GENES MEET GEOLOGY: FISH PHYLOGEOGRAPHIC PATTERN REFLECTS ANCIENT, RATHER THAN MODERN, DRAINAGE CONNECTIONS
ABSTRACT Abstract We used DNA analysis of the freshwater Galaxias vulgaris complex (Pisces: Galaxiidae) to test a geological hypothesis of drainage evolution in South Island, New Zealand. Geological evidence suggests that the presently north-flowing Nevis River branch of the Clutha/Kawarau River system (Otago) once flowed south into the Nokomai branch of the Mataura system (Southland). The flow reversal is thought to have resulted from fault and fold activity associated with post-Miocene uplift. Mitochondrial DNA sequence data (control region and cytochrome b genes; 76 individuals; maximum divergence 7.1%) corroborate this geomorphological hypothesis: The Nevis River retains a freshwater fish species (Galaxias gollumoides; five sites; 10 haplotypes) that is otherwise restricted to Southland (nine sites; 15 haplotypes). There is no indication that the Nevis River lineage of G. gollumoides lives elsewhere in the Clutha/Kawarau system (> 30 sites). Likewise, two widespread Clutha lineages (G. ‘sp’ D; G. anomalus–G. pullus) are apparently absent from the Nevis (> 30 sites). In particular, G. ‘sp D’ lives throughout much of the Clutha (12 sites, 23 haplotypes), including a tributary of the Kawarau, but is absent from the Nevis itself. Conventional molecular clock calibrations (based on a minimum Nevis-Mataura haplotype divergence of 3.0%) indicate that the Nevis flow reversal may have occurred in the early-mid Pleistocene, which is roughly consistent with geological data. The broad phylogeographic structure evident in the Clutha system is consistent with the sedentary nature of nonmigratory galaxiids. Our study reinforces the value of combining biological and geological data for the formulation and testing of historical hypotheses.
- SourceAvailable from: Graham P Wallis[show abstract] [hide abstract]
ABSTRACT: There are two distinct genotypes and morphotypes of fish belonging to Galaxias vulgaris Stokell (sensu lato) in narrow sympatry in Healy Stream, a tributary of the Kye Burn in the Taieri River system, northeastern Otago. Therefore the present taxonomy of the species misrepresents the diversity of these fish. Analysis of allozymes using gel electrophoresis indicates at least four groups of populations. The two sympatric forms are clearly distinct species; the taxonomic status of the other geno‐ and morphotypes is more equivocal, but three species are recognised and described or redescribed here, on the basis of morphological and genetic evidence: 1. Galaxias vulgaris Stokell (type locality: Rubicon River, Waimakariri River system, North Canterbury), present in and north of the Waitaki River;2. G. anomalus Stokell (type locality: a drain tributary of the Manuherikia River, between Ophir and Omakau, Clutha River), present in the Clutha and Taieri Rivers in Otago. in the rivers of Southland as far west as the Waiau River;3. G. depressiceps n.sp. (type locality: Healy Creek, tributary of the Kye Burn, Taieri River system) also present in the Clutha and Taieri Rivers, the Kakanui River in north Otago, and all the rivers of Southland also as far west as the Waiau River.Other populations in Otago are also variously distinct genetically or morphologically from the above three species. Decisions on their taxonomic status await further study.Journal of the Royal Society of New Zealand. 01/1996; 26(3):401-427.
- [show abstract] [hide abstract]
ABSTRACT: Drainage systems of the Great Plains and western Gulf Slope underwent substantial changes through diversions and stream captures during the Pleistocene, either as the result of the glacial advances or through independent geologic processes. The distributions of a variety of fishes that range across west-central North America, such as the plains killifish (Fundulus zebrinus), are thought to be the product of this Pleistocene influence. We examined the geographic pattern of genetic variation in F. zebrinus using three allozyme loci (n = 793), mitochondrial DNA restriction fragment length polymorphisms (RFLPs, n = 352), and sequencing of the mitochondrial cytochrome oxidase I (COI, n = 23) in an attempt to understand the roles of dispersal and vicariance. The phylogeographic patterns were concordant between the allozyme and mitochondrial data with the exception of the population in the North Canadian River. The populations fell into three geographic assemblages, which we designated as northern, central, and southern. A large phylogenetic break (average Roger's D = 0.702; average sequence divergence in RFLPs = 4.6%; average sequence divergence in COI = 5.5%) separated the northern/central and southern assemblages. The northern region was likely colonized sometime during the mid-Pleistocene. Fish in the Brazos and Pecos Rivers probably reached these drainages through stream captures of the Red River. The large phylogenetic break between the northern/central and southern clades supports previous attempts to recognize two species of plains killifish: F. zebrinus and F. kansae.Evolution 03/2001; 55(2):339-50. · 4.86 Impact Factor
? 2001 The Society for the Study of Evolution. All rights reserved.
Evolution, 55(9), 2001, pp. 1844–1851
GENES MEET GEOLOGY: FISH PHYLOGEOGRAPHIC PATTERN REFLECTS ANCIENT,
RATHER THAN MODERN, DRAINAGE CONNECTIONS
JONATHAN M. WATERS,1,2DAVE CRAW,3JOHN H. YOUNGSON,3AND GRAHAM P. WALLIS1
1Department of Zoology, University of Otago, PO Box 56, Dunedin, New Zealand
3Department of Geology, University of Otago, PO Box 56, Dunedin, New Zealand
hypothesis of drainage evolution in South Island, New Zealand. Geological evidence suggests that the presently north-
flowing Nevis River branch of the Clutha/Kawarau River system (Otago) once flowed south into the Nokomai branch
of the Mataura system (Southland). The flow reversal is thought to have resulted from fault and fold activity associated
with post-Miocene uplift. Mitochondrial DNA sequence data (control region and cytochrome b genes; 76 individuals;
maximum divergence 7.1%) corroborate this geomorphological hypothesis: The Nevis River retains a freshwater fish
species (Galaxias gollumoides; five sites; 10 haplotypes) that is otherwise restricted to Southland (nine sites; 15
haplotypes). There is no indication that the Nevis River lineage of G. gollumoides lives elsewhere in the Clutha/
Kawarau system (? 30 sites). Likewise, two widespread Clutha lineages (G. ‘sp D’; G. anomalus–G. pullus) are
apparently absent from the Nevis (? 30 sites). In particular, G. ‘sp D’ lives throughout much of the Clutha (12 sites,
23 haplotypes), including a tributary of the Kawarau, but is absent from the Nevis itself. Conventional molecular
clock calibrations (based on a minimum Nevis-Mataura haplotype divergence of 3.0%) indicate that the Nevis flow
reversal may have occurred in the early-mid Pleistocene, which is roughly consistent with geological data. The broad
phylogeographic structure evident in the Clutha system is consistent with the sedentary nature of nonmigratory
galaxiids. Our study reinforces the value of combining biological and geological data for the formulation and testing
of historical hypotheses.
