SPECIATION, PHYLOGEOGRAPHY, AND GENE FLOW
IN GIANT SALAMANDERS (DICAMPTODON)
By
CRAIG A. STEELE
A dissertation submitted in partial fulfillment of the requirements for degree of
DOCTOR OF ZOOLOGY
WASHINGTON STATE UNIVERSITY Department of Biological Sciences
i
DECEMBER 2006
To the Faculty of Washington State University:
The members of the Committee appointed to examine the dissertation of
CRAIG A. STEELE find it satisfactory and recommend that it be accepted.
____________________________________ Chair ____________________________________ ____________________________________ ____________________________________
ii
ACKNOWLEDGMENTS
Many people helped make the completion of this onerous and seemingly never ending
project possible. Their contributions to my research projects were greatly appreciated in my
many times of need and I recognize their help.
First, I’d like to thank my chair and my committee for their comments and constructive
criticisms and for always pushing me to achieve my full potential. Their efforts and guidance
helped me reach a level of excellence that I would not have been able to achieve eon my own.
Several people deserve special recognition for their expertise and involvement in the
project. Bryan Carstens was indispensable for the completion of the first chapter. I am grateful
for his patient tutoring of phylogenetic and phylogeographic methods. He and his research
colleges also provided many tissues samples, DNA sequences, and species specific primers for
my projects. Andy Giordano also deserves special recognition for his efforts in genotyping
hundreds of samples and his involvement in the data analysis required for the final chapter on
gene flow.
The following people provided much appreciated help with field work and had the
physical stamina needed to catch hundreds of salamanders: Alma Hanson, and Cyndi White.
Thanks to E.D. Brodie, Jr. for providing samples from the remote Shoat Springs location and
Mike Patterson for guiding me to the hard-to-reach Fox Creek locality in Northwest Oregon to
find Cope’s giant salamanders. The following provided indispensable explanations of analytical
techniques and training of standard lab techniques: Steve Mech, and Don Traul, and Kristen
Lew. Insightful discussions about data analysis, appropriate analytical techniques and the
iii
troubleshooting of programs were provided by: Devin Drown, Matt King, Eric Roalson, Mike
Alfaro, Melanie Murphy, Steve Spear, and Caren Goldberg.
I thank the Washington and Oregon Departments of Fish and Wildlife and the Idaho
Department of Fish and Game for issuing the permits necessary for the collection of samples in
the field. I also extend my appreciation to the Museum of Vertebrate Zoology at University of
California, Berkeley for generously providing tissue for analysis. The all-important funding was
provided through Washington Department of Fish and Wildlife, the James R. King Fellowship,
the Brislawn Fellowship and interdepartmental stipends by the School of Biological Sciences.
Finally, I would like to thank my wife, Maria Ortega, who not only spent several summer
“vacations” with me looking for salamanders, but also proofread nearly everything I wrote,
contributed to the creation of some figures, and provided much needed emotional support during
the roughest of times. Thank you so much, cariño.
Thank you everyone!
Craig A. Steele
iv
SPECIATION, PHYLOGEOGRAPHY, AND GENE FLOW
IN GIANT SALAMANDERS (DICAMPTODON)
Abstract
by Craig A. Steele, Ph.D. Washington State University
December 2006
Chair: Andrew Storfer
Giant salamanders of the genus Dicamptodon occur in the Pacific Northwest of North
America. The variety of geographic distributions and life history traits displayed among this
genus provide opportunities to test hypotheses concerning regional biogeography, effects of
Pleistocene glaciation, comparative phylogeography, and patterns of gene flow. A genus-level
phylogeny was constructed to test competing biogeographic hypotheses concerning the disjunct
distribution of the Idaho giant salamander (D. aterrimus), and a Pleistocene speciation
hypothesis for the Cope’s giant salamander (D. copei). Results indicate speciation and
distribution of D. aterrimus is attributable to the orogeny of the Cascade Mountains rather than
recent inland dispersal and that D. copei is distantly related to other coastal species and likely
originated much earlier than the Pleistocene. Patterns of intraspecific variation were examined
for the widespread Pacific giant salamander (D. tenebrosus) and hypotheses concerning the
location and number of Pleistocene refugia were tested. Results indicate that D. tenebrosus was
restricted to two Pleistocene refugia, one in the Columbia River valley and another in the
Klamath-Siskiyou Mountains, and has recently expanded northward from these refugia into its
v
current distribution. Phylogeographic patterns for D. copei were compared to that of the
codistributed Van Dyke’s salamander (Plethodon vandykei). Results reveal that sympatric
populations displayed identical phylogeographic topologies, suggesting shared evolutionary
histories, but topologies were ultimately incongruent due to several highly divergent allopatric
populations of D. copei. Comparative patterns of genetic population structure were examined for
sympatric populations of D. tenebrosus and D. copei. Results indicate that the metamorphosing
species, D. tenebrosus, displayed a lack of population structure while the non-metamorphosing
species, D. copei, displayed a larger degree of population structure. These results help explain
the phylogeographic patterns presented for each species. The large distribution and post-glacial
expansion by D. tenebrosus was facilitated by its high dispersal ability while the low dispersal
ability of D. copei lead to a small and fragmented geographic range and greater phylogeographic
structure within its range. These results suggest that understanding life history variation on a
local scale can lead to a better understanding of the mechanistic underpinnings of species’
distributions in general.
vi
TABLE OF CONTENTS
Page
ACKNOWLEDGEMENTS.................................................................................................. iii
ABSTRACT.......................................................................................................................... v
LIST OF TABLES................................................................................................................ viii
LIST OF FIGURES ............................................................................................................. ix
CHAPTER
INTRODUCTION................................................................................................. 1
1. TESTING HYPOTHESES OF SPECIATION TIMING IN DICAMPTODON COPEI
AND DICAMPTODON ATERRIMUS (CAUDATA: DICAMPTODONTIDAE)...... 4
2. COALESCENT-BASED HYPOTHESIS TESTING SUPPORTS MULTIPLE
PLEISTOCENE REFUGIA IN THE PACIFIC NORTHWEST FOR THE PACIFIC
GIANT SALAMANDER (DICAMPTODON TENEBROSUS)................................ 32
3. EVIDENCE FOR PHYLOGEOGRAPHIC INCONGRUENCE OF
CODISTRIBUTED SPECIES BASED ON SMALL DIFFERENCES IN GEOGRAPHIC
DISTRIBUTIONS..................................................................................................... 68
4. SCALING UP FROM LIFE HISTORY DYNAMICS TO PHYLOGEOGRAPHIC
PATTERNS: A COMPARATIVE STUDY OF TWO SYMPATRIC SALAMANDER
TAXA….................................................................................................................... 100
vii
LIST OF TABLES
1. Genetic Distances……….................................................................................................. 26
2. Locality Information for D. tenebrosus Samples.............................................................. 60
3. Results of Nested Clade Analysis for D. tenebrosus........................................................ 61
4. Genetic Distances within D. tenebrosus............................................................................ 62
5. Genetic Distances within D. copei.................................................................................... 92
6. Results for tests of phylogenetic concordance between D. copei and P. vandykei .......... 93
7. Results of Nested Clade Analysis for D. copei................................................................. 94
8. Pairwise FST values for D. tenebrosus and D. copei ......................................................... 117
9. Summary statistics for D. copei microsatellites ............................................................... 120
10. Summary statistics for D. tenebrosus microsatellites..................................................... 124
viii
LIST OF FIGURES
1. Distribution of Species....................................................................................................... 27
2. Constraint Trees for Phylogenetic Hypotheses.................................................................. 28
3. Phylogeny for Dicamptodon.............................................................................................. 29
4. Different estimates of Dicamptodon phylogeny................................................................ 30
5. Bayesian Posterior Probabilities of Different Root placements for Phylogeny................. 31
6. Distribution of D. tenebrosus............................................................................................. 63
7. Different Pleistocene Hypotheses for D. tenebrosus......................................................... 64
8. Phylogeny for D. tenebrosus..............................................................................................65
9. Haplotype Network for D. tenebrosus............................................................................... 66
10. Results of Nested Clade Analysis for D. tenebrosus....................................................... 67
11. Distribution of D. copei................................................................................................... 95
12. Phylogeny for D. tenebrosus............................................................................................96
13. Haplotype Network for D. copei...................................................................................... 97
14. Historical demographic patterns for Dicamptodon copei................................................ 98
15. Map of study area for comparative gene flow ................................................................ 117
16. Graphical output from the program STRUCTURE for D. copei......................................... 118
17. Graphical output from the program STRUCTURE for D. tenebrosus..................................119
ix
INTRODUCTION
Understanding the patterns of species distributions and the processes that lead to
those patterns is critical for understanding the evolutionary history of organisms, the past
and present ecological or environmental constraints of their distributions, and identifying
distinct lineages for conservation. This dissertation provides insights into the ecological
and evolutionary conditions that can shape or constrain geographical distributions and
genetic structuring of organisms. Chapters of this dissertation examine the genetic
structuring at different population scales and use the patterns detected to test hypotheses
concerning biogeography, phylogeography, and population level gene flow. Salamanders
of the genus Dicamptodon were chosen as study organisms for these projects not because
of their charisma, but rather because the variety of geographic distributions and life
history traits displayed among this genus provide ideal opportunities to test a variety of
evolutionary hypotheses that can advance our understanding about conditions that affect
the abundance and distribution of species.
Chapter one is an examination of alternative biogeographic hypotheses for the
Idaho giant salamander (D. aterrimus). This member of the genus is unique in that it
occurs in the Rocky Mountains of Idaho and is far removed from other members of the
genus that have a coastal distribution. Two alternative hypotheses exist for explaining the
disjunct distribution of D. aterrimus and invoke either ancient vicariance through the rise
of Cascade Mountains 3-5 million years ago, predicting reciprocal monophyly of inland
and coastal lineages, or more recent inland dispersal during the Pleistocene, whereby the
inland lineage is nested within a coastal lineage. Another hypothesis posits that the
Cope’s giant salamander (D. copei) speciated relatively recently from northern
1
populations of the Pacific giant salamander during the Pleistocene. This hypothesis
predicts that the D. copei lineage would be nested within D. tenebrosus. This chapter
focuses on constructing a genus-level phylogeny that is then used to test the
biogeographic and speciation hypotheses proposed for theses species.
Chapter two moves from a genus-level phylogeny to a species-level phylogeny
and focuses on the most broadly distributed species in the genus, the Pacific giant
salamander (D. tenebrosus). The large range of this species makes it ideal for testing
hypotheses about the effect of Pleistocene glaciation on the genetic structuring of
regional fauna. Alternative hypotheses exist concerning the number and location of
Pleistocene refugia in the Pacific Northwest. Putative refugia include a northern refugium
in the Columbia River valley and a southern refugium in the Klamath-Siskiyou
Mountains. Chapter two focuses on estimating a phylogeny for D. tenebrosus and uses a
coalescent simulation approach to test the competing hypotheses that during the
Pleistocene D. tenebrosus was restricted into either a single northern refugium, a single
southern refugium, or into the two refugia.
Chapter three continues to examine genetic structuring at a species-level but shifts
to the Cope’s giant salamander (D. copei). This species has the smallest distribution of
the genus and is codistributed with another salamander, the Van Dykes’s salamander
(Plethodon vandykei). Previous studies of P. vandykei support two reciprocally
monophyletic lineages corresponding to coastal populations and inland populations.
