and S. trigonicus, and in two additional terapontids, using mito-
chondrial DNA markers (Shelley et al., 2020). The pattern appears to
be consistent across the region, although river catchments that drain
the nort hern plateau appe ar to be the mos t fr agmented under current
sea levels (Shelley et al., 2020). These catchments are constrained
by the plateau along their entire length and the particularly tortuous
coastline acts to markedly separate neighbouring river mouths.
The estimated mean age of divergence for each of our study
species ranges between 0.2 and 8.2 Myr ago (Shelley, Swearer,
et al., 2018b; Unmack et al., 2013), so it can be inferred that each
species has had multiple opportunities to disperse. It seems likely
that some instances of dispersal were followed by population ex-
tinctions given the presence of small fragmented populations of
species on the plateau that may occur hundreds of kilometres from
their main distribution (Shelley, Morgan, et al., 2018a). Of particu-
lar relevance here, small populations of H. epirrhinos and M. gracilis
occur in the lower King Edward River, suggesting that (a) they likely
dispersed th ere from the la rge pop ulati on in the Dr ysdale River, and
(b) they have not flourished in their new environment. So, it appears
only some species have successfully colonised neighbouring catch-
ments despite historical dispersal opportunities for the broader fish
communit y. Potential reasons for this are discussed below.
4.2 | Evidence of dispersal syndromes interacting
with landscape features to influencing genetic
structure within catchments
We provide evidence of contrasting patterns of intra- catchment gene
flow and genetic structure between sympatric species that are con-
sistent with predic ted dispersal potential. For instance, Syncomistes
and Hephaestus species pairs (predicted to have moderate and high
dispersal potentials, respectively) showed patterns of either pan-
mixia or shallow genetic struc turing within the Drysdale River, as did
H. jenkinsi in the Mitchell River. By contrast, Melanotaenia species
(predicted to have poor dispersal potential) exhibited patterns of
significant genetic structuring and limited gene flow between most
river reaches in those same catchments. To emphasise this point,
74% of all pairwise estimates of FST among sampling locations within
catchments were significant for the Melanotaenia species. By con-
trast, <40% of pair wise estimates of FST among sampling locations
within catchments were significant for Hephaestus and Syncomistes
species. Other multispecies population genetic studies on fishes and
invertebrates also have demonstrated patterns of gene flow and
genetic structuring consistent with dispersal traits (Alp et al., 2012;
Chester et al., 2015; Harris et al., 2015). Thus, this result supports
our first hypothesis that species dispersal potential influences pat-
terns of population genetic structure, providing an important exam-
ple for obligate Australian freshwater fishes.
We also provide evidence that waterfalls can act as significant
barriers to gene flow, depending on their scale and nature. We
demonstrate evidence for strong genetic structuring and a lack
of recent migration between populations above and below King
Edward Falls (13 m straight drop) in all wide- range species regard-
less of dispersal syndromes. Although significant genetic struc turing
and a lack of recent migration was observed among sites above and
below Mitchell Falls (four tiers, c. 15 m straight drop) in M. austra-
lis, the falls had lit tle effect on migration and genetic structuring in
H. jenkinsi which has a greater dispersal potential. By contrast, the
shorter Solea Falls on the Drysdale River (c. 10 m straight drop) had
little effect on pat terns of gene flow and genetic structure within
and between the congeneric groups. Evidence of upstream migra-
tion across even the largest of the falls is an interesting finding in
the context of the region's fish communities considering that these
hei ghts cannot be pas se d by leaping, and that none of th e st ud y sp e-
cies exhibit climbing abilities. However, this apparent anomaly is not
novel in a global context as some freshwater fishes have been shown
to overcome substantially more imposing barriers such as Niagara
Falls (56 m straight drop; Lujan et al., 2020). In the Kimberley, ex-
treme monsoonal rainfall events can lead to substantial overland
flow that may open temporary avenues for dispersal around wa-
terfall structures, whereas heightened water levels associated with
these events may lessen the effective height of the barriers allowing
upstream passage. Regardless, the ability of a species to capitalise
on these opportunities is expected to be influenced by dispersal
ability, as demonstrated by the contrasting patterns of population
connectivity and migration observed in the Drysdale and Mitchell
Rivers (Papadopoulou & Knowles, 2016).