We used DNA analysis of the freshwater Galaxias vulgaris complex (Pisces: Galaxiidae) to test a geological
Biogeography, galaxiid, mitochondrial DNA, phylogenetics, river capture, vicariance.
Received December 12, 2000.Accepted May 9, 2001.
The common river galaxias complex (Galaxias vulgaris
sensu lato; hereafter G. vulgaris complex; see McDowall and
Wallis 1996; McDowall 1997; McDowall and Chadderton
1999; Waters and Wallis 2001a,b) consists of up to 10 closely
related sedentary freshwater fish species that are restricted
to New Zealand’s South Island. Unlike several diadromous
(migratory) species in the genus that have been the subject
of previous biogeographic papers (McDowall 1978; Craw
1979), members of the G. vulgaris complex complete their
entire life cycle in freshwater (McDowall 1990; Allibone and
Townsend 1997). The Otago and Southland provinces of
southern South Island house considerable diversity, with at
least eight major lineages of the Galaxias vulgaris complex
(Waters and Wallis 2001a,b), compared to single lineages
from central and northern South Island (Waters and Wallis
The biogeography of galaxiid fishes has intrigued biolo-
gists for more than a century (see Darwin 1872). In seeking
to explain the wide Southern Hemisphere range of this group
(Berra 1981), some biogeographers support a vicariant
(Gondwanan) model (Croizat et al. 1974; Rosen 1974, 1978;
Craw 1979), whereas others favor marine dispersal (Mc-
Dowall 1978; Berra et al. 1996). Much of the controversy
reflects ideology more than biological reality. Some vicari-
ance biogeographers argue that dispersal events are not test-
able (e.g., Ball 1975; Michaux 1991), only invoking dispersal
for taxa with distributions that cannot be explained by vi-
cariance. Such views are fast becoming obsolete with the
advent of molecular phylogeographic analysis (Avise 2000).
Indeed, both dispersal and vicariance models find support
from recent molecular analyses of galaxiid fishes (Waters et
Despite the long-standing debate concerning the Gond-
wanan distribution of galaxiid fishes, biogeographers have
only recently narrowed their focus to vicariance at the intra-
continental level (e.g., within and among drainages). New
Zealand’s rivers provide considerable scope for the testing
of vicariance hypotheses. Geological upheaval in South Is-
land has promoted the formation of new drainage divides,
leading to flow reversals in some rivers, including parts of
the Clutha/Kawarau River catchment, New Zealand’s largest
drainage system (Fig. 1A, B). Some hypothesized drainage
changes are merely rearrangements within the Clutha/Ka-
warau system. For instance, the presently north-flowing Car-
drona River (Fig. 1B, Fig. 2) previously drained south into
the Kawarau River (Craw et al. 1999). More dramatically,
the presently north-flowing Nevis River branch of the Clutha/
Kawarau system (Otago; Fig. 1B) may have once flowed
south into the Nokomai River branch of the Mataura River
(Southland; Fig. 1B, Fig. 2). Flow reversal could have oc-
curred as a result of ongoing uplift and compression of the
topography (Fig. 2). The Nevis valley profile is broadly con-
vex and far from equilibrium. The upper reaches have a gentle
gradient, with loss of altitude of only 250 m over 30 km.
The river then plunges steeply to the Kawarau River, losing
400 m over 12 km in a steeply incised rugged gorge. The
convex profile suggests that river capture occurred at the top
of this gorge (Fig. 2).
The Nevis River reversal hypothesis constitutes an excel-
lent model for phylogeographic analysis. Under vicariance
(e.g., Croizat et al. 1974) it is hypothesized that the fauna of
a translocated river will retain phylogenetic similarity with
that of its ancestral catchment. Furthermore, with absence of
subsequent migration or range expansion, the original drain-
RIVER CAPTURE AND FISH BIOGEOGRAPHY
DNA lineages of the Galaxias vulgaris complex, and (B, inset) the distribution of haplotypes in the Nevis-Nokomai region; the outline
of Figure 2 is also indicated. The historical south-flowing direction of the Nevis River drainage is marked with an arrow. Small dots
represent locations where galaxiid fish were apparently absent. Sample site abbreviations refer to the Appendix. N, Nevis River; M,
Mataura River; C, Clutha River.
Map of the study area in southern South Island, New Zealand, showing (A) the broad distribution of five major mitochondrial
age divide will continue to be represented as a point of faunal
disjunction. Thus, biogeographic analysis of sedentary fresh-
water taxa may have the potential to elucidate drainage evo-
Mayden (1988) examined the distributions of endemic
freshwater fishes in the Central Highlands of North America
and concluded that their biogeographic relationships reflect
historical rather than present-day drainage connections.How-
ever, biogeographic analyses based solely on distribution can
lead to spurious conclusions (Trewick 2000; Waters and Wal-
lis 2000; Wallis and Trewick 2001). Phylogeographic studies
incorporate molecular data and are potentially far more in-
formative (Waters and Wallis 2000). In recent years, a num-
ber of phylogeographic studies have concluded that river cap-
ture is an important biogeographic process (e.g., Mayden and
Matson 1992; Musyl and Keenan 1992; Waters et al. 1994;
Gollmann et al. 1997; Echelle and Echelle 1998; Hurwood
and Hughes 1998; Strange 1998; Durand et al. 1999; En-
gelbrecht et al. 2000; Kreiser et al. 2001). However, these
studies have tended to be descriptive rather than predictive
and are dominated by post hoc discussion of geological fac-
tors that might explain biogeographic relationships.
