Although the drainage history of intermountain western North America has been
studied extensively, traditional geological and paleontological approaches have slowed in
deciphering that history because most Miocene and Pliocene evidence is buried by
Pleistocene and other recent alluvial sediments. To fill this void, phylogeography has
emerged as a novel approach to the study of drainage history. By using genetic analysis
to examine the evolutionary relationships among various populations of a given species,
especially an aquatic organism, researchers can construct genetic-based distribution
patterns that ostensibly reflect historic inter-basin connections. This approach can
potentially fill in some of the gaps left by other methods and can also help geologists
focus their investigations in areas most likely to have been important dispersal pathways.
As with any scientific endeavor, a suitable research subject is vital to the
successful outcome of the proposed study. Among candidate species, the speckled dace,
Rhinichthys osculus, is an excellent choice for phylogeographic study of the drainage
history of western North American due to five characteristics:
1. Restriction to aquatic habitats. As with any freshwater fish, speckled dace reside
exclusively in freshwater lakes and rivers.
2. Widespread Distribution. Fish classified as speckled dace are found in 32 of
Smith’s 48 intermountain drainages and are native to the major drainages from
Sonora, Mexico and the Colorado River basin in Arizona and New Mexico to the
Columbia River in British Columbia (Mckell 2003). This has obvious advantages
for the study of the intermountain region.
3. Low Vagility. Ironically, despite their widespread distribution, speckled dace tend
to remain stationary once they invade a particular geographic region. Thus, the
evolutionary history of specific populations is directly connected to the geological
history of their geographic location (Bernatchez and Wilson 1998).
4. High Ecological Diversity. Speckled dace are not restricted to any particular type
of freshwater habitat, but occur in locations as diverse as cold mountain streams
and isolated desert springs (Mckell 2003). This means that dace are not only
distributed over a large geographic range, but cover that range completely,
inhabiting nearly every possible freshwater location.
5. High Potential Reproductive Rate. A mature female speckled dace (at least two
years old) can produce anywhere from 450-2000 eggs in a single season (Peden
and Hughes 1981). The success of speckled dace can be attributed in part to their
prolific production of progeny, making them resistant to local extinction and good
candidates for invasion of other habitats due to overpopulation.
Speckled dace have long been recognized as a relatively undefined species. Smith et
al. (2002) list 16 ‘subspecies’ of Rhinichthys osculus that are classified based on
morphological characteristics, such as number of pharyngeal teeth or type of frenum and
barbels. Smith et al. (2002) further admit that many of the geographic and morphological
variants of R. osculus could be defined as completely separate species depending on what
species model is used to evaluate them. If a phylogenetic model is used, for instance,
various clades of the clearly nonmonophyletic speckled dace could easily be redefined as
individual species (Oakley 2004, Smith et al. 2002).
Similar studies of speckled dace have previously been undertaken by Mckell (2003),
and Pfrender et al. (2004). The current study extends the database of speckled dace
cytochrome b sequences to include parts of the geographic range not yet been examined
(see Figure 1) and adds some resolution to the clade designations made by McKell. The
current study also lends further support to the classification of different speckled dace
populations as distinct species within the Rhinichthys genus.
MATERIALS AND METHODS
Speckled dace were collected from 28 locations in Utah, Idaho, Nevada, and
Wyoming by means of electrofishing. Once incapacitated, the dace were overdosed with
Tricaine Methanesulfonate (MS-222) and subsequently preserved by immersion in a 95%
At Brigham Young University, speckled dace specimens were accessioned into
the fish collection of the Monte L. Bean Life Science Museum. Collections from each
sample site were given unique population numbers and individual fish were assigned
their own accession database number. Tissue samples, predominately fin clips but
including occasional muscle samples, were then collected from a proportion of the fish in
each population. In the case of smaller populations (less than 10 fish), the entire
population was sampled. In large populations, tissue samples were obtained from
between 10-12 fish in most cases, though a few populations were sampled more
Total genomic DNA was isolated from the tissue samples using the following
procedure: 600 µl of Cell Lysis Solution was added to about 30 mg of tissue in a 1.5 ml
tube and homogenized using scissors. The tissue homogenate was then incubated at
65°C for 15 minutes. After incubation, 3 µl of proteinase K solution (20 mg/ml) were
added to the lysate and incubated overnight at 55°C. The next day, 3 µl of RNase A were
added to the samples and incubated for 15 minutes at 37°C. Samples were then cooled to
room temperature and mixed with 200 µl of Protein Precipitation Solution by 20 seconds
of vigorous vortexing. Each sample was then centrifuged at 15,000g for 3 minutes,
forming a protein pellet at the bottom of the tubes. The DNA-containing supernatant was
then poured into a fresh 1.5 ml tube containing 600 µl of absolute isopropanol. The tubes
were inverted 50 times to mix the solution and centrifuged at 15,000g for another 3
minutes. The supernatant was discarded and the DNA pellet washed by addition of 600
µl of 140 proof ethanol, which was then centrifuged for 1 minute at 15,000g. The
supernatant was once again discarded. The DNA pellets were dried for 30 minutes and
then resuspended in 200 µl of sterile distilled water. DNA isolation was confirmed by
DNA visualization in a 1% agarose gel.