4.3 | The potential role of ecological niche width in
determining gene flow and range size
Our result s suggest that dispersal potential is not the sole intrinsic
factor contributing to variation in patterns of gene flow, genetic
structure, and species range extent. For example, species within
|
13
SHELLE Y Et aL.
each congeneric pair exhibit drastically different range sizes de-
spite having similar dispersal potential, inferred from species traits
and observed patterns of gene flow within the Dr ysdale River.
Furthermore, H. epirrhinos (high dispersal potential) and M. graci-
lis (low dispersal potential) have dispersed between the Drysdale
and King Edward River, although populations are diminutive in the
latter. Their inability to thrive in the King Edward River may be a
consequence of biological (e.g., low genetic diversity and fitness of
founder populations and/or adaptive genetic differences) or eco-
logical factors (e.g., resource competition and/or predation from dif-
ferent aquatic taxa) limiting population grow th (Miller et al., 2019;
Szűcs et al., 2017).
Furthermore, it may be linked to the specialised ecological re-
quirements of these range- restricted species (Koivula et al., 2002;
Law ton et al., 2011; Or tego et al ., 2010). Notable differen ces in eco-
logical niche (i.e., habitat and dietary requirements) were apparent
between all wide (generalist ecological niche) and narrow- range
(specialist ecological niche) species pairs. Such differences would be
expected to influence colonisation success as generalists typically
utilize a wider variety of resources than specialists, allowing them to
perform bet ter across a variety of habitats (Caley & Munday, 20 03;
Futuyma & Moreno, 1988). Conversely, the specific resource re-
quirements of specialists are more likely to lead to environment mis-
matches (Le Feuvre et al., 2021; Slatyer et al., 2013).
4.4 | Implications for evolution and conservation
Our results add to a growing body of evidence suggesting the
Kimberley Plateau is an active evolutionary cradle where periodic
connection and isolation of catchments during Plio- Pleistocene
glacial cycles have driven patterns of narrow- range endemism that
are unparalleled in Australia (Shelley et al., 2020). Disjunct pat-
terns in species distributions and genetic structuring also have
eme rg ed (Sh elley, Morgan, et al., 2018a), the dive rs it y of which has
probably resulted from a combination of each species’ ability to
disperse and establish in new and changing habitats, interacting
with temp oral variation in climate and physical connectivity within
and between catchments. We obser ved a consistent pattern of
strong genetic structuring and population isolation across catch-
ments in a range of species differing in dispersal potential, niche
width, and range extent. Consequently, is it expected that such
patterns would apply to other freshwater fish fauna and catch-
ments distributed across the plateau. Therefore, it is problematic
to interpret contemporary connectivity from species distribu-
tions in the region. For conservation purposes, fish communities
from different catchments should be assumed to be isolated until
proven otherwise. These communities are expected to be largely
self- recruiting entities and likely to respond independently to en-
vironmental pressures, thus warranting independent management
consideration. This study highlights the importance of conserving
biodiversity from different catchments in order to preserve pat-
terns of endemism, genetic diversity, and evolutionary potential in
the region and provides important context for those trying to de-
scribe and manage freshwater biodiversity in the Kimberley land-
scape, Australia generally, and abroad.
ACKNOWLEDGMENTS
We acknowledge the help of Martin Gomon and Dianne Bray from
Museum Victoria in providing access to tissue samples. All work had
animal ethics approval from the Research Ethics and Integrity of-
fice at the University of Melbourne, ID 1212470.1. All collections
by the authors were made under Government of Western Australia,
Department of Fisheries permit, ref. 220/12. Collections by the au-
thors in National Parks were made under Department of Parks and
Wi ldl i fe Pe rmi t SF0086 85 (2 012– 20 13 ) and SF 0 0 98 77 (2 014– 201 5).
DATA AVAILAB ILITY STATE MEN T
The data that support the findings of this study are available from
the corresponding author upon reasonable request. Genetic data are
publicly available and accessible in the DRYAD archives under acces-
sion: https://doi.org/10.5061/dryad.3r228 0gh8.