The widespread G. vulgaris complex represents a prom-
ising system for studies of freshwater vicariance. We have
previously shown that although loss of diadromy represents
an important prerequisite for speciation, some clades (up to
six taxa) have radiated subsequent to loss of diadromy (Wa-
ters and Wallis 2001b). The process of headwater capture
presents an important means for a freshwater-limited species
to increase its range and undergo genetic change (possibly
leading to speciation) in allopatry. Here we test the Nevis
River flow-reversal hypothesis using phylogeographic anal-
ysis of mitochondrial DNA (mtDNA) sequences. Further-
more, we aim to assess the role of river capture in generating
MATERIALS AND METHODS
Samples of the G. vulgaris complex were obtained from
40 sites in southern South Island (Fig. 1; Appendix). Spec-
imens were collected with pole-nets or an electrofishing ap-
paratus, anaesthetised, and stored frozen or placed in 95%
ethanol. Sequence analyses (see below) incorporated a total
of 76 individuals, including 50 fish (25 sites) from the Clutha
system, 18 fish (seven sites) from the Mataura system, along
with representatives from an additional eight sites in South-
land and Stewart Island (Fig. 1).
Tissue was digested using a CTAB-proteinase K buffer,
and DNA was purified by chloroform extraction and isopro-
panol precipitation. A total of 759 bp from the mitochondrial
control region was amplified and sequenced with primers P4
(5?-TAAACTACCCTCTGCTCCC-3? and S-Phe (5?-GCT-
TTAGTTAAGCTACG-3?. The complete mitochondrial cy-
tochrome b gene (1160 bp) was amplified with primers cytb-
Glu (5?-GAAAAACCACCGTTGTTATTCA-3? and cytb-Thr
(5?-CGACTTCCGGATTACAAGACC-3?, and a 768-bp seg-
JONATHAN M. WATERS ET AL.
related sediment basins (after Beanland and Barrow-Hurlbert 1988;
Turnbull 1988) and its southern extension into the Nokomai river
catchment (Kerr et al. 2000). The Nevis River was captured near
the Kawarau River and now flows north, but previously it flowed
south into the Nokomai River. The divide between these rivers was
formed by collision of broad upfolds of the Hector and Garvie
Geological map of the Nevis-Cardrona Fault System and
ment was sequenced with the former primer. Amplifications
were carried out in a PTC-100 thermal cycler (MJ Research,
Watertown, MA). Polymerase chain reaction conditions were
identical for both genes, with 40 cycles of denaturation at
94?C (30 sec), annealing at 47?C (30 sec), and extension at
72?C (30 sec). Sequencing reactions were constructed as rec-
ommended using the ABI Prism Big-Dye kit (Perkin-Elmer,
Foster City, CA). Completed sequencing reactions were pu-
rified by ethanol precipitation and run overnight on an ABI
Prism 377 autosequencer.
Sequence alignments were performed by eye. Phylogenetic
congruence of control region and cytochrome b datasets was
examined using the partition homogeneity test of Farris et
al. (1994) using PAUP (ver.4.0b4, Swofford 1999). Five hun-
dred partition replicates were analyzed under maximum par-
simony (MP) using the heuristic option and 10 replicates of
random addition for each replicate. Pairwise sequence di-
vergences were calculated using the Kimura (1980) two-pa-
rameter model of nucleotide sequence evolution implemented
in PAUP. Phylogenetic trees were constricted with MP and
distance (minimum evolution) methods implemented in
PAUP. MP analyses were performed with all nucleotide po-
sitions equally weighted and a transition:transversion ratio
of 2:1. MP analyses were performed with indels treated as a
fifth base, with a weighting of two relative to transitions.
Trees were recovered using the heuristic option implemented
in PAUP, with 20 replicates of random sequence addition.
Phylogenetic confidence in distance and MP topologies was
estimated by bootstrapping (Felsenstein 1985) with 1000 rep-
licate datasets. Tasmanian Galaxias brevipinnis control re-
gion and cytochrome b sequences (GenBank accession nos.
AF267338, AF267353) were included as an outgroup for phy-
logenetic analyses. Sequences from related members of the
G. vulgaris complex (found near but outside the study region)
were also included for phylogenetic comparison (G. depres-
siceps and G. vulgaris; GenBank accession nos. AF159823,
AF267367, AF159816, AF159820, AF267356, AF267357).
Partition homogeneity tests failed to reject phylogenetic
congruence of the control region (759 bp) and cytochrome
b (768 bp) datasets (P ? 0.367). The apparent phylogenetic
congruence justified the combination of the two partial gene
sequences in a phylogenetic analysis of 1527 bp of the mi-
tochondrial genome (79 haplotypes).
Both distance and MP analyses produced well-resolved
mtDNA phylogenies (Fig. 3). In addition to the outgroup (G.
brevipinnis) and reference taxa (G. vulgaris and G. depres-
siceps), five major mtDNA lineages/species were detected.
Specifically, bootstrap analysis strongly supported the recip-
rocal monophyly of G. gollumoides, combined G. anomalus
? G. pullus, G. ‘sp D’, G. ‘southern’, and G. ‘teviot’ (Fig.
3). As reported elsewhere (Waters and Wallis 2001b), cy-
tochrome b divergences among lineages consistently ex-
ceeded those for control region (Table 1). For combined se-
quence data, divergences among haplotypes ranged from
0.0% (several pairs of identical haplotypes, typically within
sites) to 7.1% (between G. gollumoides and G. ‘southern’).