The mitochondrial DNA cytochrome b (cyt b) gene was amplified from isolated
total genomic DNA using the polymerase chain reaction (PCR). PCR was performed on
a PTC-225 Peltier Thermal Cycler using the following protocol: a master mix was
created containing 2 µl of 10X PCR buffer, 2 µl dNTPs, 1 µl forward primer, 1 µl reverse
primer, 0.15 µl Taq polymerase, and 11.85 µl sterile distilled water for each sample to be
amplified. Eighteen microliters of that master mix were then added to 2 µl of total
genomic DNA from each sample. The PCR reaction was run through 35 cycles of 20 sec
at 94°C, 30 sec at 50°C, and 90 sec at 72°C. An initial DNA denaturation of 4 min at
95°C and a final extension of 7 min at 72°C proceeded and followed the 35-cycle
protocol, respectively. The cyt b gene was amplified in two segments using two primer
sets: HA (CAACGATCTCCGGTTTACAAGAC) and LD (CCATTCGTCATCGCCGGTGC); LA
(GTGACTTGAAAAACCACCGTTG) and HD (GGGTTGTTTGATCCTGTTTCGT).
Amplified PCR product was used to perform the Big Dye Primer Cycle
Sequencing Ready Reaction Kit M13 (Applied Biosystems, Foster City, CA) in
preparation for sequencing. The big dye reaction was filtered with Sephadex onto a 96-
well plate and sequenced by Brigham Young University DNA Sequencing Center on
either a Perkin-Elmer ABI Prism 377 automated sequencer or an ABI 3100 automated
sequencer. Sequencer 4.6 was used to analyze and align the nucleotide sequences.
Cyt b DNA sequences of dace colleted from the Salina River were aligned against
speckled dace cyt b sequences generated by Matthew McKell in 2003. These Salina
sequences were then used as references for the alignment of subsequent dace cyt b
sequences. Using the Sequencer software, the speckled dace cyt b nucleotide sequences
were separated into contigs based on population and compared for sequence divergence.
Each population was separated into unique haplotypes and a consensus sequence for each
haplotype was generated for use in the statistical parsimony (TCS), Bayesian inference
(BI), and maximum likelihood (ML) analyses. Cyt b sequences from longnose dace,
Rhinichthys cataractae, and Utah chub, Gila atraria were used as outgroups (M.
McKell, personal communication).
TCS version 1.21 (Templeton et al. 1992) was used to create phylogenetic
networks among the various speckled dace cyt b haplotypes.
MrBayes 3.04b4 was used to perform Bayesian inference without an a priori
evolution model. The BI analysis was sampled every 1000 trees for 10,000,000
generations. Each sample point prior to stationarity (ranging from 17-60) was later
thrown out and a consensus phylogenetic tree based on 50% majority rule for nodal
support was created with PAUP*4.0b10 software using the remaining trees (Swofford
Modeltest (Posada and Crandall 1998) was used to determine the GTR evolution
model as the most appropriate for the speckled dace data set. The ML analysis was run
using the PHYML program based on the GTR a priori model and a consensus tree was
created using PAUP*4.0b10 software (Swofford 2002).
Of the 28 locations (see Figure 1; Table 1) from which speckled dace were
sampled, only fish from 18 locations were successfully sequenced. Five populations
(Muddy Creek, Warren Bridge, Mary’s River, North Fork Little Humbolt River, and
Blackfoot River) yielded no DNA on successive isolation attempts. It is presumed that
the tissue from those five populations was inadequately preserved. Other populations
(Pacific Creek, Mammoth Creek, Black’s Fork, Raft River, and Teton River) generated
DNA, but the cyt b gene could not be amplified in repeated attempts using the protocols
and primers discussed above.