ORCID
James J. Shelley https://orcid.org/0000-0002-2181-5888
Owen J. Holland https://orcid.org/0000-0002-2244-9373
Stephen E. Swearer https://orcid.org/0000-0001-6381-9943
Timothy Dempster https://orcid.org/0000-0001-8041-426X
Matthew C. Le Feuvre https://orcid.org/0000-0001-9592-5927
Craig D. H. Sherman https://orcid.org/0000-0003-2099-0462
Adam D. Miller https://orcid.org/0000-0002-1632-7206
REFERENCES
Alp, M., Keller, I., Westram, A. M ., & Robinson, C. T. (2012). How river
struc ture and biologic al traits influence gene flow: A popula-
tion genetic study of two stream invertebrates with differing
dispersal abilities. Freshwater Biology, 57, 969– 981. htt ps://doi.
org /10.1111/j .136 5- 2427.2012 .0 2758 .x
Blacket, M. J., Robin, C., Good, R. T., Lee, S. F., & Miller, A. D. (2012).
Universal primers for fluorescent labelling of PCR fragments— an
efficient and cost- effective approach to genotyping by fluores-
cence. Molecular Ecology Resources, 12, 456– 463. https://doi.
org /10.1111/j .1755- 09 98.2011. 03104.x
Bohonak, A. J. (1999). Dispersal, gene flow, and population struc-
ture. The Quarterly Review of Biology, 74, 21– 45. ht tps://doi.
org /10.1086/392950
Booth Jones, K. A., Nicoll, M. A. C., Raisin, C., Dawson, D. A ., Hipp erson,
H., Horsburgh, G. J., … Norris, K. (2017). Widespread gene flow
between oceans in a pelagic seabird species complex. Molecular
Ecology, 26, 5716– 5728. htt ps://doi.org/10.1111/mec.14330
Bradbury, I. R., L aurel, B., S nelgrove, P. V. R., Bent zen, P., & Campana, S.
E. (20 08). Global patterns in marine dispersal estimates: The influ-
ence of geogra ph y, ta xo nomic cat eg or y and li fe his to r y. Proceedings
of the Royal Society B: Biological S ciences, 275, 18 03– 1809. https://
doi.org/10.1098/rspb.2008.0216
Brookfield, J. (1996). A simple new method for es timating null allele fre-
quency from heterozygote deficiency. Molecular Ecology, 5, 453–
455. ht tp s://doi. org /10.10 46/ j.1365- 294X.1996.00 098.x
Caley, M. J., & Munday, P. L. (20 03). Growth trades off with habitat spe-
cialization. Proceedings of the Royal Society B: Biological Sciences,
270, S175– S177. https://doi.org/10.1098/rsbl.2003.0040
14
|
SHELLEY Et aL.
Chester, E. T., Miller, A. D., Valenzuela, I., W ickson, S. J., & Robson, B.
J. (2015). Drought survival str ategies, disper sal potential and per-
sistence of invertebrate species in an intermittent stream land-
scape. Freshwater Biology, 60, 2066 – 2083. ht tps://doi.o rg /10.1111/
fwb.12630
Clober t, J., Baguette, M., Benton, T. G., & Bullock, J. M. (2012). Dispersal
ecology and evolution. Oxford University Press.
Closs, G. E., & Lake, P. S. (1996). Drought, differential mor talit y and the
coexistence of a native and an introduced fish species in a south
east Australian intermit tent stream. Environmental Biology of Fishes,
47, 17– 26. https://doi.org/10.1007/BF000 02376
Coates, W. D., Hale, R., & Morrongiello, J. R. (2019). Dispersal decisions
and personalit y in a freshwater fish . Animal Behaviour, 157, 209–
218. https://doi.org/10.1016/j.anbeh av.2019.07.012
Comte, L., & Olden, J. D. (2018). Evidence for dispersal syndromes in
freshwater fishes. Proceedings of the Royal Society B: Biological
Sciences, 285, 20172214. https://doi.org/10.1098/rspb.2017.2214
Cook, B. D., Unmack, P. J., Huey, J. A., & Hughes, J. M. (2014). Did com-
mon disjunct populations of freshwater fishes in nort hern Australia
form from the same biogeographic events? Freshwater Science, 33,
263– 272 . http s: //do i. org /10.10 86/6749 36
Davis, A . M., Pusey, B. J., & Betancur- R, R. (2020). Effects of adoption of
freshwater residency on life- histor y ecology of terapontid grunt-
ers. Freshwater Biology, 65, 1139– 1152 . https: //doi.org/10 .1111/
fwb.1349 9
Earl, D. A., & Vonholt, B. M. (2012). STRUCTURE HARVESTER: A website
and program for visualizing STRUCTURE output and implement-
ing the Evanno method. Consevation Genetic Resources, 4, 359– 361.