Comparative phylogeography of codistributed species provides understanding about the
role of climatic, geological, and ecological forces in shaping the geographic distribution
and intraspecific variation of species comprising an ecosystem. We hypothesized that D.
2
copei would have a congruent phylogeographic pattern due to ecological similarities and
similar habitat requirements to P. vandykei. Chapter three focuses on estimating a
phylogeny for the D. copei and then compares the topology with that for P. vandykei to
determine if these codistributed species shared similar evolutionary histories.
Chapter four moves down in scale from genetic patterns at the species-level to
patterns at the population level. This chapter examines the relationships between
variation in life history traits and the corresponding patterns of genetic population
structure. Two of the species of giant salamander, D. copei and D. tenebrosus, have
different dispersal potentials. One species, D. tenebrosus, commonly metamorphoses into
a terrestrial adult while D. copei remains in an aquatic state throughout its life. These
different life history traits result in contrasting dispersal potential from natal streams.
Chapter four examines how the different rates of dispersal affect gene flow and
population structuring for each species and tests the hypothesis that the low dispersal
species, D. copei, will display more genetic population structure than the high dispersal
species, D. tenebrosus. Results obtained at the population level are then used to explain
the phylogeographic patterns observed at the species-level for each organism. The low
dispersal potential of D. copei not only explains its small and fragmented distribution in
the Pacific Northwest, but also the large degree of phylogeographic structure in its small
range. Low dispersal has in effect made D. copei susceptible to a high degree of
population structuring. In contrast, the higher dispersal potential for D. tenebrosus can
help explain its large and continuous distribution as well as its phylogeographic pattern of
rapid post-glacial expansion. In this case, the high dispersal potential has a homogenizing
effect on the genetic structuring of the species.
3
The top-down approach of examining genetic structure at the genus, species and
population level allows for the testing of a variety of evolutionary hypotheses that could
only be answered by examining genetic patterns at the appropriate scale. In some cases
(e.g. chapter 4), the results seen at one scale help explain evolutionary patterns seen at
another. This demonstrates the utility of multi-scale approaches to population genetics.
Results provide not only answers to specific evolutionary hypotheses concerning the
study organism, but also provide insight into the role of environmental and ecological
factors shaping the abundance and distribution of species in general.
4
CHAPTER ONE
TESTING HYPOTHESES OF SPECIATION TIMING IN DICAMPTODON
COPEI AND DICAMPTODON ATERRIMUS (CAUDATA: DICAMPTODONTIDAE)
Abstract
Giant salamanders of the genus Dicamptodon are members of the mesic forest
ecosystem that occurs in the Pacific Northwest of North America. We estimate the
phylogeny of the genus to test several hypotheses concerning speciation and the origin of
current species distributions. Specifically, we test competing a priori hypotheses of
dispersal and vicariance to explain the disjunct inland distribution of the Idaho giant
salamander (D. aterrimus) and to test the hypothesis of Pleistocene speciation of Cope’s
giant salamander (D. copei) using Bayesian hypothesis testing. We determined that
available outgroups were too divergent to root the phylogeny effectively, and we
calculated Bayesian posterior probabilities for each of the 15 possible root placements for
this four-taxon group. This analysis placed the root on the branch leading to D. aterrimus,
indicating that current distribution and speciation of D. aterrimus fit the ancient
vicariance hypothesis and are attributable to the orogeny of the Cascade Mountains rather
than recent inland dispersal. Furthermore, test results indicate that D. copei is distantly
related to other coastal lineages and likely originated much earlier than the Pleistocene.
These results suggest that speciation within the genus is attributable to ancient geologic
events, while more recent Pleistocene glaciation has shaped genetic variation and
distributions within the extant species.
5
Introduction
A current trend in evolutionary biology has been the examination of speciation,
current range distributions, and patterns of genetic subdivision in the context of
Pleistocene climate change and the associated cycles of glacial advance and retreat
(Steinfartz et al. 2000; Sullivan et al. 2000; Austin et al. 2002; Church et al. 2003; Crespi
et al. 2003; Starkey et al. 2003; Zamudio and Savage 2003). Several studies invoke pre-
Pleistocene events or conditions to explain patterns of genetic structure, speciation, and
disjunct populations in the eastern and southeastern portions of the United States (Avise
and Walker 1998; Zamudio and Savage 2003; Donovan et al. 2000; Austin et al. 2000)
and recently in the western U. S., especially in the Pacific Northwest (Green et al. 1996;
Demboski and Cook 2001; Soltis et al. 1997).
The Pacific Northwest (PNW) of North America has been influenced by
numerous geological processes that have resulted in a complex and varied topography.
The combination of geologically ancient mountain ranges overlain with recent
Pleistocene glaciation provides a complex, yet well-defined historical context with which
to interpret genetic data (Cracraft, 1988; Riddle, 1996). As a result, tractable predictive
hypotheses are possible with respect to speciation and phylogeography of the region
(Brunsfeld et al., 2001). Within the PNW, coniferous rainforest ecosystems occur along
the western coast of North America, from southern Alaska to northern California, with a
disjunct inland forest in the northern Rocky Mountains (NRM) of British Columbia,
Idaho, and extreme western Montana. Mesic forests have long been established in the
PNW, dating to the mid Eocene in the NRM and were established in their present coastal
range by the early Pliocene (5-2 mya; Graham, 1993). The uplift of the Cascades
6
established a rain shadow that caused xerification of the Columbia basin prior to the
Pleistocene (2 mya; Daubenmire, 1952; Graham, 1993), which effectively divided mesic
forests into a coastal element and an inland element. Subsequent Pleistocene glaciation
resulted in severe southern compression of the PNW mesic forests during glacial maxima
(Waitt and Thorson, 1983; Delcourt and Delcourt, 1993; Soltis et al., 1997), and would
have forced mesic forest organisms into refugia.
The giant salamanders of the genus Dicamptodon are endemic to mesic forests of
the PNW. Members of this genus provide an ideal study system for examining
biogeographic hypotheses since the species are widespread throughout the western
United States with several species endemic to particular geographic locales. The genus
was originally considered monotypic (Tihen, 1958) but subsequent morphological
(Nussbaum, 1970; Nussbaum, 1976) and molecular studies (Daugherty et al., 1983;
Good, 1989) have resulted in recognition of four species (Fig. 1). D. copei is found
primarily in the Olympic Peninsula and Coast Range of Washington, D. ensatus is
restricted to regions surrounding the San Francisco Bay, and D. tenebrosus is widespread
and ranges from the Cascade Mountains in British Columbia in the north through
Washington and Oregon into California. D. tenebrosus forms a contact zone with D.
ensatus north of San Francisco and is sympatric with D. copei in parts of western
Washington and extreme northern Oregon. The fourth species, D. aterrimus, occurs in a
disjunct portion of the mesic forest ecosystem in northern Idaho and is geographically
isolated from the rest of the genus. Results of allozyme studies have consistently shown
that the highest genetic distances within the genus occur between D. aterrimus and
coastal species (Daugherty et al., 1983; Good, 1989), but relationships within coastal
7
lineages have not yet been resolved (Good, 1989). Here, we use mitochondrial DNA
sequence data to resolve relationships within Dicamptodon with two complementary
analyses. First, we test for monophyly of each of the four described species and second,
we test a priori hypotheses regarding speciation for D. aterrimus and D. copei derived
from biogeographic studies in the PNW mesic forest ecosystem.
The competing hypotheses relevant to speciation of D. aterrimus in the inland
mesic posit either pre-Pleistocene vicariance or post-Pleistocene dispersal (Brunsfeld et
al., 2001). The ancient vicariance hypothesis invokes pre-Pleistocene isolation of
ancestral D. aterrimus from the rest of the genus, associated with xerification of the
Columbia basin following the Cascade orogeny. It predicts deep genetic divergence and
reciprocal monophyly between D. aterrimus and coastal Dicamptodon species (Fig. 2a),
and requires that D. aterrimus persisted in a refuge located in one or more of the river
canyons south of glacial maxima throughout the Pleistocene. Phylogeographic studies of
two other PNW amphibian lineages endemic to the mesic forests, Ascaphus truei/A.
montanus (Neilson et al., 2001) and Plethodon vandykei/P. idahoensis (Carstens et al.,
2004), provide support for the ancient vicariance hypothesis. Alternatively, D. aterrimus
could be a post-Pleistocene arrival to the NRM, with either a southern dispersal route
through the central Oregon highlands or a northern route through southern British
Columbia and northern Washington as glaciers retreated. These inland dispersal
hypotheses predict a topology where D. aterrimus is nested within the clade from which
its ancestors originated: either the clade of northern D. tenebrosus haplotypes for the
inland dispersal-north hypothesis (Fig. 2b), or southern D. tenebrosus haplotypes for the
inland dispersal-south hypothesis (Fig. 2c).
8
A second taxon for which a priori hypotheses have been erected is D. copei
(Nussbaum, 1976), the only obligate neotene (aquatic gilled adult) within Dicamptodon.
Nussbaum (1976) proposed that ancestral populations of D. tenebrosus occurred
throughout western Washington and the Olympic mountains. During Pleistocene glacial
maxima, the Puget Sound lobe of the Cordilleran glacier isolated the coastal and Olympic
peninsular populations from the northern Cascadian populations of Dicamptodon. A
harsh terrestrial environment along with abundant pluvial habitat on the coast (Booth,
1987) would have favored an aquatic lifestyle and putatively led to speciation of the
neotenic D. copei (Nussbaum, 1976). This hypothesis predicts that D. copei would be
nested within northern populations of a parapyletic D. tenebrosus (Fig. 2d). Alternatively,
if speciation of D. copei predates the Pleistocene, reciprocal monophyly of D. tenebrosus
and D. copei would be predicted, with relatively deep divergence between these taxa.
In this study, we use DNA sequence data to estimate phylogenetic relationships
within Dicamptodon and explicitly test a priori hypotheses related to the speciation of D.
aterrimus and D. copei. In doing so, we evaluate the relative influence of pre-Pleistocene
and Pleistocene geological events on speciation within the genus.
Materials and methods
DNA extraction, amplification, and sequencing
We obtained tissue samples from throughout the geographic range of each species
and included: 12 D. copei from 5 populations, 46 D. tenebrosus from 16 populations, 10
D. ensatus from 5 populations, and used sequence from 6 D. aterrimus from Carstens et
al. (2005) that represented the greatest divergence within this species (Fig. 1; Appendix
9
1). Sample sizes ranged from 1-5 with an average of 2.8 samples per population. The
Ambystomatidae is traditionally considered to be the sister taxon to the
Dicamptodontidae (Larson, 1991), and we used sequence from Ambystoma mexicanum as
a putative outgroup.