In most cases, sampling of multiple individuals yielded only
one major lineage/species per site. However, three sites (M4,
C10, C5, Fig. 1) each had two major mtDNA lineages.
Clutha system samples (25 sites, 50 haplotypes) were dom-
inated by two widespread lineages, G. ‘sp D’ and G. anom-
alus–G. pullus (Fig. 1). G. ‘sp D’ haplotypes were recorded
from 12 of 25 Clutha sites (23 haplotypes). The geographic
range of this lineage extends from the lower reaches (Po-
mahaka River; C18–C20; Fig. 1) to the headwaters of the
Clutha (C1–C3). Phylogenetic analysis revealed a high de-
gree of phylogeographic structure within G. ‘sp D’, with hap-
lotypes grouped into lower-Clutha (C12, C15, C18–C20),
RIVER CAPTURE AND FISH BIOGEOGRAPHY
cytochrome b sequences of the Galaxias vulgaris complex. Bootstrap values (1000 replicates) ? 70% are given for distance analysis
(above nodes) and maximum parsimony (below nodes). The geographic distribution of major lineages is indicated. Abbreviations represent
locality codes as presented in the Appendix.
Phylogram from distance analysis (minimum evolution; Kimura two-parameter distances) of combined control region and
TABLE 1. Mean sequence divergence among lineages of the Galaxias
vulgaris complex. Divergences were calculated with the Kimura (1980)
two-parameter model of sequence evolution and are given as per-
centages. Values below the diagonal are for cytochrome b (768 bp),
whereas values above the diagonal are for control region (759 bp).
gol spD sthanotev
mid-Clutha (C6–C8), and upper-Clutha (C1–C3, C5) clades
(Figs. 1, 3). Divergences among upper and lower Clutha G.
‘sp D’ haplotypes were as high as 2.2%, and the mean di-
vergence was 1.2 ? 0.7%. The G. anomalus–G. pullus clade
was represented by 14 haplotypes (eight sites), with a max-
imum divergence of 1.8%. This clade exhibited strong phy-
logeographic structure and ranged from the lower (Waitahuna
River; C16–C17; Fig. 1) to the middle reaches of the Clutha
system (Manuherikia River; C4; Fig. 1).
Galaxias gollumoides (10 haplotypes; five sites) was re-
corded only from a single tributary of the Clutha, the Nevis
River (N1–N5; Fig. 1). Nevis River G. gollumoides constitute
JONATHAN M. WATERS ET AL.
a monophyletic group (100% bootstrap support) sister to
Southland G. gollumoides (MP bootstrap 98%; Fig. 3). Little
mtDNA variation was detected among the Nevis River hap-
lotypes (0.2 ? 0.1%; maximum 0.4%). In contrast, diver-
gences among Southland haplotypes of G. gollumoides were
considerable (mean 1.2 ? 0.5%; maximum 2.1%). Despite
their combined monophyly, Nevis River and Mataura River
(Southland; 10 haplotypes) G. gollumoides were quite di-
vergent (3.4% ? 0.3%; range 3.0–4.1%).
Two lineages, G. gollumoides and G. ‘southern’, were
found to be widespread in rivers south of the Clutha (Fig.
1). In particular, G. gollumoides mtDNA was recorded from
four sites in the Mataura River (M1, M4, M6, M7; Fig. 1),
three additional Southland rivers (MC1, CT2, W1), and two
Stewart Island rivers (F1, RB1; Fig. 1). Galaxias gollumoides
haplotypes were also recorded from the Nevis River (Clutha
system; see above; Fig. 1), whereas G. ‘southern’ was not
found in the Clutha system.
In contrast to the marked structure evident in the Clutha
system, the galaxiid fauna of the Mataura River (Southland;
Fig. 1) displays little obvious phylogeographic structure. For
instance, Mataura G. gollumoides contain at least four di-
vergent mtDNA lineages, three of which were recorded from
a single site (M4). In addition, some Mataura haplotypes (M1,
M4) cluster with haplotypes from Stewart Island (F1) and
the Catlins River (CT2). Neither Mataura G. ‘southern’ nor
Mataura G. gollumoides are monophyletic, and neither clade
exhibits phylogenetic structure that is concordant with sam-
pling locations (Fig. 3).
Geological evidence strongly suggests that the presently
north-flowing Nevis River branch of the Kawarau/Clutha
River system once flowed south into the Nokomai branch of
the Mataura system (Southland; Fig. 1). Our genetic evidence
corroborates this geomorphological hypothesis. Specifically,
the Nevis River retains a freshwater fish mtDNA lineage (G.
gollumoides) that is otherwise restricted to Southland.Despite
considerable sampling effort (Fig. 1B), there is no indication
that the Nevis River lineage of G. gollumoides (five sites; 10
haplotypes) lives elsewhere in the Clutha. Likewise, two
widespread Clutha lineages (G. ‘sp D’, G. anomalus–G. pul-
lus) are apparently absent from the Nevis. In particular, G.