Although the entire cytochrome b gene nucleotide sequence (1140 bp) was
generated for most of the successfully sequenced dace populations, the ends of the gene
(72 bp on the 5’ side, 92 bp on the 3’ side) showed chromatograms sufficiently unreliable
across a significant proportion of the 181 individual fish as to render those portions
questionable for phylogenetic analysis. Thus, a total of 976 bp (beginning at position 73)
of the total cyt b gene were used in the TCS, BI, and ML analyses. Of those, 221
nucleotides were variable among the sequenced populations; there were 179 variable
nucleotides at the third codon position, 9 variable nucleotides at the second codon
position, and 33 variable nucleotides at the first codon position. Sixty unique haplotypes
were identified from the 181 fish. Only one haplotype, Yell1, was shared among
populations; it was also found in La Chappelle and Bear River speckled dace.
The BI (see Figure 2) and ML (see Figure 3) analyses show nearly identical
results and mainly differ in the bootstrap values of their phylogenetic branching. For
simplification, the BI consensus tree will serve as the reference for the following
exposition of the results and discussion. The BI analysis separates the speckled dace
populations into three distinct and well-supported clades: the Northern Bonneville Basin
clade, the Lahontan Basin clade, and the Southern Bonneville Basin clade. A fourth
clade is composed of longnose dace from Goose Creek; these fish were used as an
The Goose Creek longnose dace clade is basal to all three clades of speckled dace.
The Southern Bonneville Basin clade is the next to break off, followed by the split
between the Lahontan Basin and Northern Bonneville Basin. Strong bootstrap support
exists for all of these clades. The Northern Bonneville Basin clade separates into two
subgroups, one containing the Cotton Creek haplotypes and the other subgroup composed
of the Fish Creek (FshSn), La Chapelle, Bear River, Beaver Creek, Main Creek, and
Yellow Creek haplotypes. The latter group also contains the only haplotype shared
between two or more populations (Yell1).
The Lahontan Basin clade further separates into a subgroup composed of all the
Truckee River, McDermitt Creek, and Deep Creek haplotypes, as well as some of the
East Fork (E. Fk.) Carson River haplotypes. The other branch of the Lahontan Basin
clade contains the remaining E. Fk. Carson haplotypes and all of the East Walker River
haplotypes. Both groupings have strong bootstrap values, 73% and 100%, respectively.
The Southern Bonneville Basin clade contains dace from Lake Creek, Fish Creek
(Fish1), Clear Creek, and Salina Creek of the Bonneville Basin, and dace from the Fish
Creek (FshP) that is a tributary to the Price River of the Colorado River Basin. The Fish
Creek (FshP) dace resolve separately from the other fish in this clade and form a basal
polytomy to the other haplotypes.
The TCS analysis (see Figure 4) shows four major networks that correlate with
the four distinct clades evident from the BI phylogeny. The smallest of these major
networks (I) contains the exact same haplotypes as found in the Northern Bonneville
Basin clade. A second network (II) consists of most of the haplotypes found in the
Lahontan Basin clade, excluding two of the E. Fk. Carson haplotypes and two of the East
Walker River haplotypes, the four of which do not fall into any specific network, but are
distantly connected to the second network (II) via the outlying McDermitt haplotypes.
The Cotton Creek haplotypes also form their own isolated network, though they are most
closely connected to the first network (I). A third network (III) coincides with the
Southern Bonneville Basin clade and includes haplotypes from the Salina River, Clear
Creek, Lake Creek, and Fish Creek (FshP). Finally, the longnose dace Goose Creek
haplotypes form their own network (IV).
The clades that emerge from the BI, ML, and TCS analyses all show a strong
geographic signature. These groupings coincide and support some of the clade
designations described by McKell (2003). The Lahontan Basin clade, for instance, is
likely encompassed by the Western clade defined by McKell. Similarly, the Northern
Bonneville Basin clade can be presumed to be continuous with McKell’s Northern clade,
and the Southern Bonneville Basin clade appears to overlap McKell’s Southeastern clade.
In McKell’s phylogenetic analysis, he points out that within his Western clade (or
the Lahontan Basin clade as described here), there is a great deal of sequence divergence.
Indeed, that is also evident from the analysis presented in this paper. Sequence
divergence among the haplotypes in the Lahontan Basin clade ranges from less than one
percent (between Deep Creek and Truckee) to as much as much as 6.5% (E. Walker and
McDermitt), whereas the total sequence divergence among the Northern clade (excluding
the Cotton haplotypes) is a little more than 2%. It seems, however, that the Western
clade could be further resolved by excluding the Gandy Marsh population. Gandy Marsh
is located close to Lake Creek. Considering that speckled dace from Lake Creek resolve
in the Southern Bonneville Basin clade, it is possible that the Lake haplotypes act as a
bridge connecting the Gandy Marsh to the Southern Bonneville Basin clade.