DNA was extracted from 10-20 mg tail clips, which had been stored in 90%
EtOH, either with the DNeasy Tissue kit (Qiagen, Inc.; Valencia, CA), following
manufacturer’s instructions for rodent tails or using a standard phenol/chloroform
extraction protocol (Sambrook et al., 1989). To amplify the Cytochrome b gene (Cyt b),
we used the following primers from Carstens et al. (2005): tRNA-Threonine (5’—
TTCAGCTTACAAGGCTGATGTTTT—3’); tRNA-Glucine (5’—TTGTATTCAACTATAAAAAC—
3’); forward internal 5’—TCCACCCATACTTTTCTTATAAAGA—3’; reverse internal 5’—
TAATTAGTGGATTTGCTGGTGTAA—3’). Amplicons were purified using polyethylene
glycol precipitation, and sequencing reactions were performed with the BigDye Kit
version 2.0 (Applied Biosystems, Inc.; Foster City, CA) with 20-40 ng of PCR product in
15 µl volumes. CentriSep columns (Princeton Separations; Adelphia, NJ) were used to
filter sequencing reactions, and samples were run on an ABI 377 automated sequencer
using 5% Long Ranger polyacrylamide gels. Cyt b was sequenced in both the 5’ and 3’
directions, and edited and aligned with Sequencer 3.0 (GeneCodes; Ann Arbor, MI).
Sequences were deposited in GenBank (Appendix 1).
Phylogenetic Analyses
We generated maximum parsimony (MP) and maximum likelihood (ML)
estimates of the phylogeny to identify major clades and to test phylogenetic hypotheses in
10
Dicamptodon. We pruned all redundant haplotypes, and performed searches with PAUP*
4.0 (Swofford, 2002), both with A. mexicanum as an outgroup and without an outgroup.
The MP searches were conducted with stepwise-addition starting trees (150 random-
addition replicates) and TBR branch-swapping. For the ML analysis, we used DT-
MODSEL (Minin et al., 2003) to select a model of sequence evolution; this method
incorporates fit, a penalty for over-parameterization, and performance into model
selection. It also selects simpler models than other automated model-selection methods
(e.g., Modeltest; Posada and Crandall, 1998) and estimates phylogeny as accurately as
more complex models (Abdo et al., in press). We then conducted heuristic searches under
ML with the chosen model, and TBR branch-swapping, and 10 random addition-
sequence replicates. Nodal support for both the MP and ML tree was assessed using non-
parametric bootstrap values (Felsenstein, 1985), computed from 200 replicates. We
estimated the phylogeny of Dicamptodon with A. mexicanum as an outgroup using both
ML and MP. For the MP analysis, we translated nucleotide sequence into amino acids,
and conducted a MP search on these data to attempt rooting based only on slowly
evolving characters.
Bayesian Hypothesis Testing
Recent advances in phylogenetic methods allow evolutionary biologists to
conduct tests of a priori hypotheses with several approaches. However, regardless of the
method used, testing hypotheses shown in Figure 2 requires rooting the Dicamptodon
phylogeny. Although the family Ambystomatidae is likely to be the sister taxon to
Dicamptodontidae (Larson, 1991; Larson and Dimmick, 1993), fossil evidence (Estes,
11
1981) suggests that Dicamptodon have been independent of the Ambystoma lineage for a
considerable period of time and may be too divergent to serve as a reliable root. We
explored two approaches for rooting our phylogeny estimate: outgroup rooting and
rooting under the molecular clock hypothesis. We examined the effectiveness of rooting
the phylogeny with A. mexicanum by conducting Bayesian estimation (using MRBAYES;
Huelsenbeck and Ronquist, 2001) and determining the posterior probability of each of 15
possible root placements. In addition, we conducted a likelihood-ratio test (Felsenstein,
1988) of the molecular clock hypothesis and conducted Bayesian searches under a strict
molecular clock to determine the posterior probability of each of the 15 possible root
placements for a 4-taxon tree. In each analysis, we assumed each of the four species to be
monophyletic groups, based on the results of our ML tree, and filtered the posterior
distribution of topologies from Bayesian searches described below with filters that
corresponded to each possible root.
In addition, we used MRBAYES (Huelsenbeck and Ronquist, 2001) to assess the
posterior probability of each of the four a priori hypotheses described in the introduction:
ancient vicariance, inland dispersal-north, and inland dispersal-south hypotheses for D.
aterrimus, and Pleistocene speciation hypothesis for D. copei. Achieving stationarity with
respect to topology is critical for Bayesian hypothesis testing because we are assessing
topological predictions. The topology parameter may be particularly susceptible to non-
stationarity (Huelsenbeck et al., 2002), so we employed a stationarity test used by
Carstens et al. (2004), which is similar to one proposed by Huelsenbeck et al. (2002). We
conducted four independent heated runs (each composed of 4 Metropolis-coupled chains)
and started each run with a different random tree. We ran the chains for 3.1 x 106
12
generations and sampled trees every 1000 generations. If the four independent runs have
each converged on the true joint posterior-probability distribution, the four samples of
trees should represent independent samples drawn from that distribution. To assess this
expectation statistically, we saved the last 3000 trees from each run and computed the
symmetric-difference distance between each tree in the sample and our ML tree using
PAUP* 4.0. We then conducted a standard ANOVA on the 4 groups of tree-to-tree
distances to assess whether the four chains could have been drawn independently from
the same underlying joint posterior probability distribution. While a non-significant result
for this test would not guarantee that the runs have reached stationarity with respect to
topology, it would provide much stronger evidence of such than would the standard
examination of lnL plots (Huelsenbeck et al., 2002). To complete the hypothesis test, we
then imported the sample of trees from the Bayesian analysis into PAUP* and filtered
them with constraint trees predicted by each of the a priori hypotheses. The proportion of
trees in the sample consistent with the topology predicted by each hypothesis is the
Bayesian conditional probability that the hypothesis is correct.
Results
Sequencing
We sequenced all of Cyt b and a portion of the tRNA(Thr), corresponding to
positions 14109-15249 of the A. mexicanum mitochondrial genome, for 68 individual
giant salamanders. We translated the nucleotide data to amino acids, checked for stop
codons, and aligned the amino acids with other salamander Cyt b sequences to verify that
the pattern of molecular evolution was consistent with the mitochondrial DNA of
13
salamanders and inconsistent with the presence of nuclear psuedogenes. Data from Cyt b
and the tRNA(Thr), were combined into a single data set with a total of 1174 bases.
Several individuals had identical haplotypes (Appendix 1). In the final data set there were
6 D. aterrimus, 7 D. copei, 5 D. ensatus, and 25 D. tenebrosus haplotypes. Genetic
distances corrected with the HKY+I+Γ model of sequence evolution (see below) as well
as uncorrected percent sequence difference are shown in Table 1. Uncorrected divergence
ranged from 0.043 to 0.067 within Dicamptodon, and from 0.206 to 0.222 between
Dicamptodon and Ambystoma.
Phylogenetic Analyses
We selected the HKY+I+Γ model of sequence evolution using DT-MODSEL
(Minin et al., 2003) with equilibrium base frequencies of πA = 0.311 ; πC = 0.194 ; πG =
0.123 ; πT = 0.371; transition—transversion ratio = 3.698); proportion of invariable sites
= 0. 754; and Γ-distribution shape parameter (α = 1.57). The ML phylogeny estimate has
a likelihood score of –lnL = 3189.8281. When we enforced the molecular clock and
conducted a ML search, the resulting tree had a –lnL = -3237.8343. The likelihood ratio
test indicated that we could reject the molecular clock hypothesis (δ = 96.0124; p <
0.001). Other than a few relationships within northern D. tenebrosus, the MP phylogeny
(not shown) is identical to the ML phylogeny (Fig. 3). There is strong bootstrap support
for monophyly of haplotypes sampled from each of the four described species [D.
aterrimus (ML = 100% of the replicates, MP = 100 %); D. copei (ML = 90%, MP =
99%); D. ensatus (ML = 83%, MP = 84%); D. tenebrosus (ML = 94%, MP= 97%)], but
little support for relationships among the four species.
14
Bayesian Hypothesis Testing
Genetic distance between A. mexicanum and Dicamptodon (1.37 to 1.64
substitutions/site and 21.6% - 22.2% uncorrected divergence) lead us to question the
appropriateness of Ambystoma as an outgroup (Table 1). We explored this by adding
other salamander Cyt b sequences to the data matrix and estimating the phylogeny with
neighbor joining and uncorrected distances as a fast way to explore the sister-group
relationship between Ambystoma and Dicamptodon. In every case, A. mexicanum was the
sister taxon to Dicamptodon, but separated by an extremely long branch (data not
shown). Thus, while A. mexicanum was the best available outgroup, it may not be a
particularly good outgroup. Consequently, we compared the results of three different
rooting methods. A ML search of the data, using A. mexicanum sequence to root the
phylogeny, recover a paraphyletic D. ensatus as the sister taxon to a group containing all
other Dicamptodontidae (Fig. 4a). The phylogram illustrates the discrepancy between the
length of branches within the ingroup compared to the length of the outgroup branch. A
strict consensus of the most parsimonious trees in the parsimony search of amino acids,
again using A. mexicanum sequence to root the phylogeny, placed D. copei outside a
clade comprising all other dicamptodontids (Fig. 4b). We used Bayesian methods, again
with A. mexicanum as an outgroup, to estimate the posterior probabilities for each
possible rooting of the genus, and found little support for any of the root placements (D.
tenebrosus = 0.464; D. ensatus =0.391; D. copei =0.072; D. aterrimus = 0.027). There is
therefore little support for any root using the outgroup strategy.
Huelsenbeck et al. (2002) demonstrated that a Bayesian rooting under a clock
provides reliable roots, and that this conclusion is robust to violations of the clock
15
assumption. Therefore, we used MRBAYES to compute the probability of each of the 15
possible root placements assuming monophyly of each species. These analyses indicated
that D. aterrimus represents the earliest divergence in the genus and is the sister lineage
to the other giant salamanders (Fig. 5). This placement of D. aterrimus is consistent with
previous work using allozymes (which also found lack of an appropriate outgroup; Good,
1989), and is also the only strong signal for any rooting. We thus consider D. aterrimus
to be the sister taxon to the rest of the genus for hypothesis testing, but stress that this
placement is tentative.
The four independent Bayesian runs had average tree-to-tree distances from the
ML tree of 31.34, 31.56, 32.05, and 31.56. The ANOVA indicated that these values were
not significantly different (FOBS = 1.504; 0.1 > p > 0.05), a result which we interpreted as
evidence that independent Metropolis-coupled MCMC chains were sampling topologies
from the same joint posterior probability distribution and have likely achieved
stationarity with respect to topology. We discarded trees from the first 100,000 burn-in
generations, and combined 3000 trees from each run into a set of 12,000 trees that were
used to test the a priori hypotheses. For D. aterrimus, we could reject the inland dispersal
north hypothesis (p < 0.0001) and the inland dispersal south hypothesis (p < 0.0001), but
not the ancient vicariance hypothesis (p = 0.9936). For D. copei, we found that we could
reject the hypothesis of Nussbaum (1976) that proposed Pleistocene isolation from
northern D. tenebrosus (p < 0.0001) but could not reject the monophyly of either D. copei
(p = 1.0) or D. tenebrosus (p = 1.0). These results suggest that speciation within
Dicamptodon was largely shaped by pre-Pleistocene events.