h t t p s : / / d o i . o r g / 1 0 . 1 0 0 7 / s 1 2 6 8 6 - 0 1 1 - 9 5 4 8 - 7
Evanno, G., Regnaut, S., & Goudet, J. (2005). Detec ting the number of
clusters of individuals using the software S TRUCTURE: A sim-
ulation study. Molecular Ecology, 14, 2611– 2620. https://doi.
org/10.1111/j.1365- 294X.20 05.02553.x
Ferreira, D. G., Souza- Shibatta, L., Shibatta, O. A., Sofia, S. H., Carlsson, J.,
Pinheiro Dias, J. H. Makrakis, S., & Makrakis, M. C. (2017). Genetic
struc ture and diversity of migrator y freshwater fish in a fragmented
Neotropical river system. Review of Fish Biology and Fisheries, 27,
2 0 9 – 2 3 1 . h t t p s : / /d o i . o r g / 1 0 . 1 0 0 7 / s 1 1 1 6 0 - 0 1 6 - 9 4 4 1 - 2
Frankham, R., Ballou, J. D., Ralls, K., Eldridge, M. D. B., Dudash, M. R.,
Fenster, C. B., Lacy, R. C ., & Sunnucks, P. (2017). Genetic manage-
ment of fragmented anim al and plant populations. Oxford Universit y
Press.
Frankham, R., Ballou , S. E. J. D., Briscoe, D. A., & Ballou, J. D. (2010).
Introduction to conservation genetics, 2nd ed. C ambridge Universit y
Press, C ambridge, UK.
Futuyma, D. J., & Moreno, G. (1988). The evolution of ecological spe-
cialization. A nnual Review of Ecology and Systematics, 19, 207– 233.
htt ps://doi.or g/10.1146/annur ev.es.19.110188.001231
Harris, L. N., Bajno, R., Gallagher, C. P., Koizumi, I., Johnson, L. K.,
Howland, K. L., Taylor, E. B., & Reist, J. D. (2015). Life- histor y
characteristics and landscape attributes as drivers of genetic
variation, gene flow, and fine- scale population structure in north-
ern Dolly Varden (Salvelinus malma malma) in Canada. Canadian
Journal of Fisheries and Aquatic Sciences, 72, 147 7– 1493 . ht tp s://doi.
o r g / 1 0 . 1 1 3 9 / c j f a s - 2 0 1 5 - 0 0 1 6
Hubisz, M. J., Falush, D., Stephens, M., & Pritchard, J. K. (2009). Inferring
weak population structure wit h the assistance of sample group in-
formation. Molecular Ecology Resources, 9, 1322– 1332. http s://doi.
org /10.1111/j .1755- 09 98.200 9.02 591.x
Hughes, J. M., Huey, J. A., & Schmidt, D. J. (2013). Is realised connec-
tivity among populations of aquatic fauna predictable from po-
tential connectivity? Freshwater Biology, 58, 951– 966. https://doi.
org /10.1111/f wb .12099
Jombar t, T. (20 08). adegenet: A R package for the multivariate analy-
sis of genetic markers. Bioinformatics, 24, 1403– 1405. https://doi.
org/10.1093/bioin forma tics/btn129
Jombart, T., & Ahmed, I. (2011). adegenet 1.3- 1: New tools for the anal-
ysis of genome- wide SNP data. Bioinformatics, 27, 3070– 3071.
https://doi.org/10.1093/bioin forma tics/btr521
Jost, L. (2008). GST and its relatives do not measure differ-
entiation. Molecular Ecology, 17, 4015– 4026. https://doi.
org /10.1111/j .136 5- 294X .200 8.03 887.x
Kennard, M. J., Pusey, B. J., Olden, J. D., Mackay, S. J., Stein, J. L ., &
Marsh, N. (2010). Classification of natural flow regimes in Australia
to support environmental flow management. Freshwater Biology, 55,
171– 193. https ://doi.or g/10.1111/ j.1365- 2427.20 09.023 07.x
Koivula, M., Kukkonen, J., & Niemelä, J. (20 02). Boreal carabid- beetle
(Coleoptera, Carabidae) assemblages along the clear- cut originated
succession gradient. Biodiversity & Conservation, 11, 1269– 1288.