‘sp D’, found throughout much of the Clutha (12 sites, 23
haplotypes), is reported from near but not in the Nevis (Ban-
nock Burn; C1; Fig. 1B). Thus, it seems that the phylogeo-
graphic relationships of the Clutha galaxiid fauna reflect his-
torical rather than present-day drainage patterns. The pres-
ence of G. gollumoides some way north in the Nevis River
(N4, N5) is consistent with the geological hypothesis of a
northerly capture point (Fig. 2). However, the negative eco-
logical effects of introduced brown trout (Salmo trutta) upon
galaxiids (Townsend and Crowl 1991) in the Nevis and Ka-
warau systems prevent us from elucidating an exact point of
faunal disjunction. Recent surveys have documented more
than 70 sites that apparently lack native fishes (Fig. 1B),
presumably because many of the sites are now dominated by
Headwater taxa are potentially highly susceptible to river
capture (Waters et al. 1994; Bishop 1995). The restricted
Nevis River distribution of Clutha G. gollumoides may in-
dicate that this is a relatively sedentary headwater lineage
that is ecologically excluded from lower in the system. In-
deed, a contact-zone study in a small tributary of the Mataura
River (M4; Fig. 1B; Waters et al. 2001 indicates that G.
gollumoides dominates upper reaches, whereas G. ‘southern’
is more abundant downstream. We suggest that G. gollu-
moides, and not G. ‘southern’, was translocated into the Clu-
tha because only the former is predominantly a headwater
Geological Timing and Molecular Clock Calibrations
The Nevis-Nokomai River area is underlain by schist bed-
rock that was eroded during uplift to yield a low relief erosion
surface in the Miocene (?20–30 million years ago). This
erosion surface has been disrupted by renewed and ongoing
uplift associated with the rise of the Southern Alps (Stirling
1990), which began in the late Miocene. A lake complex
occupied large parts of what is now the Clutha/Kawarau
catchment during the Miocene, and this lake complex was
filled with sediment from rising mountain ranges during the
late Miocene and Pliocene as the present rugged topography
was initiated (Douglas 1986; Turnbull 1988). Uplift persisted
and disrupted these sediments and the underlying low-relief
erosion surface (Fig. 2).
The Nevis-Cardrona Fault System (Beanland and Barrow-
Hurlbert 1988; Kerr et al. 2000; Fig. 2) is an important set
of north-northeast striking faults associated with Miocene–
Recent uplift. This fault system marks a topographical bound-
ary between gently deformed erosion surface (east) and deep-
ly eroded rugged mountains (west). The two principal strands
of the system are uplifting mountain ranges to the east and
west, and the intervening narrow structural depression hosts
the Cardrona, Nevis, and Nokomai Rivers. Along northerly
portions of fault system, the west side has been most uplifted,
with more gently upfolded mountain ranges to the east. How-
ever, the southern end of the system has the opposite ge-
ometry, with most uplift to the east of the Nokomai River
and a gently upfolded range to the west.
River capture occurred because of the ongoing uplift and
compression of the topography. The structural depression
within the Nevis-Cardrona Fault System has become pro-
gressively narrower, and thus the river valleys have been
narrowing with time. The present divide between the Nevis
and Nokomai rivers formed as a direct result of this com-
pression, where upfolded mountains (Hector and Garvie
Ranges) on either side of the Nevis-Cardrona Fault System
collided (Fig. 2). This divide occurs at a scissorlike hinge in
the fault system between west-side-up and east-side-up por-
tions (see above; Fig. 2). The divide is presently about 100
m above river level on either side, suggesting at least 300,000
to 500,000 years of uplift after capture, assuming uplift rates
of 0.3–0.5 mm/year typical of the region (Wellman 1979;
McSaveney et al. 1992).
DNA sequence data enable us to crudely date the Nevis
River flow reversal. According to standard (but uncorrected)
rates of divergence for vertebrate mtDNA (e.g., 2% per mil-
lion years; Brown et al. 1982), the minimum Nevis-Mataura
RIVER CAPTURE AND FISH BIOGEOGRAPHY
divergence of 3.0% implies a phylogenetic split in the early
Pleistocene (1.5 million years ago). A more rapid divergence
rate based on fish control region (3.6% per million years;
Donaldson and Wilson 1999) yields a younger date for the
change in drainage pattern (0.8 million years ago).Geological
evidence suggests that the faster divergence rate may be more
appropriate here. The convex profile of the Nevis River in a
region of rapid erosion supports the younger date. Capture
of the Nevis by the Kawarau may have been facilitated by
glaciation of the Kawarau in the Pleistocene.
Phylogeographic Structure and Drainage Evolution
The Clutha River is a large drainage system with few ob-
vious historical barriers to migration. Nevertheless, the non-
migratory galaxiid fauna of the Clutha exhibits strong phy-
logeographic structure. Only in a few instances is there ev-
idence of dispersal over relatively recent times (e.g., a hap-
lotype shared by C1 and C3, ?60 km apart). Anthropogenic
barriers to dispersal such as introduced trout (Townsend and
Crowl 1991) and hydroelectric impoundments (see Mc-
Dowall 1990) are far too recent to explain the deep phylo-
geographic structure that is generally evident within Clutha
clades. Even the widespread G. ‘sp D’ and G. anomalus–G.
pullus exhibit divergences (up to 2.2%) that are suggestive
of dispersal over geological rather than recent time frames,
according to standard molecular clock calibrations.
Some of the observed local phylogeographic structure
might reflect stochastic lineage sorting (e.g., paraphyly of
adjacent samples from the Beaumont River; C13, C14; Figs.
1A, 3). Additionally, waterfalls may represent barriers to up-
stream migration on local scales (Currens et al. 1990; King
and Wallis 1998). For instance, a substantial series of wa-
terfalls separates G. ‘sp D’ populations from upper Pool Burn
(C8; Fig. 1A) and lower Pool Burn (C7; Fig. 1A; Esa et al.
2000), possibly explaining their divergence (Fig. 3). More
generally, the broad phylogeographic structure observed in
G. ‘sp D’ and G. anomalus–G. pullus probably reflects the
sedentary nature of nonmigratory galaxiid taxa.
The findings that neither Mataura River G. ‘southern’ (four
sites; seven haplotypes) nor Mataura G. gollumoides (four
sites; 10 haplotypes) are monophyletic and that neither ex-
hibits obvious phylogeographic structure contrast strongly
with results for the Clutha galaxiid fauna. Thus, the hypoth-
esis of historical connection between Nevis and specifically
the Nokomai is less clearly supported than would be the case
if Nevis and Nokomai haplotypes of G. gollumoides consti-
tuted a monophyletic group. The lack of structure/monophyly
in the Mataura system may reflect changes in drainage pat-
terns during the Pleistocene. Freshwater interchange between
Southland and Stewart Island drainages probably occurred
during periods of low sea level at glacial maxima (see Flem-
ing 1979). In addition, there may have been gene flow across
the unstable alluvial plains of Southland. For instance, the
Oreti River (Fig. 1A) was connected to the Mataura River
between 350,000 and 120,000 years ago (McIntosh et al.