16
Discussion
Our data provide additional insight to the systematics of the salamander genus
Dicamptodon and the biogeography of the PNW region by rejecting speciation
hypotheses that the genetic structure of the genus was primarily shaped by post-
Pleistocene events. For both D. aterrimus and D. copei, we were unable to reject
hypotheses that posited pre-Pleistocene isolation. The finding of monophyly in D. copei,
which at one time was considered polyphyletic (Daugherty et al., 1983), as well as the
other species, suggests that these lineages have been on separate evolutionary trajectories
for a significant amount of time. The rejection of the inland dispersal hypotheses for D.
aterrimus and the strong support for the ancient vicariance hypothesis is congruent with
the pattern that has been observed in other mesic-forest amphibians (including the tailed-
frog, Ascaphus truei/ A. montanus [Neilson et al., 2001] and the Plethodon vandykei/ P.
idaohensis complex [Carstens et al., 2004]).
Our findings, combined with previous research on amphibian members of this
mesic-forest ecosystem, strongly suggest that Dicamptodon was once widespread
throughout the PNW during the Miocene. Physical evidence such as dicamptodontine
fossils and trackways occurring as far east as Montana and North Dakota further support
this conclusion (Peabody 1954, 1959; Estes, 1981). While pre-Pleistocene xerification in
the Columbian Plateau was apparently responsible for separating inland and coastal
populations, there was a recent opportunity for gene flow between the two regions during
the Pleistocene along a northern corridor of mesic forests during the glacial maxima
approximately 25,000-10,000 years ago (Richmond et al., 1965; Barnosky et al., 1987).
Such a corridor has been invoked to explain subtle morphological similarities between
17
inland and north-coastal populations of Dicamptodon (Nussbaum, 1976). However, the
mesic-forest amphibians of the PNW are probably limited in their ability to disperse long
distances overland because they are either stream-breeding (Dicamptodon, Ascaphus) or
closely associated with seeps and streams (P. vandykei / P. idahoensis). While
phylogenetic patterns suggest inland dispersal along a northern corridor in several plant
species and small mammals (reviewed in Brunsfeld et al., 2001), there is no genetic
evidence for such a pattern in Dicamptodon, Ascaphus, or the Plethodon vandykei/P.
idahoensis complex.
Rejection of the Pleistocene-speciation hypothesis for D. copei suggests that
northern D. tenebrosus and D. copei are not as closely related as once thought, and that
speciation of D. copei occurred earlier than previously hypothesized (Nussbaum, 1976).
Genetic distance between D. copei and other Dicamptodon supports earlier speculation
by Daugherty et al. (1983) that D. copei is an ancient lineage, and offers clues about the
relative timing of divergence events. High divergences between D. copei and other
members of the genus suggest that D. copei diverged at approximately the same time as
D. aterrimus was isolated from coastal Dicamptodon. The basic premise of Nussbaum’s
hypothesis may be correct but would require that populations in the northern Cascades,
from which the ancestors of D. copei diverged, were unable to escape advancing glaciers
and that modern populations of D. tenebrosus have recently expanded into the north
Cascades. Testing this hypothesis will require additional D. tenebrosus sampling, explicit
phylogeographic modeling and coalescent-based hypothesis testing following Knowles
(2001) and Carstens et al. (2005).
18
Support for the sister-group relationship between D. tenebrosus and D. ensatus is
high in our analyses, as indicated by MP (72) and ML (67) bootstrapping and Bayesian
posterior probability (p = 0.9483). No obvious geographic barrier exists between D.
tenebrosus and D. ensatus that would suggest allopatric speciation. However, the split
occurs along the ‘North Coast Divide’ (Nussbaum, 1976), a low ridge that delineates the
southern range limit in some taxa and divides a variety of species into sub-species
(reviewed in Good, 1989). Such taxa include the transition of mountain kingsnake
subspecies Lampropeltis zonata zonata to the intergrade zone of L. z. zonata x
multicincta (Zweifel, 1952; McGurty, 1988) and the boundary between the Northern
Alligator lizard subspecies Elgaria coerulea coerulea and E. c. shastensis (Smith, 1995).
Recent geologic activity in this region of northern California is characterized by erosion
(Wahrhaftig and Birman, 1965), and it may be that the North Coast Divide delimited the
boundary of a coastal refuge for D. ensatus during the Pleistocene. D. ensatus has close
associations with the same redwood forest (Sequoia sempervirens) habitat as fossil
dicamptodontine salamanders and is thought to be more similar morphologically to
ancestral Dicamptodon than are other extant species (Nussbaum, 1976). It may be that D.
ensatus persisted throughout the Pleistocene in a southern refugium containing redwood
habitat similar to that occupied by ancestral Dicamptodon, while D. tenebrosus was
isolated in separate refugia to the north. Thus, the secondary contact between these forms
in northern California (Good, 1989) is the result of recent range expansion from their
respective refugia. Again, testing these hypotheses will require additional sampling and
explicit coalescent modeling.
19
Understanding the geological events that contributed to the speciation of extant
lineages is one of the primary goals of biogeographic research. This research is
complicated when the taxa in question are not closely related to other extant species, but
advances in computational phylogenetics allow for the testing of hypotheses even when
an appropriate outgroup is not available. As a result, our estimate of the Dicamptodon
phylogeny permits the following hypothesized history. During the Pliocene (5-2 mya),
the ancestors of D. aterrimus were isolated from other Dicamptodon by xerification of
the Columbia basin following the orogeny of the Cascades. By the end of the Pliocene or
early Pleistocene, the ancestors of D. copei were isolated from other coastal Dicamptodon
(~2 mya), and evolved obligate neoteny in a pluvial environment, likely as described by
Nussbaum (1976). The remaining coastal Dicamptodon lineages were divided into at
least two populations by Pleistocene climatic change; the southern populations evolved
into D. ensatus and northern populations evolved into D. tenebrosus. The support of the
ancient vicariance hypothesis for D. aterrimus and other mesic-forest amphibians
provides evidence that amphibian species in the PNW were similarly affected by pre-
Pleistocene events.
References
Abdo, Z., Minin, V., Joyce, P., Sullivan, J, in press. Accounting for uncertainty in the tree
topology has little effect on the decision theoretic approach to model selection in
phylogeny estimation. Mol. Biol. Evol.
Austin, J. D., Lougheed, S.C., Neidrauer, L., Chek, A.A., Boag, P.T., 2002. Cryptic
20
lineages in a small frog: the post-glacial history of the spring peeper, Pseudacris
crucifer (Anura: Hylidae). Mol. Phylogenet. Evol. 25, 316–329.
Avise, J. C., Walker, D., 1998. Pleistocene phylogeographic effects on avian populations
and the speciation process. Proc. R. Soc. Lond. B 265, 457–463.
Barnosky, C.W., Anderson, P.M., Bartlein, P. J., 1987. The northwestern U.S. during
deglaciation; vegetational history and paleoclimatic implications. In: Ruddiman,
W.F., Wright, Jr, H.E., (Eds.), North America and Adjacent Oceans During the
Last Deglaciation. The Geological Society of America, Boulder, pp. 289–321.
Booth, D.B, 1987. Timing and processes of deglaciation along the southern margin of the
Cordilleran ice sheet. In: Ruddiman, W.F., Wright, Jr, H.E., (Eds.), North
America and Adjacent Oceans During the Last Deglaciation. The Geological
Society of America, Boulder, pp. 71–90.
Brunsfeld, S.J., Sullivan, J., Soltis, D.E., Soltis, P.S., 2001. Comparative phylogeography
of Northwestern North America: A synthesis. In: Silvertown, J., Antonovics, J.
(Eds.), Integrating ecological and evolutionary processes in a spatial context.
Blackwell Science, Oxford, pp. 319–339.
Carstens, B.C., Stevenson, A.L., Degenhardt, J.D., Sullivan J., 2004. Testing Nested
Phylogenetic and Phylogeographic Hypotheses in the Plethodon vandykei Species
Group. Syst. Biol. 53, 781–792.
Carstens, B.C., Degenhardt J.D., Stevenson A.L., Sullivan J., 2005. Accounting for
Coalescent Stochasticity in Testing Phylogeographic Hypotheses: Modeling
Pleistocene Population Structure in the Idaho Giant Salamander Dicamptodon
aterrimus. Mol. Ecol.
21
Church, S.A., Kraus, J.M., Mitchell, J.C., Church, D.R., Taylor, D.R., 2003. Evidence
for multiple Pleistocene refugia in the postglacial expansion of the eastern tiger
salamander, Ambystoma tigrinum tigrinum. Evolution 57, 372–383.
Cracraft, J., 1988. Deep-history biogeography: retrieving the historical pattern of
evolving biotas. Syst. Zool. 37, 221–236.
Crespi, E.J., Rissler, L.J., Browne, R.A., 2003. Testing Pleistocene refugia theory:
phylogeographic analysis of Desmognathus wrighti, a high-elevation salamander
in the southern Appalachians. Mol. Ecol. 12, 969–984.
Daubenmire, R., 1952. Plant geography of Idaho. In: Davis, R.J. (Ed.), Flora of Idaho.
Brigham Young University Press, Provo, pp 1–17.
Daugherty, C.H., Allendorf, F.W., Dunlap, W.W., Knudsen, K.L., 1983. Systematic
implications of geographic pattern of genetic variation in the genus Dicamptodon.
Copeia 1983, 679–691.
Delcourt, P.A., Delcourt, H.R., 1993. Paleoclimates, paleovegetation, and paleofloras
during the late Quaternary. In: Flora of North America Editorial Committee
(Eds.), Flora of North America, Oxford University Press, NY, pp 71–94.
Demboski, J.R., Cook, J.A., 2001. Phylogeography of the dusky shrew, Sorex
monticolus (Insectivora, Soricidae): insight into deep and shallow history in
northwestern North America. Mol. Ecol. 10, 1227–1240.
Donovan, M.F., Semlitsch, R.D., Routman, E.J., 2000. Biogeography of the southeastern
United States: a comparison of salamander phylogeographic studies. Evolution
54, 1449–1456.
Estes, R, 1981. Gymnophiona, Caudata. In: Wellnhofer, P. (Ed.), Handbuch der
22
Paläoherpetologie, Part 2, Gustav-Fisher-Verlag, Stuttgart and New York, pp 1–
115.
Felsenstein, J., 1978. Cases in which parsimony or compatibility methods will be
positively misleading. Syst. Zool. 27, 401–410.
Felsenstein, J., 1985. Confidence limits on phylogenies: An approach using the bootstrap.
Evolution 39, 783–791.
Felsenstein, J., 1988. Phylogenies from molecular sequences: Inference and reliability.
Annu Rev Genet 22, 521–565.
Good, D.A., 1989. Hybridization and cryptic species in Dicamptodon (Caudata:
Dicamptodontidae). Evolution 43, 728–744.
Graham, A, 1993. History of the vegetation: Cretaceous (Maastrichtian)—Tertiary. In:
Flora of North America Editorial Committee (Eds.), Flora of North America,
Oxford University Press, NY, pp 57–70.
Green, D.M., Sharbel, T.F., Kearsley, J., Kaiser, H., 1996. Post-glacial range fluctuation,
genetic subdivision an speciation in the western North American spotted frog
complex, Rana pretiosa. Evolution 50, 374–390.
Huelsenbeck, J.P., Ronquist, F., 2001. MRBAYES: Bayesian inference of phylogenetic
trees. Bioinformatics 17, 754–755.
Huelsenbeck, J.P., Lagret B., Miller R.E., Ronquist F., 2002. Potential applications and
pitfalls of Bayesian inference of phylogeny. Syst. Biol. 51, 673–688.