Although McKell’s Southeastern clade and the Southern Bonneville Basin clade
show considerable geographic overlap, it is strange that the Lake, Clear, and Fish Creek
(Fish1) populations show such strong affinities in the phylogeny presented here, whereas
the Gandy Marsh population (located slightly North of Lake Creek, but still in the same
system) does not appear related to the Sevier, Bear, or San Pitch.
Another notable aspect of the Southern Bonneville Basin clade is that the dace
from Fish Creek (FshP) are most closely related to the Southern Bonneville Basin fish.
Because Fish Creek (FshP) is part of the Colorado River Basin, this relationship supports
the hypothesis that there existed an ancient connection between the Colorado River Basin
and the Southern Bonneville Basin, two basins that are now isolated from one another.
Within the Lahontan Basin clade, it is interesting to note that the haplotype from
Deep Creek, which is part of the Owyhee system, is included in that clade. The Owyhee
system currently interdigitates with the Lahontan Basin in Nevada and it is known that a
Pliocene connection existed between the Lahontan Basin and the Snake River (Smith
1975). The inclusion of Deep Creek in the Lahontan Basin clade is consistent with this
former connection. Furthermore, the sequence divergence between the Deep Creek and
McDermitt haplotypes, which are the most proximate of the Lahontan Basin clade
populations, ranges between 1-2%. Allowing for one percent sequence divergence per
million year separation, 1-2% divergence correlates to approximately half a million to a
million years of separation (Smith et al. 2002).
Another curious feature of the Lahontan Basin clade is that some of the E. Fk.
Carson haplotypes (EFCar2, EFCar3) resolve with some of the East Walker River dace,
indicating migration between those two rivers. Moreover, three other E. Fk. Carson
haplotypes (EFCar5, EFCar6. EFCar7) associate with the Truckee haplotypes. Thus, it
seems reasonable that fish from the Truckee migrated south into the Carson, and then
later migrated further into the Walker, giving rise to the Walker haplotypes. The most
recent aquatic connection between these areas occurred between 10 and 20 thousand
years ago and helps explain the close relationship between these fish (Grayson 1993).
The Lahontan Basin clade and the Northern Bonneville Basin clade are clearly
more related to one another than either is to the Southern Bonneville Basin speckled dace.
There must therefore have been an ancient connection between what is now the Lahontan
Basin and the Northern Bonneville Basin. To further investigate this possible connection,
more sampling and characterization of speckled dace in the Snake River Basin, southern
Idaho, and eastern Oregon is needed. Further analysis of the Snake River system should
help to resolve the connection between the Northern Bonneville Basin and Lahontan
Perhaps the most striking aspect of speckled dace is their genetic diversity. If the
one percent sequence divergence per million years asserted by Smith et al. (2002) can be
relied upon, then it is clear that the geographic variants of R. osculus as presently
constituted have been diverging for a long time (See Appendix A).
The TCS analysis (see Figure 4) adds further support to this idea. The Northern
Bonneville Basin network is separated from the Lahontan Basin network by at least 25
steps. The Lahontan Basin is itself 45 steps from the GC clade and 35 steps from the
Southern Bonneville Basin clade. Considering the relationship between clade and
geography, and the extent of the sequence divergence between these clades, it is
reasonable that some of the subspecies classifications of R. osculus should be elevated to
In fact, most of subspecies listed by Smith et al. (2002) were collected from small
geographic regions in Nevada. The total range of what is now considered speckled dace,
however, covers essentially all the Pacific drainages in the United States (Lee et al. 1980).
Because the fish outside of Nevada are almost entirely morphologically undescribed, it is
likely that genetic and morphological diversity to be found in these other speckled dace
could form criteria upon which the R. osculus could be subdivided into multiple species.
It is possible that some of the previously characterized subspecies that correlate
with one of the phylogenetic clades described in this study can simply be reclassified as
species themselves. Rhinichthys osculus adobe (Jordan and Evermann), for instance,
which corresponds geographically with some of the Southern Bonneville Basin clade
dace, might be upgraded to the rank of separate Rhinichthys species.
Alternatively, a morphological examination of the speckled dace populations used
in this study would be useful in determining if any phenotypic differences correlate to the
above clades. New species could then be determined based on those analyses. This is an
approach that has proven fruitful in the examination and potential classification of
various populations of leatherside chub, Lepidomeda copei, into separate species.
Johnson et al. (2004) examined leatherside chub on the basis of various species models
and concluded that the northern and southern clades were distinct species. This was
based, in part, on the 8% sequence divergence between the two clades.