It has been suggested that drainage evolution is best es-
tablished by independent biological and geological data, an
approach that avoids circularity (Bishop 1995). Genetic data
have so far contradicted the capture of Waiau River head-
waters by the Buller River at Lewis Pass, a hypothesis that
seemed well-supported by faunal distribution (Waters and
Wallis 2000). Conversely, preliminary genetic and geological
data are both suggestive of a composite origin for the Taieri
River in New Zealand (Wallis et al. 1998; Youngson et al.
The current study provides strong support for a vicariance
(low-dispersal) model in the Clutha system: The Nevis River
retains faunal evidence of a vicariant event that may have
occurred more than a million years ago. Similarly, geological,
distributional, and phylogeographic evidence (Mayden 1988;
Kreiser et al. 2001) together provide compelling evidence of
preglacial drainage patterns in North America’s Great Plains.
Conversely, the lack of phylogeographic structure in the Ma-
taura drainage probably reflects dispersal and suggests that
this system is less informative for tests of vicariant hypoth-
eses. The contrasting levels of phylogeographic structure ob-
served in the Clutha and Mataura systems may reflect drain-
age size: The relatively large geographic scale of the Clutha
system could have promoted greater structure. Alternatively,
the unstable drainages of the Southland plain may have pre-
vented the accumulation of phylogeographic structure in the
south, as suggested for similar braided drainages in Canter-
bury (Wallis et al. 2001). In either case, our data provide
some of the strongest evidence to date that vicariant processes
have played substantial roles in the evolution of New Zea-
land’s aquatic fauna. While either biological or geological
data alone suggest freshwater vicariance, the combination of
datasets reveals a far more compelling case. Our study re-
inforces the value of both types of information for the for-
mulation and testing of historical hypotheses.
R. Allibone, C. Arbuckle, L. Chadderton, W. Cooper, Y.
bin Esa, J. Hollows, M. Nielson, H. Rhodes, S. Trewick, and
M. Tubbs helped to collect some of the specimens used for
this study. The manuscript was improved by comments from
B. Bowen and two anonymous reviewers. The work was sup-
ported by contract UOO-705 from the Marsden Fund ad-
ministered by the Royal Society of New Zealand and a seed-
ing grant from the Ecological Research Group (University of
Otago). Fish were collected under University of Otago An-
imal Ethics approval 43/99.
Allibone, R. M., and C. R. Townsend. 1997. Distribution of four
recently discovered galaxiid species in the Taieri River, New
Zealand: the role of macrohabitat. J. Fish Biol. 51:1235–1246.
Avise, J. C. 2000. Phylogeography: the history and formation of
species. Harvard Univ. Press, Cambridge, MA.
Ball, I. R. 1975. Nature and formulation of biogeographical hy-
potheses. Syst. Zool. 24:407–430.
Beanland, S., and S. A. Barrow-Hurlbert. 1988. The Nevis-Cardrona
fault system, central Otago, New Zealand: late Quaternary tec-
tonics and structural development. N. Z. J. Geol. Geophys. 31:
Berra, T. M. 1981. An atlas of distribution of the freshwater fish
families of the world. Univ. of Nebraska Press, Lincoln, NE.
Berra, T. M., L. Crowley, W. Ivantsoff, and P. A. Fuerst. 1996.
Galaxias maculatus: an explanation of its biogeography. Mar.
Freshwater Res. 47:845–849.
JONATHAN M. WATERS ET AL.
Bishop, P. 1995. Drainage rearrangement by river capture, be-
heading and diversion. Prog. Phys. Geogr. 19:449–473.
Brown, W. M., E. M. Prager, A. Wang, and A. C. Wilson. 1982.
Mitochondrial DNA sequences of primates: tempo and mode of
evolution. J. Mol. Evol. 18:225–239.
Craw, R. C. 1979. Generalised tracks and dispersal in biogeography:
a response to R. M. McDowall. Syst. Zool. 28:99–107.
Craw, D., J. H. Youngson, and P. O. Koons. 1999. Gold dispersal
and placer formation in an active and oblique collisional moun-
tain belt, Southern Alps, New Zealand. Econ. Geol. 94:605–614.
Croizat, L., G. J. Nelson, and D. E. Rosen. 1974. Centers of origin
and related concepts. Syst. Zool. 23:265–287.
Currens, K. P., C. B. Schreck, and H. W. Li. 1990. Allozyme and
morphological divergence of rainbow trout (Oncorhynchus my-
kiss) above and below waterfalls in the Deschutes River, Oregon.
Darwin, C. 1872. On the origin of species by means of natural
selection, or the preservation of favoured races in the struggle
for life. 6th ed. John Murray, London.
Donaldson, K. A., and R. R. Wilson. 1999. Amphi-panamic gem-
inates of snook (Percoidei: Centropomidae) provide a calibration
of the divergence rate in the mitochondrial DNA control region
of fishes. Mol. Phylogenet. Evol. 13:208–213.
Douglas, B. J. 1986. Lignite resources of central Otago. New Zea-
land Energy Research and Development Committee publication
no. 104. Univ. of Auckland, Auckland.
Durand, J. D., A. R. Templeton, B. Guinand, A. Imsiridou, and Y.
Bouvet. 1999. Nested clade and phylogeographic analyses of the
chub, Leuciscus cephalus (Teleostei: Cyprinidae), in Greece: im-
plications for Balkan Peninsula biogeography. Mol. Phylogenet.