Knowles, L.L. 2001. Did the Pleistocene glaciations promote divergence? Tests of
explicit refugial models in montane grasshoppers. Mol. Ecol. 10: 691-701.
Larson, A., 1991. A molecular perspective on the evolutionary relationships of the
23
salamander families. Evol. Biol. 25, 211–277.
Larson, A., Dimmick, W.W., 1993. Phylogenetic relationships of the salamander
families: A analysis of congruence among morphological and molecular
characters. Herpetol. Monogr. 7, 77–93.
McGurty, B.M., 1988. Natural history of the Californian mountain kingsnke
Lampropeltis zonata. In: De Lisle, H.F., Brown P.R., Kaufman B., McGurty B.M.
(Eds.), Proceedings of the Conference on California Herpetology, Southwestern
Herpetologists Society, Van Nuys, pp. 73–88
Minin, V., Abdo, Z., Joyce, P., Sullivan, J., 2003. Performance-based selection of
likelihood models for phylogeny estimation. Syst. Biol 52, 674–683
Neilson, M., Lohman K., Sullivan, J., 2001. Phylogeography of the tailed frog (Ascaphus
truei): Implications for the biogeography of the Pacific Northwest. Evolution 55,
147–160.
Nussbaum, R.A., 1970. Dicamptodon copei, n. sp., from the Pacific Northwest, U.S.A.
(Amphibia: Caudata: Ambystomatidae). Copeia 1970, 506–514.
Nussbaum, R.A., 1976. Geographic variation and systematics of salamanders of the
genus Dicamptodon Strauch (Ambystomatidae). Miscellaneous Publications No.
149, Museum of Zoology, University of Michigan, Ann Arbor.
Peabody, F.E., 1954. Trackways of living and fossil salamanders from the Paleocene of
Montana. J. Paleontol. 28, 79–83.
Peabody, F.E., 1959. Trackways of living and fossil salamander. University of California
Publications in Zoology 63,1–72.
Richmond, G.M., Fryxell, R., Neff, G.E., Weis, P.L., 1965. The Cordilleran ice sheet of
24
the northern Rocky Mountains and related Quaternary history. In: Wright, Jr,
H.E., Frey, D.G. (Eds.), The Quaternary of the United States, Princeton
University Press, Princeton, pp 231–242
Riddle, B.R., 1996. The molecular phylogeographic bridge between deep an shallow
history in continental biotas. Trends Ecol. Evol. 11, 207–211.
Sambrook, J., Fritsch, E.F., Maniatis, T., 1989. Molecular Cloning: A Laboratory Manual
(2nd Ed.). Cold Springs Harbor Laboratory Press, Plainview, New York.
Smith, H.M., 1995. Handbook of Lizards – Lizards of the United States and Canada.
Cornell University Press, Ithaca and London.
Soltis, D.E., Gitzendanner, M.A., Strenge, D.D., Soltis, P.S., 1997. Chloroplast DNA
intraspecific phylogeography of plants from the Pacific Northwest of North
America. Plant Syst. Evol. 206, 353–373.
Starkey, D.E., Shaffer, H. B., Burke, R.L., Forstner, M.R.J., Iverson, J.B., Janzen, F.J.,
Rhodin, A.G.J., Ultsch, G.R., 2003. Molecular Systematics, phylogeography, and
the effects of Pleistocene glaciation in the painted turtle (Chrysemys picta)
complex. Evolution 57, 119–128.
Steinfartz, S., Veith, M., Tautz, D., 2000. Mitochondrial sequences analysis of
Salamandra taxa suggests old splits of major lineages and postglacial
recolonization of Central Europe from distinct source populations of Salamadra
salamandra. Mol. Ecol. 9, 397–410.
Sullivan, J., Arellano, E., Rogers, D.S., 2000. Comparative phylogeography of
Mesoamerican highland rodents: Concerted versus independent responses to past
climatic fluctuations. Amer. Nat. 155, 755–768.
25
Swofford, D.L., 2002. PAUP*. Phylogenetic Analysis Using Parsimony (and Other
Methods). Version 4. Sinauer Associates, Sunderland, Massachusetts.
Tihen, J. A., 1958. Comments on the osteology and phylogeny of ambystomatid
salamanders: Bulletin of the Florida State Museum, Biological Sciences 3, 1–50.
Wahrhaftig, C., Birman., J.H., 1965. The Quaternary of the Pacific Mountains System in
California. In: Wright, Jr, H.E., Frey, D.G. (Eds.), The Quaternary of the United
States, Princeton University Press, Princeton, pp 299–340.
Waitt, Jr., R.B., Thorson, R.M., 1983. The Cordilleran ice sheet in Washington, Idaho,
and Montana. In: Wright, Jr., H.E., Porter, S.C. (Eds.), Late Quaternary
Environments of the United States, the Late Pleistocene, University of Minnesota
Press, Minneapolis, pp 53–70.
Zamudio, K.R., Savage, W.K., 2003. Historical isolation, range expansion, and
secondary contact of two highly divergent mitochondrial lineages in spotted
salamanders (Ambystoma maculatum). Evolution 57, 1631–1652.
Zweifel, R.G, 1952. Pattern variation and evolution of the mountain kingsnake,
Lampropeltis zonata. Copeia 1952, 152–168.
26
Table 1 Genetic Distances. Shown above the diagonal are genetic distances corrected under the HKY+I+Γ model of sequence evolution in units of substitutions per site. Uncorrected percent sequence divergences are shown below the diagonal.
D. aterrimus D. copei D. ensatus D. tenebrosus A. mexicanum
D. aterrimus — 0.0999 0.0666 0.0960 1.3726
D. copei 0.0656 — 0.0658 0.0820 1.6416
D. ensatus 0.0503 0.0434 — 0.0589 1.2578
D. tenebrosus 0.0670 0.0572 0.0455 — 1.5946
A. mexicanum 0.2155 0.2223 0.2061 0.2207 —
27
Fig. 1. Approximate distribution of the four species in the salamander genus Dicamptodon. Numbers indicate approximate sampling sites and correspond to numbers in Appendix 1. Populations sometimes included several nearby localities; refer to Appendix 1 for specific locality information.
28
Fig. 2. Constraint trees for phylogenetic hypotheses of interspecific relationships among Dicamptodon. The ancient vicariance hypothesis for the speciation of D. aterrimus (2a.) predicts reciprocal monophyly between the inland species of D. aterrimus and the reaming coastal clades. (2b.) The inland dispersal-north hypothesis for the speciation of D. aterrimus predicts a close relationship between D. aterrimus and northern populations of D. tenebrosus (population no. 1–3) while the inland dispersal-south hypothesis (2c.) predicts a close relationship between D. aterrimus and southern populations of D. tenebrosus. The Pleistocene speciation hypothesis for D. copei (2d.) predicts the close relationship between D. copei and northern populations of D. tenebrosus.
29
Fig. 3. The optimal ML phylogeny for the salamander genus Dicamptodon. An unrooted phylogeny (3a.) demonstrates the long branches between the four species. A ML phylogeny, rooted with D. aterrimus (3b.) from 43 unique mtDNA haplotypes of 1174 bp of cyt b and tRNA-threonine with lnL = 3189.8281. Numbers on branches are ML (above) and MP (below) bootstrap support of nodes retained in >50% of 200 replicates.
30
Fig. 4. Estimates of the Dicamptodon phylogeny using A. mexicanum as the outgroup. A maximum likelihood search of the sequence data showing a paraphyletic D. ensatus as the sister taxon to a group comprising all other dicamptodontids (4a.), a strict consensus of the most parsimonious trees from a search of amino acid data showing a paraphyletic D. copei outside a clade containing all other Dicamptodontidae(4b.).
31
Fig. 5. Bayesian posterior probabilities of each of the 15 possible root placements on a 4 taxon tree. The root placement with the highest posterior probability is the branch leading to D. aterrimus.
32
CHAPTER TWO
COALESCENT-BASED HYPOTHESIS TESTING SUPPORTS MULTIPLE
PLEISTOCENE REFUGIA IN THE PACIFIC NORTHWEST FOR THE
PACIFIC GIANT SALAMANDER (DICAMPTODON TENEBROSUS).
Abstract Phylogeographic patterns of many taxa are explained by Pleistocene glaciation.
The temperate rainforests of the Pacific Northwest of North America provide an excellent
example of this phenomenon, and competing phylogenetic hypotheses exist regarding the
number of Pleistocene refugia influencing genetic variation of endemic organisms. One
such endemic is the Pacific Giant Salamander, Dicamptodon tenebrosus. In this study, we
estimate this species’ phylogeny and use a coalescent modeling approach to test five
hypotheses concerning the number, location, and divergence times of purported
Pleistocene refugia. Single refugium hypotheses include: a northern refugium in the
Columbia River Valley and a southern refugium in the Klamath-Siskiyou Mtns. Dual
refugia hypotheses include these same refugia but separated at varying times: last glacial
maximum (20,000 years ago), mid-Pleistocene (800,000 years ago), and early Pleistocene
(1.7 mya). Phylogenetic analyses and inferences from nested clade analysis reveal
distinct northern and southern lineages expanding from the Columbia River Valley and
the Klamath-Siskiyou Mtns., respectively. Results of coalescent simulations reject both
single refugium hypotheses and the hypothesis of dual refugia with a separation date in
the late Pleistocene but not hypotheses predicting dual refugia with separation in early or
mid-Pleistocene. Estimates of time since divergence between northern and southern
33
lineages also indicate separation since early to mid-Pleistocene. Tests for expanding
populations using mismatch distributions and ‘g’ distributions reveal demographic
growth in the northern and southern lineages. The combination of these results provides
strong evidence that this species was restricted into, and subsequently expanded from, at
least two Pleistocene refugia in the Pacific Northwest.
Introduction
One of the main objectives of phylogeography is to infer the processes that have
lead to the genetic patterns observed in populations across the landscape (Avise 2000).
The cycles of glacial advance and retreat during the Pleistocene had an undeniable effect
on genetic structuring within species (Hewitt 1996; Ibrahim et al. 1996; Avise et al. 1998)
and among species groups (Brunsfeld et al. 2001; Carstens et al. 2005a). The role of
Pleistocene refugia during glacial advances was especially important in generating and
maintaining genetic diversity. The separation of ancestral populations into isolated
refugia allowed for the formation of distinct evolutionary lineages within species (Hewitt
2000). Identifying the number and location of Pleistocene refugia is important in
determining the patterns of post-glacial expansion from Pleistocene refugia (Hewitt
1999), identifying distinct lineages or populations for conservation or management
purposes (Wagner et al. 2005), and providing insights into the evolutionary history of
ecosystems (Carstens et al. 2005a).
The world’s largest expanse of temperate rainforest occurs within the Pacific
Northwest of North America and provides a prime example of an ecosystem shaped by
Pleistocene glacial processes. Within this ecosystem are a multitude of endemic
34
organisms for which several competing phylogenetic hypotheses exist regarding the
number of Pleistocene refugia in structuring genetic variation of these species (Brunsfeld
et al. 2001). One such endemic of this coniferous rainforest ecosystem is the Pacific
Giant Salamander, Dicamptodon tenebrosus. Its widespread range (Fig. 1) from
southwestern British Columbia to northwestern California makes it an ideal organism for
testing hypotheses on the number and location of Pleistocene refugia in the Pacific
Northwest, as well as investigating post-glacial expansion routes from these refugia.