htt ps://doi.org/10.10 23/A:1016 0 18702894
Lawton, R. J., Messmer, V., Pratchett, M. S., & Bay, L. K. (2011). High gene
flow across large geographic scales reduces extinction risk for a
highly specialised coral feeding butterflyfish. Molecular Ecology, 20,
358 4– 3598. h tt ps://doi.or g/10 .1111/ j.13 65- 294X.2011. 05207.x
Le Feuvre, M. C., Dempster, T., Shelley, J. J., Davis, A. M., & Swearer,
S. E. (2021). Range restriction leads to narrower ecological niches
and greater extinction risk in Australian freshwater fish. Biodiversity
and Conservation, 30, 2955– 2976. https://doi.org/10.1007/s1053 1-
0 2 1 - 0 2 2 2 9 - 0
Le Feuvre, M. C., Dempster, T., Shelley, J. J., & Swearer, S. E. (2016).
Macroecological relationships reveal conservation hotspots and
extinction- prone species in Australia’s freshwater fishes. Global Ecology
and Biogeography, 25, 176– 186. https://doi.org/10.1111/geb.12397
Luiz, O. J., Allen, A. P., Robertson, D. R., Floeter, S. R., Kulbicki, M.,
Vigliola, L., Becheler, R., & Madin, J. S. (2013). Adult and larval
traits as determinants of geographic range size among tropical reef
fishes. Proceedings of the National Academy of Sciences, 110, 16498–
16502 . ht tps://doi.o rg/10.1073/pnas.130 40 74110
Lujan, N. K., Weir, J. T., Noonan, B. P., Lovejoy, N. R., & Mandrak, N . E .
(2020). Is Niagara Falls a barrier to gene flow in riverine fishes? A test
using genome- wide SNP dat a from seven native species. Molecular
Ecology, 29, 1235– 1249. https://doi.org/10.1111/mec.15406
Marshall, D. J., Monro, K., Bode, M., Keough, M. J., & Swearer, S.
(2010). Phenotype– environment mismatches reduce con-
nectivity in the sea. Ecology Letters, 13, 128– 140. https://doi.
org /10.1111/j .1461- 0248 .20 09.0140 8. x
Meirmans, P. G. (2009). GenoDive v2.0 b14. Computer software distrib-
uted by the author. Available from: http://www.bentl eydru mmer.
nl/softw are/softw are/GenoD ive.html
Miller A. D., Good R. T., Coleman R. A., Lancaster M. L., & Week A. R.
(2013). Microsatel lite loci and the complete mito chondrial DNA se-
quence characterized through next generation sequencing and de
novo genome assembly for the critically endangered orange- bellied
parrot, Neophema chrysogaster. Molecular Biology Reports, 40, (1),
3 5 – 4 2 . h t t p : / / d x . d o i . o r g / 1 0 . 1 0 0 7 / s 1 1 0 3 3 - 0 1 2 - 1 9 5 0 - z
Miller, A . D., Hoffmann, A. A., Tan, M. H., Young, M., Ahrens, C.,
Cocomaz zo, M., Rattray, A., Ierodiaconou, D. A., Treml, E., &
Sherma n, C . D. (2019). Loca l and regional sc ale habitat heterogen e-
it y co n t r i b u t e to genetic a daptation in a co m m e r c i a l l y imp or t ant m a-
rine mollusc (Haliotis rubra) from southeastern Australia. Molecular
Ecology, 28, 3053– 3072. https://doi.org/10.1111/mec.15128
Mondol, M. R . K., Schmidt, D. J., Huey, J. A., & Hughes , J. M. (2014).
Development of micros atellite markers for western rainbowfish
(Melanotaenia australis) using the Ion Torrent second generation
sequencing approach. Conservation Genetics Resources, 6, 829– 831.
h t t p s : / / d o i . o r g / 1 0 . 1 0 0 7 / s 1 2 6 8 6 - 0 1 4 - 0 2 7 1 - z
Ortego, J., Aguirre, M. P., & Cordero, P. J. (2010). Population genet-
ics of Mioscirtus wagneri, a grasshopper showing a highly frag-
mented distribution. Molecular Ecology, 19, 472– 483. https://doi.
org /10.1111/j .136 5- 294X .200 9.04512.x
Papadopoulou, A., & Knowles, L. L . (2016). Toward a paradigm shift in
comparative phylogeography driven by trait- based hypotheses.
|
15
SHELLE Y Et aL.