Echelle, A. A., and A. F. Echelle. 1998. Evolutionary relationships
of pupfishes in the Cyprinodon eximius complex (Atherinomor-
pha: Cyprinodontiformes). Copeia 1998:852–865.
Engelbrecht, C. C., J. Freyhof, A. Nolte, K. Rassmann, U. Schliew-
en, and D. Tautz. 2000. Phylogeography of the bullhead Cottus
gobio (Pisces: Teleostei: Cottidae) suggests a pre-Pleistocene
origin of the major central European populations. Mol. Ecol. 9:
Esa, Y. B., J. M. Waters, and G. P. Wallis. 2000. Introgressive
hybridization between Galaxias depressiceps and Galaxias sp D
(Teleostei: Galaxiidae) in Otago, New Zealand: secondary con-
tact mediated by water races. Conserv. Genet. 1:329–339.
Farris, J. S., M. Ka ¨llersjo ¨, A. G. Kluge, and C. Bult. 1994. Testing
significance of incongruence. Cladistics 10:315–319.
Felsenstein, J. 1985. Confidence limits on phylogenies: an approach
using the bootstrap. Evolution 39:783–791.
Fleming, C. A. 1979. The geological history of New Zealand and
its life. Auckland Univ. Press, Auckland.
Gollmann, G., Y. Bouvet, Y. Karakousis, and C. Triantaphyllidis.
1997. Genetic variability in Chrondostoma from Austrian,
French and Greek Rivers (Teleostei, Cyprinidae). J. Zool. Syst.
Evol. Res. 35:165–169.
Hurwood, D. A., and J. M. Hughes. 1998. Phylogeography of the
freshwater fish, Mogurnda adspersa, in streams of northeastern
Queensland, Australia: evidence for altered drainage patterns.
Mol. Ecol. 7:1507–1517.
Kerr, L. C., D. Craw, R. J. Norris, J. H. Youngson, and P. Wopereis.
2000. Structure, geomorphology and gold concentration in the
Nokomai Valley, Southland, New Zealand. N. Z. J. Geol. Geo-
Kimura, M. 1980. A simple method for estimating evolutionary rate
of base substitutions through comparative studies of nucleotide
sequences. J. Mol. Evol. 16:111–120.
King, T. M., and G. P. Wallis. 1998. Fine-scale genetic structuring
in endemic galaxiid fish populations of the Taieri River. N. Z.
J. Zool. 25:17–22.
Kreiser, B. R., J. B. Mitton, and J. D. Woodling. 2001. Phylo-
geography of the plains killifish, Fundulus zebrinus. Evolution
Mayden, R. L. 1988. Vicariance biogeography, parsimony, and evo-
lution in North American freshwater fishes. Syst. Zool. 37:
Mayden, R. L., and R. H. Matson. 1992. Systematics and bioge-
ography of the Tennessee shiner, Notropis leuciodus (Cope) (Te-
leostei, Cyprinidae). Copeia 1992:954–968.
McDowall, R. M. 1978. Generalized tracks and dispersal in bio-
geography. Syst. Zool. 27:88–104.
———. 1990. New Zealand freshwater fishes: a natural history and
guide. Heinemann, Auckland.
———. 1997. Two further new species of Galaxias (Teleostei: Gal-
axiidae) from the Taieri River, southern New Zealand. J. Roy.
Soc. N. Z. 27:199–217.
McDowall, R. M., and W. L. Chadderton. 1999. Galaxias gollu-
moides (Teleostei: Galaxiidae), a new fish species from Stewart
Island, with notes on other non-migratory freshwater fishes pre-
sent on the island. J. Roy. Soc. N. Z. 29:77–88.
McDowall, R. M., and G. P. Wallis. 1996. Description and rede-
scription of Galaxias species (Teleostei: Galaxiidae) from Otago
and Southland. J. Roy. Soc. N. Z. 26:401–427.
McIntosh, P. D., D. N. Eden, and S. J. Burgham. 1990. Quaternary
deposits and landscape evolution in northeast Southland, New
Zealand. Palaeogeogr. Palaeoclimatol. Palaeoecol. 81:95–113.
McSaveney, M. J., R. Thomson, and I. M. Turnbull. 1992. Timing
of relief and landslides in central Otago. Pp. 1–6 in D. H. Bell,
ed. Sixth International Symposium on Landslides Proceedings,
Christchurch, New Zealand.
Michaux, B. 1991. Distributional patterns and tectonic develop-
ments in Indonesia: Wallace reinterpreted. Aust. Syst. Bot. 4:
Musyl, M. K., and C. P. Keenan. 1992. Population genetics and
zoogeography of Australian freshwater golden perch, Macquaria
ambigua (Richardson 1845) (Teleostei: Percichthyidae), and
electrophoretic identification of a new species from the Lake
Eyre basin. Aust. J. Mar. Freshw. Res. 43:1585–1601.
Rosen, D. E. 1974. Phylogeny and zoogeography of salmoniform
fishes and relationships of Lepidogalaxias salamandroides. Bull.
Amer. Mus. Nat. Hist. 153:265–326.
———. 1978. Vicariant patterns and historical explanation in bio-
geography. Syst. Zool. 27:159–188.
Stirling, M. W. 1990. The Old Man Range and Garvie Mountains:
tectonic geomorphology of the Central Otago peneplain, New
Zealand. N. Z. J. Geol. Geophys. 33:233–244.
Strange, R. M. 1998. MtDNA variation in johnny darters (Pisces:
Percidae) from eastern Kentucky supports stream capture for the
origin of upper Cumberland River fishes. Amer. Midl. Nat. 140:
Swofford, D. L. 1999. PAUP: phylogenetic analysis using parsi-
mony (and other methods). Ver. 4.0b4. Sinauer Associates, Sun-
Townsend, C. R., and T. A. Crowl. 1991. Fragmented population
structure in a native New Zealand fish: an effect of introduced
brown trout? Oikos 61:347–354.