Previous studies have confirmed the monophyly of D. tenebrosus and revealed some
geographic structure (Daugherty et al. 1983; Good 1989; Steele et al. 2005), but
relationships among theses lineages are not well-resolved. In this study we use a
coalescent modeling approach to test statistically competing phylogeographic hypotheses
concerning the number, location, and divergence time among Pleistocene refugia in the
Pacific Northwest for this species.
Specific hypotheses proposed by Brunsfeld et al. (2001) include the possibility of
single or dual refugia. Location of the purported refugia is uncertain because post-glacial
expansion from refugia resulted in a contiguous distribution across the landscape, thereby
removing any clues as to the location of the refugia. However, genetic patterns revealed
in previous studies suggest at least two refugia. A southern refugium is thought to exist in
the Klamath-Siskiyou Mtns based on a study of six plant species (Soltis et al. 1997).
Evidence also suggests another refugium located farther north. Proposed locations of a
northern refugium have included: the Olympic Peninsula, Vancouver Island, and Haida
Gwaii (Queen Charlotte Islands) (Soltis et al. 1997; Byun et al. 1999; Demboski et al.
1999). All of these localities are unlikely northern refugia for D. tenebrosus because
35
these areas are well outside its known distribution. However, another purported northern
refugium is the Columbia River Valley. Genetic studies conducted in a variety of fish
species have identified the lower Columbia River and its tributaries as a probable
refugium (Brown et al. 1992; Bickham et al. 1995; Taylor et al. 1999; McCusker et al.
2000; Haas and McPhail 2001). Considering that D. tenebrosus is a stream-breeding
salamander and that terrestrial adults are closely associated with streams, we propose the
Columbia River Valley to also be a plausible refugium for this species.
In this study, we test five hypotheses concerning the number, location, and
divergence times of Pleistocene refugia for D. tenebrosus. Our hypotheses (Fig. 2)
include: 1) a single northern refugium in the Columbia River Valley; 2) a single southern
refugium in the Klamath-Siskiyou Mtns; 3) two refugia, one in the Columbia River
Valley and the other in the Klamath-Siskiyou Mtns, separated at last glacial maximum
(20,000 years ago); 4) these same two refugia but separated since the mid-Pleistocene
(800,000 years ago); and, 5) the two refugia separated since the early Pleistocene (1.7
million years). By constructing evolutionary models based on these hypotheses and then
coalescing simulated data under these models, we can then determine the probability that
the observed data are generated by these evolutionary scenarios.
Material and methods
Sample collection and DNA amplification
We obtained tissue samples of 82 individuals from 31 localities throughout the
range of D. tenebrosus (Fig. 1), including localities in purported refugia of the Columbia
River Valley and the Klamath-Siskiyou Mtns. Samples were obtained primarily from the
36
Museum of Vertebrate Zoology at Berkeley but were supplemented with field-collected
tissues. Sequences of the three remaining members of the genus (D. aterrimus, D.
ensatus, and D. copei) were used as outgroups (Steele et al. 2005).
DNA was extracted using standard phenol/chloroform extractions (Sambrook et
al. 1989). Thirty-nine sequences for a ~1100 bp section of the cytochrome b gene (cyt b)
were obtained from an earlier study (Steele et al. 2005) and are deposited in GenBank.
Amplification of the same cyt b region from an additional 43 samples was performed
using the two primer sets in Carstens et al. (2005b): tRNA-Threonine (5´-
TTCAGCTTACAAGGCTGATGTTTT-3´) with a reverse internal (5´-
TAATTAGTGGATTTGCTGGTGTAA-3´) and tRNA-Glucine (5´-
TTGTATTCAACTATAAAAAC-3´) with a forward internal (5´-
TCCACCCATACTTTTCTTATAAAGA-3´). We also amplified a ~750-bp portion of
the mitochondrial control region (CR) for all 82 samples using a modified 007 primer (5´-
GCACCCAAAGCCAAAATTTTCA-3´) and the 651 primer (5´-
GTAAGATTAGGACCAAATCT-3´) (Shaffer and McKnight 1996). Amplicons were
purified using centrifugal filters (Millipore; Bedford MA) and sequencing reactions were
performed using BigDye Kit version 3.1 (Applied Biosystems; Foster City, CA) with 20-
40 ng of PCR product in 10 ul reaction volumes. Sequencing reactions for cyt b and CR
were performed in both 5´ and 3´ directions, purified with a 70% isopropyl wash, and run
on either an ABI 377 or ABI 3730 automated sequencer. Sequences were aligned and
edited with SEQUENCHER 4.1 (Gene Codes; Ann Arbor, MI). Sequences are deposited in
Genbank (Appendix 1).
37
Phylogeny reconstruction
We analyzed the data using maximum parsimony (MP), maximum likelihood
(ML), and Bayesian analyses. Redundant haplotypes were removed and we used DT-
MODSEL (Minin et al. 2003) to select a model of evolution. Maximum parsimony and
ML analyses were performed with PAUP* 4.0.b10 (Swofford 2002) using a heuristic
search with TBR and 10 random-addition replicates. For the MP analysis, we weighted
all sites equally and treated gaps as missing data. The HKY+I+G model of evolution
were the best fit for the cyt b and CR data as well as the combined data. The ML analysis
was performed under this model where I = 0.752, G = 0.735, a transition/transversion
ratio of 3.22, and the following equilibrium base frequencies: A = 0.3226, C = 0.1768 G
= 0.1537, T = 0.3469. Data sets were tested for congruence with a partition homogeneity
test in PAUP and resulted in a nonsignificant value of P = 0.01. Critical values for this
test are thought to be between 0.01 and 0.001 (Cunningham 1997). Branch support of MP
and ML analyses were assessed from 200 non-parametric bootstrap replicates.
To estimate Bayesian posterior probabilities of nodes, we used MR.BAYES 3.1
(Huelsenbeck and Ronquist 2001; Ronquist and Huelsenbeck 2003) to conduct 4.5x10 6
generations of a Bayesian run under an HKY+I+G model with default flat priors and
sampling every 100th generation. DT-MODSEL selected the HKY+I+G model of evolution
for both the cyt b and CR data sets but differed in values for transition/transversion ratio,
proportion of variable sites, and among site rate heterogeneity for each gene. We
therefore partitioned the cyt b and CR sequences and unlinked the data sets, thereby
allowing these parameters to vary across the two data sets during analysis. Two
independent runs were performed simultaneously on the data with each run using one
38
cold and three heated chains. Examination of the posterior probability distributions
suggested the Markov chain reached stationarity within 100,000 generations but we
discarded the first 25% of the samples (1,125,000 generations) as ‘burn in’ to ensure
stationarity. The average standard deviation of split frequencies between the two
independent runs at completion was 0.0038 and suggested convergence of the two runs
on a stationary distribution.
Testing Pleistocene hypotheses
Our approach to hypothesis testing is similar to that of Carstens et al. (2005b) in
their testing of competing Pleistocene hypotheses for the Idaho giant salamander (D.
aterrimus). We used MESQUITE 1.05 (Maddison and Maddison 2004) to conduct
coalescent simulations of the combined data set within each of the five hypotheses of
Pleistocene refugia (Fig. 2). We simulated 1000 coalescent genealogies within the
predicted population history of each hypothesis and then simulated DNA, using model
parameters determined from ingroup sequences only to better reflect intraspecific
evolution, on each of the simulated gene trees.
Performing the simulations required an estimate of effective population size (Ne).
To estimate Ne we used a coalescent approach implemented in MIGRATE 2.0.6 (Beerli
2004) to calculate the parameter θ where θ =2Neµ. A mutation rate (µ) for the cyt b
region was estimated as µ = 1.6x10-8 based on degree of divergence of this gene region
between the Idaho Giant Salamander (D. aterrimus) and the Pacific Giant Salamander (D.
tenebrosus) and calibrated to the orogeny of the Cascade Mtns that lead to the separation
of the species (Steele et al. 2005). We therefore used cyt b data and its associated
39
mutation rate to estimate an Ne for the entire species and for populations in the purported
refugia of the Columbia River Valley and the Klamath-Siskiyou Mtns. Simulations under
a single refugium hypothesis used an Ne equivalent to the proportion of the Ne estimated
from populations located in the purported refugium to that of the Ne estimated for all
populations sampled. The three hypotheses with two Pleistocene refugia (see Fig. 2)
differ in time since divergence between refugia. Simulations were conducted using three
divergence dates that correspond to a split early in the Pleistocene (1.7 mya), a mid-
Pleistocene split (800,000 years ago), and a recent Pleistocene split (20,000 years ago)
corresponding to the last glacial maxima (Wait and Thorson 1983). A generation length
of 4 years (Nussbaum et al. 1983) was used to convert divergence times in years to
coalescent times in generations.
Slatkin and Maddison’s (1989) S statistic, which measures the discord between a
gene tree and subdivision of populations, was used to assess significance of each
hypothesis. This statistic treats the defined populations as categorical variables and is a
measure of the minimum number of migration events (i.e. sorting events) between
populations as implied by the gene tree. Coalescent simulations of the gene tree within
the population tree provided by each hypothesis produced a distribution of expected
values of the S statistic under the proposed degree of population subdivision and
divergence time. Values of the S statistic calculated from the observed gene tree were
compared to this distribution in order to determine the significance of discord between
simulated gene trees and the population divisions presented in each hypothesis.
Nested clade and population level analyses
40
To test for significant association of haplotypes with geography we performed a
Nested Clade Analysis (NCA) (Templeton et al. 1987, 1995). NCA analyses are a
common tool in phylogenetic studies and are often useful in inferring historical
phylogeographic processes. NCA has been criticized for lacking statistical assessment
among alternative phylogeographic inferences (Knowles and Maddison 2002) but
Templeton (2004) maintains that a slightly revised inference key reduces error and
provides an accurate assessment of phylogeographic processes, especially when specific
a priori scenarios are unknown. We employ the NCA as an opportunity to reinforce
results obtained from coalescent simulations. Congruent results among coalescent
simulations and NCA increases confidence in the accuracy of inferences made about past
phylogenetic processes.
A minimum spanning network was constructed using TCS 1.18 (Clement et al.
2000) and haplotypes were nested using rules of Templeton et al. (1987) and Templeton
and Sing (1993). Geographical localities for each population were calculated using
latitude and longitude. GEODIS 2.2 (Posada et al. 2000) was used to test for significant
association of haplotypes and geography. We followed the inference key in Templeton
(2004) for clades with significant geographical associations.
Genetic diversity of haplotypes was explored using AMOVA performed in
ARLEQUIN 2.0 (Excoffier et al. 1992; Schneider et al. 2000). We partitioned samples into
population groupings based on results of the previous analyses and considered two
alternative groupings: one corresponding to the major lineages identified by the
phylogeny and the other corresponding to clades identified by NCA.