Proceedi ngs of the National Academy of Sciences, U.S.A., 113, 8018–
8024. https://doi.org/10 .1073/pn as.16 010 69113
Pepper, M., & Keogh, S. J. (2014). Biogeography of the Kimberley,
Western Australia: A review of landscape evolution and biotic re-
sponse in an ancient refugium. Journal of Biogeography, 41, 1443–
1455. h tt ps ://doi.or g/10.1111/ jbi.12324
Prunier, J. G., Dubut, V., Loot, G., Tudesque, L., & Blanchet , S. (2018). The
relative contribution of river network structure and anthropogenic
stressors to spatial patterns of genetic diversity in t wo freshwater
fishes: A multiple- stressors approach. Freshwater Biology, 63, 6– 21.
https ://doi.or g/10.1111/ fw b.13 034
Raymond, M ., & Rousset , F. (1995). GENEPOP, population genetics
software for exact tests and ecumenicism (Version 3.4). Journal
of Heredit y, 86, 2 4 8 – 2 4 9 . h t t p s : / / d o i . o r g / 1 0 . 1 0 9 3 / o x f o r d j o u r n a l s .
jhe re d. a111573
Rice, W. R. (1989). Analyzing tables of statistical tests. Evolution, 43, 223–
225. https://doi.org/10.2307/2409177
Riginos , C., Buckley, Y. M., Blomberg, S. P., & Treml, E. A. (2014). Dispersal
capacity predicts both population genetic struc ture and species
richness in reef fishes. The American Natu ralist, 184, 52– 64. htt ps://
doi.org/10.10 86/676505
Ronce, O. (2007). How does it feel to be like a rolling stone? Ten ques-
tions about dispersal evolution. Annual Review of Ecology Evolution
and Systematics, 38, 231– 253. ht tps://doi.org/10.1146/annur
ev.ecols ys.38.091206.095611
Ryman, N., & Palm, S. (2006). POWSIM: a computer program for as-
sessing statistical power when testing for genetic differen-
tiation. Molecular Ecology Notes, 6(3), 60 0– 602. https://doi.
org /10.1111/j .1471- 828 6. 20 06.01378.x
Shell ey, J. J., Dem ps te r, T., Le Feu vre, M . C. , Un ma ck , P. J., L af fa n, S. W., &
Swea rer, S. E. (2 019). A rev isio n of the bi or egio nalis ati on of fre shwa -
ter fish communities in the Austr alian Monsoonal Tropics. Ecology
and Evolution, 9, 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., Allen, M. G. (2018a). A field gu ide to the freshwater
fishes of the Kimbe rley. Murdoch University Print Production Team.
Shelley, J. J., Swearer, S. E., Adams, M ., Dempster, T., Le Feuvre, M. C.,
Hammer, M. P., & Unmack, P. J. (2018b). Cr yptic biodiversity in the
freshwater fishes of the Kimberley endemism hotspot, nor thwest-
ern Australia. Molecula r Phylogenetics and Evolution, 127, 843– 858.
https://doi.org/10.1016/J.YMPEV.2018.06.032
Shelley, J. J., Swearer, S. E., Dempster, T., Adams, M., Le Feuvre, M. C.,
Hammer, M. P., & Unmack, P. J. (2020). Plio- Pleis tocene sea- level
changes drive speciation of freshwater fishes in nor th- western
Australia. Journal of Biogeography, 47, 1727– 1738. https://doi.
org /10.1111/j bi.138 56
Slatkin, M. (1987). Gene flow and the geographic struc ture of natural
populations. Science, 236, 787– 792. https://doi.org/10.1126/scien
ce.3 57619 8
Slatyer, R . A., Hirst, M., & Sexton, J. P. (2013). Niche breadth predicts
geographical range size: A general ecological pat tern. Ecology
Letter s, 16, 1104– 1114. ht tps://doi .org /10.1111 /ele .1214 0
Stevens, V. M., Whitmee, S., Le Galliard, J., Clobert, J., Böhning- Gaese,
K., Bonte, D., & Baguette, M. (2014). A comparative analysis of dis-
persal syndromes in terrestrial and semi- terrestrial animals. Ecology
Letter s, 17, 1039– 1052 . ht tps://doi.org /10.1111/ele .123 03
Szűcs, M., Melbourne, B. A., Tuff, T., Weiss- Lehman, C., & Hufbauer, R. A.