Trewick, S. A. 2000. Molecular evidence for dispersal rather than
vicariance as the origin of flightless insect species on the Chat-
ham Islands, New Zealand. J. Biogeogr. 27:1189–1200.
Turnbull, I. M. 1988. Geological map of New Zealand 1:63, 360:
sheet S133-Cromwell. Department of Scientific and Industrial
Wallis, G. P., and S. A. Trewick. 2001. Finding fault with vicari-
ance: a critique of Heads (1998). Syst. Biol. 50: In press.
Wallis, G. P., R. M. Allibone, J. H. Youngson, C. Arbuckle, and
D. Craw. 1998. Biogeography of galaxiid fish from the Taieri:
genes meet geology. Pp. 47–48 in J. Hamel, ed. Research in the
Taieri catchment. Ecology Research Group Occasional Paper no.
1. Univ. of Otago, Dunedin.
Wallis, G. P., K. F. Judge, J. Bland, J. M. Waters, and T. M. Berra.
2001. Genetic diversity in New Zealand galaxiid fish: a test of
a biogeographic hypothesis. J. Biogeogr. 28:59–67.
Waters, J. M., and G. P. Wallis. 2000. Across the Southern Alps
by river capture? Freshwater fish phylogeography in South Is-
land, New Zealand. Mol. Ecol. 9:1577–1582.
———. 2001a. Mitochondrial DNA phylogenetics of the Galaxias
vulgaris complex from South Island, New Zealand: rapid radi-
ation of a species flock. J. Fish Biol. 58:1166–1180.
———. 2001b. Cladogenesis and loss of the marine life history
RIVER CAPTURE AND FISH BIOGEOGRAPHY
Geographic origin and Genbank accession details for Galaxias sequences. Map references refer to the New Zealand topographic map 1:50,000
series. Location codes are used in Figures 1 and 3.
Source Map referenceControl region Cytochrome b
Nevis R, Clutha R
Nevis R, Clutha R
Nevis R, Clutha R
Nevis R, Clutha R
F43 820 261
F42 921 451
F42 952 541
F42 960 530
Potter Ck, Nevis R, Clutha R
Nokomai R, Mataura R
Bushy Ck, Mataura R
Eyre Ck, Mataura R
Mokoreta R, Mataura R
Whitestone R, Waiau R
Freshwater R, Stewart Is
Robertson R, Stewart Is
Tarwood St, Catlins R
Matai St, MacLennan R
Nokomai R, Mataura R
Bullock Head St, Nokomal R, Mataura R
Bushy Ck, Mataura R
F42 006 568
F43 758 136
E43 694 288
E43 489 284
G47 090 122
D43 053 150
D48 170 543
D49 138 282
G46 323 183
G47 058 404
F43 714 091
F43 755 130
E43 694 288
Allen Ck, Mataura R
Hamilton Burn, Aparima R
Ophir Drain, Manuherikea R, Clutha R
F43 708 297
D44 265 952
D49 128 458
G41 430 615
Maori Ck, Manuherikea R, Clutha R
Waitahuna R, Clutha R
Waitahuna R, Clutha R
H42 590 467
H44 624 789
H44 662 735
Teviot R, Clutha R
Armstrongs Ck, Teviot R, Clutha R
Little Beaumont St, Beaumont R, Clutha R
Black St, Beaumont R, Clutha R
Teviot R, Clutha R
Teviot R, Clutha R
Shepherds Ck, Bannock Burn, Clutha R
Sheepskin Ck, Clutha R
Short Spur Ck, Lindis R, Clutha R
Maori Ck, Manuherikea R, Clutha R
Maori Ck, Manuherikea R, Clutha R
Pool Burn, Manuherikea R, Clutha R
Pool Burn, Manuherikea R, Clutha R
Raes Junction St, Clutha R
Tuapeka R, Clutha R
Pomahaka R, Clutha R
Flodden Ck, Pomahaka R, Clutha R
Heriot Burn, Pomahaka R, Clutha R
Daphne Brook, Catlins R
G43 441 121
G43 475 027
G44 452 889
G44 480 900
G43 396 105
G43 441 121
F42 063 558
G40 164 976
G40 417 077
H42 590 467
H42 585 382
H42 574 364
H42 574 352
G44 355 855
H44 511 801
G45 198 574
G44 231 707
G44 318 810
G46 320 184
phase in freshwater galaxiid fishes (Osmeriformes: Galaxiidae).
Waters, J. M., M. Lintermans, and R. W. G. White. 1994. Mito-
chondrial DNA variation suggests river capture as a source of
vicariance in Gadopsis bispinosus (Pisces: Gadopsidae). J. Fish
Waters, J. M., L. H. Dijkstra, and G. P. Wallis. 2000a. Freshwater
fish biogeography in the Southern Hemisphere: How important
is marine dispersal? Mol. Ecol. 9:1815–1821.
Waters, J. M., J. A. Lopez, and G. P. Wallis. 2000b. Molecular
phylogenetics and biogeography of galaxiid fishes (Ostei-
chthyes: Galaxiidae): dispersal, vicariance, and the position of
Lepidogalaxias salamandroides. Syst. Biol. 49:777–795.
Waters, J. M., Y. B. Esa, and G. P. Wallis. 2001. Genetic and
morphological evidence for reproductive isolation between sym-
patric populations of Galaxias (Teleostei: Galaxiidae) in South
Island, New Zealand. Biol. J. Linn. Soc. 73:287–298.
Wellman, H. W. 1979. An uplift map for the South Island of New
Zealand, and a model for the uplift of the Southern Alps. Pp.
13–20 in R. I. Walcott and M. M. Cresswell, eds. The origin of
the Southern Alps. Royal Society of New Zealand Bulletin no.
Youngson, J. H., D. Craw, C. A. Landis, and K. R. Schmitt. 1998.
Redefinition and interpretation of late Miocene-Pleistocene ter-
restrial stratigraphy, central Otago, New Zealand. N. Z. J. Geol.
Corresponding Editor: B. Bowen