41
After major clades of the phylogeny were identified, divergence times between
the clades were estimated using MDIV (Nielsen and Wakeley 2001). Estimating
divergence time requires a mutation rate and we again used µ = 1.6x10-8 from cyt b data
to estimate time since divergence using sequences from this gene. Initial analysis
indicated migration among major clades was nearly zero; thus, we reanalyzed the data
with the migration prior set to M = 0 and the max T = 1. We conducted 2x106 generations
of the Markov chain and repeated the analysis several times to ensure stationary. Time
since divergence was estimated as tdiv=T(θ)/2µ and a 95% confidence interval was
calculated from the distribution of posterior probabilities of θ.
Analysis of demographic history
We used FLUCTUATE (Kuhner et al. 1998) to estimate exponential growth rate (g)
of a population to test for demographic growth in clades indicated by the NCA as having
undergone range expansion. We used 10 short chains of 1,000 generations and 10 long
chains of 20,000 generations with an initial ‘g’ value of 0. Each run started with
Watterson's estimate of θ (Watterson 1975), empirical nucleotide frequencies, and with a
transition/transversion ratio (3.0292) and proportion of invariable sites (0.8855)
determined from DT-MODSEL using only ingroup sequences. The program was run
several times to ensure consistent estimation of ‘g’. Results of the ‘g’ distribution can be
biased upward (Kuhner et al. 1998); thus, to determine significant deviation from a
constant population size (g = 0) we used a conservative 99% confidence interval (±3SD
around the mean) to infer population growth.
42
Evidence for population expansion was also tested under the expansion model of
Rogers and Harpending (1992) by examining pairwise mismatch distributions.
Populations that have had constant size are thought to be multimodal in the pairwise
mismatch distribution, while populations that have undergone recent demographic
expansion are unimodal. Mismatch distributions were calculated in ARLEQUIN (Excoffier
et al. 1992; Schneider et al. 2000) for samples contained in each of the main lineages or
clades identified in the phylogeny and the NCA. Harpending’s (1994) raggedness index
was used to evaluate deviation from the null expectation of no population expansion.
Results
Summary of samples
We sequenced 1847 nucleotides of mitochondrial DNA; 1093 bases of partial cyt
b sequence and 754 bases of partial CR sequence. We found 35 distinct haplotypes from
82 individuals; 9 haplotypes were found in multiple individuals and 26 haplotypes were
represented by single individuals. The most frequently sampled haplotype, designated as
‘A’ in the phylogeny (Fig. 3), was found in 27 of the 82 individuals (33.3%) and was
present in 11 of the 31 localities. All localities that contained this widespread haplotype
are located either within the Columbia River Valley or north of the valley into
Washington State.
Phylogenetic analyses
There is clear separation of populations into two main lineages corresponding to
northern and southern localities. This topological pattern is consistent across MP, ML,
43
and Bayesian estimations of the phylogeny and is well supported by MP and ML
bootstrap support as well as Bayesian posterior probabilities (Fig. 3). There were 98
parsimony-informative sites in the complete data set and the MP analysis found 38
equally parsimonious MP trees with a tree length of 353 steps. Topology of MP trees was
similar to that of the single best ML tree (-ln 4620.7173). The Bayesian topology (Fig. 3)
was the most resolved and, except for a minor difference of relationships at the tips
within the southern clade, has an identical topology to that of MP and ML analyses.
Within the northern clade, there are two well-supported sister clades. One clade
includes an isolated population at Oak Springs, OR and the other includes localities in the
Columbia River Valley and throughout Washington State (Fig 3.). The southern clade,
which contains the remainder of localities in Oregon and California, is also split into two
weakly-supported sister clades. One lineage corresponds to coastal localities extending
from the Klamath-Siskiyou Mtns along the Oregon Coast range to the mouth of the
Columbia River and includes localities in the Cascade Mtns of Oregon. The other lineage
corresponds to localities extending southward from the Klamath-Siskiyou Mtns into
California. The overall phylogenetic pattern of two well-supported clades corresponding
to northern and southern localities is suggestive of two Pleistocene refugia for this
species.
Pleistocene hypotheses
Coalescent simulations conducted in MESQUITE were run using the estimates of
Ne calculated from the population parameter θ for the entire population and for
populations occurring in each of the purported refugia. Using MIGRATE we calculated: θ
44
Total = 0.01926, Ne Total = 601,875; θ Columbia = 0.00101, Ne Columbia = 31,563; and θ Klam-Sisk
= 0.00887, Ne Klam-Sisk = 277,188. Slatkin and Madison’s S was calculated in MESQUITE as
S = 1 for the observed data. The model of evolution for ingroup sequences used in the
coalescent simulations was: HKY+I+G, I = 0.8855, G = 0.7502, transition/transversion =
3.0292, A = 0.3201 C = 0.1739 G = 0.1564, T = 0.3492.
Results of coalescent simulations indicate that we could reject the hypothesis of a
single refugium located in the Columbia River Valley (P<0.0001) or the Klamath-
Siskiyou Mtns (P<0.0001). We could also reject the hypothesis of dual refugia with a
separation date in the late Pleistocene (P<0.0001) but not hypotheses predicting dual
refugia with a separation in the early Pleistocene (P>0.99) or mid-Pleistocene (P>0.99).
Nested clade analysis
The haplotype network consisted of two main networks that could only be joined
with a non-parsimonious connection of 25 steps (Fig. 4). These two groups corresponded
to the northern and southern lineages identified in the phylogeny. Some ambiguous
connections caused by loops are present in the network but were resolved using nesting
procedures from Templeton et al. (1992) and Templeton and Sing (1993) and ultimately
do not affect nesting design or conclusions inferred from the analysis. The overall pattern
of the NCA is consistent with phylogenetic results and coalescent hypotheses tests,
suggesting two Pleistocene refugia for this species.
Five nested haplotype networks had significant association with geography and
inferences for these groups are given in Table 2. Significant associations with geography
within the northern lineage include the nested clade 1-2 which was identified as isolation-
45
by-distance (Fig. 5). The entire northern network (clade 4-1) had significant geographic
association defined as allopatric fragmentation of the isolated population at Oak Springs,
OR. In the southern clade there were three clades with significant association with
geography. Nested clade 4-5, which encompasses localities in the Klamath-Siskiyou
Mtns, is defined as isolation-by-distance. Clade 5-1, which corresponds to coastal
localities from the Klamath-Siskiyou Mtns to the mouth of the Columbia, is defined as
range expansion. Clade 5-2 includes the nested clade 4-5 and additional populations in
the Cascade Mtns of Oregon and is also defined as range expansion.
Diversity, divergence, and demographic growth
AMOVA revealed 68.4% of the variation is explained by the north-south split.
The average nucleotide diversity estimated for the southern lineage (π = 0.009) is four
and a half times greater than the northern lineage (π = 0.002). Southern populations also
had a higher number of haplotypes (25) and polymorphic sites (S = 80) than northern
population (haplotypes = 10, S = 27). Genetic distances within and between the two
lineages is shown in Table 3.
The cyt b sequence data for the northern and southern lineages identified in the
phylogeny and NCA were analyzed in MDIV to estimate time since divergence of these
populations. Results estimated θ = 10.33 and T = 0.002. Using a mutation rte of µ =
1.6x10-8, an estimate of time since divergence between the northern and southern clades
was placed during early to mid-Pleistocene at 645,625 years ago with a 95% confidence
interval of 971,875 to 319,373 years ago.
46
We analyzed all sequence data from each of the significant clades identified by
the NCA for population growth in FLUCTUATE. Demographic expansion of the entire
northern clade (clade 4-1) was not significant when the isolated population at Oak
Springs (locality 16) was included (g = 224.82 ± 144.1), but was significant when this
allopatric population was removed (clade 3-1; g = 10,000 ± 3045.67). There was also
evidence of significant demographic expansion in each southern clade showing range
expansion in the NCA: clade 5-1 (g = 871.30 ± 126.86); clade 5-2 (g = 755.54 ± 81.48).
Results of mismatch distributions were also consistent with a pattern of demographic
growth. The model of population expansion could not be rejected for southern clades 5-1
(P = 0.53), 5-2 (P = 0.72), nor for the northern clade (P = 0.79).
Discussion
The role of Pleistocene glaciation in structuring contemporary genetic variation
has been an active area of research and phylogenetic patterns are often interpreted in the
context of postglacial expansion from glacial refugia (Hewitt 1996; Ibrahim et al. 1996;
Avise et al. 1998). Dicamptodon tenebrosus has been extensively studied using
morphological (Nussbaum 1976) and electrophoretic methods (Daugherty et al. 1983;
Good 1989) but our results are novel in the detection of two distinct lineages
corresponding to northern and southern populations. Perhaps the most intriguing
implication is the identification of the Columbia River Valley as a Pleistocene refugium
for the northern populations. The Columbia River Valley is often implicated as a
refugium for fishes (Brown et al. 1992, Taylor et al. 1999), but this has not been the case
for terrestrial taxa.
47
Postglacial expansion
Results of this study show two well supported lineages corresponding to northern
and southern populations. Estimates of divergence between these lineages indicate
separation since the early to mid Pleistocene. The localities of northern populations
encompass the purported Pleistocene refugium of the Columbia River Valley, while
apparent northward expansion of southern populations supports another purported
refugium in the Klamath-Siskiyou Mtns. Coalescent simulations also support the
hypothesis of two Pleistocene refugia for this species. The combination of these results
provides strong evidence that this species was restricted into at least two Pleistocene
refugia in the Pacific Northwest.
Isolation-by-distance of the northern populations suggests a slow and gradual
northward expansion from the Columbia River Valley to the southern banks of the Fraser
River in British Columbia which forms the northernmost boundary of the species and
apparently limits further expansion. The isolated population at Oak Springs, OR, which is
included within the northern lineage, was probably connected by suitable habitat to
populations in the Columbia River Valley during the initial north-south split and became
isolated only relatively recently. Populations in the southern refugium of the Klamath-
Siskiyou Mtns expanded northward along either side of the Willamette Valley of Oregon.
One route was along the coastal mountain ranges of Oregon to the mouth of the
Columbia River while the other was an inland route along the Oregon Cascades. The
Columbia River and its gorge appear to be effectively preventing migration and mixing
between the two lineages since no southern haplotypes were found north of the Columbia
48
River and no northern haplotypes (excluding Oak Springs) were found south of the
Columbia River Valley. The southern limit of the species is defined by a narrow zone of
secondary contact in northern California with the California Giant Salamander (D.
ensatus) (Good 1989). Populations in the Klamath-Siskiyou refugium may have
gradually expanded southward into northern California to form this contact zone with
northward expanding D. ensatus.
Pleistocene refugia
Traditionally, the Columbia River Valley has not been considered a Pleistocene
refugium for terrestrial organisms. It has only recently been identified as a Pleistocene
refugium for the Larch Mountain Salamander (Plethodon larselli), which has expanded
northward along the Cascade Mtns of Washington State (Wagner et al. 2005).
Additionally, the northern populations of the Oregon Slender Salamander (Batrachoseps
wrighti) that divergent in mitochondrial DNA are closely associated with the Columbia
River (Miller et al. 2005). However, these are examples of species with restricted
distributions adjacent to the Columbia River Valley. When north-south splits in genetic
data attributable to Pleistocene glaciation are discovered in a widely distributed species,
the typical refugia proposed for northern populations include the Olympic Peninsula, the
Queen Charlotte Islands, or southeast Alaska (Soltis et al. 1997; Conroy and Cook 2000;
Janzen et al. 2002). However, it is unlikely that Pacific Giant Salamanders resided in any
one of these refugia, because their current distribution is neither in nor near these
locations. The Columbia River and its tributaries are more often regarded as a Pleistocene
refugium for fish species (Brown et al. 1992; Bickham et al. 1995; Taylor et al. 1999;
49
McCusker et al. 2000; Haas and McPhail 2001). While some fish may prey upon larval
salamanders, presumably some of these tributaries in the Columbia River Valley were
fishless and would have provided suitable habitat for stream-breeding salamander larvae.