(2017). Genetic and demographic founder effects have long- term
fitness consequences for colonising populations. Ecology Letters,
20, 436– 44 4. https://doi.org/10.1111/ele.12743
Templeton, A. R., Shaw, K., Routman, E., & Davis, S. K. (1990). The ge-
netic consequences of habitat fragmentation. Annals of the Missouri
Botanical Garden, 13– 27. https://doi.org/10.23 07/2399621
Thacker, C. E., Unmack, P. J., Matsui, L ., & Rifenbark, N. (2007).
Comparative phylogeography of five sympatric Hypseleotris species
(Teleostei: Eleotridae) in south- eastern Aust ralia reveals a com-
plex pattern of drainage basin exchanges with little congruence
across species. Journal of Biogeography, 34, 1518– 1533. https://doi.
org /10.1111/j .136 5- 2699.20 07.01711 .x
Unmack, P. J. (2013). Biogeogr aphy. In P. Humphries, & K . Walker (Eds.),
Ecology of A ustralian fre shwater fishes (p p. 25 – 4 8). CS IRO Pu bli shi ng.
Unmack, P. J., Allen, G. R., & Johnson, J. B. (2013). Phylogeny and bio-
geography of rainbowfishes (Melanotaeniidae) from Australia and
New Guinea. Molecular Phylogenetics and Evolution, 67, 15– 27.
https://doi.org/10.1016/j.ympev.2012.12.019
Van Oosterhout, C., Hutchinson, W. F., Wills, D. P. M., & Shipley, P.
(2004). MICRO- CHECKER: Software for identifying and correcting
genotyping errors in microsatellite data. Molecular Ecology Notes, 4,
535– 53 8. ht tps://doi .org /10.1111 /j.1471- 8286. 20 04.0 0684.x
Walsh, P. S., Metzger, D. A., & Higuchi, R . (1991). Chelex 10 0 as a
medium for simple ex trac tion of DNA for PCR- based t yping
from forensic material. BioTechniques, 10, 506– 513. https://doi.
org /10.214 4/0 0011 4 018
Weir, B. S., & Cockerham, C. C. (1984). Estimating F- statistics for the
analysis of population structure. Evolution, 38, 1358– 1370. https://
doi.org/10.2307/2408641
Wilson, G. A., & Rannala, B. (20 03). Bayesian inference of recent migra-
tion rates using multilocus genotypes. Genetics, 163, 1177– 1191.
https://doi.org/10.1093/genet ics/163.3.1177
Yokoyama, Y., Purcell, A ., Lambeck, K., & Johnston, P. (2001). Shore- line
reconstruction around Australia during the Las t Glacial Maximum
and Late Glacial Stage. Quaternary International, 85, 9– 18. ht tps://
d o i . o r g / 1 0 . 1 0 1 6 / S 1 0 4 0 - 6 1 8 2 ( 0 1 ) 0 0 0 2 8 - 3
Young, M. J., Evans, J. P., & Simmons, L. W. (2011). Population genetic
struc ture and a possible role for selection in driving pheno-
typic divergence in a rainbowfish (Melanotaeniidae). Biological
Journal of the Linnean Society, 102, 144– 160. https://doi.
org /10.1111/j .1095- 8312 .2 010.01558 .x
SUPPORTING INFORMATION
Additional supporting information may be found in the online ver-
sion of the ar ticle at the publisher’s website.
How to cite this article: Shelley, J. J., Holland, O. J., Swearer,
S. E., Dempster, T., Le Feuvre, M. C ., Sherman, C. D. H., &
Miller, A. D. (2021). Landscape contex t and dispersal ability
as determinants of population genetic structure in freshwater
fishes. Freshw Biol., 00, 1– 15. ht tp s://doi.org/10 .1111/
fwb.138 44