Similarly, sequence divergences between the different clades of speckled dace are
large. For instance, divergence between the Salina or Clear Creek (Southern Bonneville
Basin) speckled dace and those from the Bear River or Fish Creek (FshSn) (Northern
Bonneville Basin) fluctuates between 7 and 8 percent. The Northern Bonneville Basin
and Lahontan Basin diverge by 6-7%, and some values for sequence divergence
measured between the Truckee (Lahontan Basin) and Salina (Southern Bonneville Basin)
dace are as high as 9%. In contrast, the divergence between the Goose Creek Longnose
dace and the speckled dace in this study averages about 9%, but is as low as 8% in some
instances (vs. Cotton or Deep).
One cautionary indicator is that the sequence divergence within one population is
quite high: some of the E. Fk. Carson River haplotypes differ from one another by as
much as 6%. These are the same haplotypes, however, that resolve with the E Walker
River haplotypes. It is possible that the E. Fk. Carson River haplotypes represent cryptic
species (i.e. two genetic lineages that overlap geographically, but are reproductively
isolated), but an analysis of nuclear markers to examine gene flow would be needed to
support such a conclusion. In the other populations, however, little variance occurs
between individuals within those population, indicating that differences between clades
may indeed represent separate species.
Advise, J.C. 2000. Phylogeography: the history and formation of species. Harvard Univ.
Press, Cambridge, MA.
Bernatchez, L., Wilson, C.C., 1998. Comparative phylogeography of Nearctic and
Palearctic fishes. Mol. Ecol. 7, 431–452.
Grayson, Donald K. 1993. The Desert’s Past: A Natural Prehistory of the Great Basin.
Johnson, Jerald B., Thomas Dowling, and Mark Belk. 2004. Neglected Taxonomy of
Rare Desert Fishes: Congruent Evidence for Two Species of Leatherside Chub.
Syst. Biol. 53(6):841–855, 2004.
Jordan, D. S., and B.W. Evermann. 1896. The Fishes of North and Middle America.
Bulletin of the United States National Museum, 47(part 1):1-618.
Lee, D. S., Gilbert, C. R., Hocutt, C. H., Jenkins, R. E., McAllister, D. E., and J. R.
Stauffer, Jr. 1980, et. seq. Atlas of North American Freshwater Fishes. North
Carolina State Museum of Natural History, Raleigh, NC.
Mckell, MD. 2003. Phylogeography of speckled dace, rhinichthys osculus (Teleostei:
Cyprinidae), in the intermountain west, USA. Master’s Thesis, Brigham Young
Oakey, David D., et al. Small Fish in a Large Landscape: Diversification of Rhinichthys
osculus (Cyprinidae) in Western North America. Copeia. Vol. 2004, No. 2, pp.
Peden, A.E., Hughes, G.W., 1981. Life history notes relevant to the Canadian status of
the speckled dace (Rhinichthys osculus). Syesis.14, 21–31.
Pfrender, Michael E., Justin Hicks, and Michael Lynch. 2004. Biogeographic patterns
and current distribution of molecular-genetic variation among populations of
speckled dace, Rhinichthys osculus (Girard). Molecular Phylogenetics and
Evolution. 30: 490–502
Posada, D., and K.A. Crandall. 1998. MODELTEST: testing the model of DNA
substitution. Bioinformatics. 14 :817-818.
Smith, G.R. 1975. Fishes of the Pliocene Glenns Ferry Formation, southwest Idaho.
Univ. Mich. Pap. Paleontol. 14:1-68.
Smith, G.R. 1978. Biogeography of Intermountain fishes. Great Basin Nat. Mem. 2:17-
Smith, G. R., T. Dowling, K. Gobalet, T. Lugaski, D. K. Shiozawa, and R. P. Evans.
2002. Biogeography and timing of evolutionary events among Great Basin fishes.
Pp. 175-234 in Robert Hershler, David B. Madsen, and Donald R. Currey (eds).
Great Basin Aquatic Systems History. Smithsonian Contributions to the Earth
Sciences, number 33. Smithsonian Institution Press. Washington D. C.
Swofford, D.L. 2002. PAUP*:phylogenetic analysis using parsimony (*and other
methods). Ver. 4.0b10. Sinauer Assoc., Sunderland, Mass.
Templeton, A.R., K.A. Crandall, and C.F. Sing. 1992. A cladistic analysis of phenotypic
associations with haplotypes inferred from restriction endonuclease mapping and
DNA sequence data. III. Cladogram estimation. Genetics. 132:619-633.