In contrast to the Columbia River Valley, the Klamath-Siskiyou Mtns have been
proposed or implicated as a Pleistocene refugium for a variety of organisms (Soltis et al.
1997; Wake 1997; Brunsfeld et al. 2001; Wilke and Duncan 2004; Kuchta and Tan
2005). The area remained unglaciated throughout the Pleistocene and is known for its
complex geology and a range of climates which have contributed to the region’s
biological diversity and endemism (Whitaker 1960; Noss et al. 1999). The restricted
distribution of the Del Norte Salamander (Plethodon elongatus), Siskiyou Salamander (P.
stormi) and the recently discovered Scott Bar salamander (P. asupak) (Mead et al. 2005)
attest to the diversity and endemism of the region.
The detection of two well defined lineages corresponding to northern and
southern populations has also been documented in other co-distributed taxa (Soltis et al.
1997; Kuchta and Tan 2005). In these cases, the highest genetic diversity was within the
southern populations and lowest genetic diversity in the northern populations. This
pattern could result from northward expansion of populations; however, the two highly
divergent clades within D. tenebrosus and other organism (Soltis et al. 1997; Kuchta and
Tan 2005) suggest separation and isolation into two Pleistocene refugia. The high genetic
diversity of southern populations of D. tenebrosus encompasses the Klamath-Siskiyou
Mtns indicates that the southern refugium was larger than the northern refugium or had a
larger ancestral population.
50
Regional phylogeography
The patterns observed from this research add to our understanding of the role of
Pleistocene refugia in regional phylogeography of the Pacific Northwest. The results
provide additional evidence of a Pleistocene refugium in the Klamath-Siskiyou Mtns and
further support the importance of a Columbia River Valley refugium for terrestrial taxa.
Genetic structure of several taxa with distributions similar to that of the Pacific Giant
Salamander have been examined but do not always show a distinct north-south split
corresponding to separation and isolation in two Pleistocene refugia. Phylogenetic
patterns in a mollusk (Wilke and Duncan 2004) and the Ensatina salamander (Wake
1997) suggest expansion from one or more southern refugia in the Klamath-Siskiyou
Mtns, while patterns in a salamander (Wagner et al. 2005) and several fish species
(Taylor et al. 1999; McCusker et al. 2000) provide examples of expansion from a
northern refugium in Columbia River Valley. Other taxa such as garter snakes (Janzen et
al. 2002), newts (Kuchta and Tan 2005), and a variety of plant species (Soltis et al. 1997)
have a more defined north-south split fitting the hypothesis of northern and southern
refugia in the Pacific Northwest proposed by Brunsfeld et al. (2001).
Various studies have examined the explicit hypotheses proposed by Brunsfeld et
al. (2001), which invoke either ancient vicariance or recent dispersal, for explaining
disjunct distributions of mesic forest taxa located in the coastal Pacific Northwest and the
inland northern Rocky Mountains (Nielson et al. 2001; Carstens et al. 2004; Steele et al.
2005; Carstens et al. 2005b). Results of these studies indicate that co-distributed
amphibians have a similar pattern of deep divergence between coastal and inland
populations consistent with the ancient vicariance hypothesis (Carstens et al. 2005a).
51
Within the northern Rocky Mountains, amphibians also have concordant phylogenetic
patterns indicating similar response to Pleistocene glaciation (Carstens et al. 2004;
Carstens et al. 2005b). The phylogenetic concordance across amphibians in this region
suggests that these organisms responded similarly to geological events. Opportunities
exist to investigate whether the patterns of concordance in amphibian phylogenies is also
apparent in coastal populations of the Pacific Northwest. Within this region are
assemblages of co-distributed species including the northwestern salamander
(Ambystoma gracile), the western red-backed salamander (Plethodon vehiculum), and the
tailed frog (Ascaphus truei). Examining genetic structure across a variety of organisms
and testing for concordant phylogenies will undoubtedly provide insights into the
evolution of communities within the Pacific Northwest and in general.
Conclusions
Pleistocene glaciation has often influenced the geographic structure of species and
studies of taxa within the Pacific Northwest regularly reveal distinct lineages often
attributed to isolation within northern and southern Pleistocene refugia. The location of a
southern refugium in the Klamath-Siskiyou Mtns is generally accepted but a variety of
locations for northern refugia exist. Results of this study indicate the Columbia River
Valley as a refugium from which northern populations of the Pacific Giant Salamander
(D. tenebrosus) expanded. This refugium has been generally established in phylogenetic
studies of fishes in the Pacific Northwest but is not well recognized as a potential
refugium for terrestrial taxa. It still remains to be seen whether co-distributed taxa also
have genetic patterns suggestive of dispersal from a Columbia River refugium.
52
References
Avise JC (2000) Phylogeography: the history and formation of species. Harvard
University Press, Cambridge, MA.
Avise JC, Walker D, Johns GC (1998) Speciation durations and Pleistocene effects on
vertebrate phylogeography. Proceedings of the Royal Society of London B, 265,
1707–1712.
Beerli P (2004) MIGRATE: Documentation and program, part of LAMARC. Version
2.0.6: Available at http://evolution.genetics.washington.edu/lamarc.html.
Bickham JW, Wood CC, Patton JC (1995) Biogeographic implications of cytochrome b
sequences and allozymes in sockeye (Oncorhynchus nerka). Journal of Heredity,
86, 140–144.
Brown JR, Beckenbach AT, Smith MJ (1992) Influence of Pleistocene glaciations and
human interaction upon mitochondrial DNA diversity in white sturgeon
(Acipenser transmontanus) populations. Canadian Journal of Fisheries and
Aquatic Sciences, 49, 358–367.
Brunsfeld SJ, Sullivan J, Soltis DE, and Soltis PS (2001) Comparative phylogeography of
northwestern North America: a synthesis. In: Integrating Ecology and Evolution
in a Spatial Context (eds. Silvertown J, Antonovics J), pp. 319–339. Blackwell
Publishing, Williston, VT.
Byun SA, Koop BF, Reimchen TE (1997) North American black bear mtDNA
phylogeography: implications for morphology and the Haida Gwaii glacial
refugium controversy. Evolution, 51, 1647–1653.
Carstens BC, Stevenson AL, Degenhardt JD, Sullivan J (2004) Testing nested
53
phylogenetic and phylogeographic hypotheses in the Plethodon vandykei species
group. Systematic Biology, 53, 781–792.
Carstens BC, Brunsfeld SJ, Demboski JR, Good JD, Sullivan J (2005a) Investigating the
evolutionary history of the Pacific Northwest mesic forest ecosystem: hypothesis
testing within a comparative phylogeographic framework. Evolution, 59, 1639–
1652.
Carstens BC, Degenhardt JD, Stevenson AL, Sullivan J (2005b) Accounting for
coalescent stochasticity in testing phylogeographic hypotheses: modeling
Pleistocene population structure in the Idaho Giant Salamander Dicamptodon
aterrimus. Molecular Ecology, 14, 255–265.
Clement M, Posada D, Crandall K (2000) TCS: a computer program to estimate gene
genealogies. Molecular Ecology, 9, 1657–1660.
Conroy CJ, Cook JA (2000) Phylogeography of a post-glacial colonizer: Microtus
Longicaudus (Rodentia: Muridae). Molecular Ecology, 9, 165–175.
Daugherty CH, Allendorf FW, Dunlap WW, Knudsen KL (1983) Systematic implications
of geographic patterns of genetic variation in the Genus Dicamptodon. Copeia,
1983, 679–691.
Demboski JR, Stone ED, Cook JA (1999) Further perspectives on the Haida Gwaii
glacial refugium. Evolution, 53, 2008–2012.
Excoffier L, Smouse P, Quattro J (1992) Analysis of molecular variance inferred from
metric distances among DNA haplotypes: Application to human mitochondrial
DNA restriction data. Genetics, 136, 343–359.
Good DA (1989) Hybridization and cryptic species in Dicamptodon (Caudata:
54
Dicamptodontidae). Evolution, 43, 728–744.
Harpending, RC (1994) Signature of ancient population growth in a low-resolution
mitochondrial DNA mismatch distribution. Human Biology, 66, 591–600.
Haas GR, McPhail JD (2001) The post-Wisconsinan glacial biogeography of bull trout
(Salvelinus confluentus): a multivariate morphometric approach for conservation
biology and management. Canadian Journal of Fisheries and Aquatic Sciences,
58, 2189–2203.
Hewitt GM (1996) Some genetic consequences of ice ages, and their role in divergence
and speciation. Biological Journal of the Linnean Society, 58, 247–276.
Hewitt GM (1999) Post-glacial re-colonization of European biota. Biological Journal of
the Linnean Society, 68, 87–112.
Hewitt GM (2000) The genetic legacy of the Quaternary ice ages. Nature, 405, 907–913.
Huelsenbeck JP, Ronquist F (2001) MRBAYES: Bayesian inference of phylogenetic
trees. Bioinformatics, 17, 754–755.
Ibrahim KM, Nichols RA, Hewitt GM (1996) Spatial patterns of genetic variation
generated by different forms of dispersal during range expansion. Heredity, 77,
282–291.
Janzen FJ, Krenz JG, Haselkorn TS, Brodie, Jr ED, Brodie, III ED (2002) Molecular
phylogeography of common garter snakes (Thamnophis sirtalis) in western North
America: implications for regional historical forces. Molecular Ecology, 11,
1739–1751.
Knowles LL, Maddison WP (2002) Statistical phylogeography. Molecular Ecology, 11,
2623–2635.
55
Kuchta SR, Tan AM (2005) Isolation by distance and post-glacial range expansion in the
rough skinned newt, Taricha granulosa. Molecular Ecology, 14, 225–244.
Kuhner M, Yamato J, Felsenstein J (1998) Maximum-likelihood estimation of population
growth rates based on the coalescent. Genetics, 149, 429–434.
Maddison WP, Maddison DR (2004) Mesquite: a modular system for evolutionary
analysis. Version 1.05 http://mesquiteproject.org.
McCusker MR, Parkinson E, Taylor EB (2000) Mitochondrial DNA variation in rainbow
trout (Oncorhynchus mykiss) across its native range: testing biogeographical
hypotheses and their relevance to conservation. Molecular Ecology, 9, 2089–
2108.
Mead LS, Clayton DR, Nauman RS, Olsen DH, Pfrender ME (2005) Newly discovered
populations of salamanders from Siskiyou County, California, represent a species
distinct from Plethodon stormi. Herpetologica, 61, 158–177.
Miller MP, Haig SM, Wagner RS (2005) Conflicting patterns of genetic structure
produced by nuclear and mitochondrial markers in the Oregon slender salamander
(Batrachoseps wrighti): Implications for conservation efforts and species
management. Conservation Genetics, 6, 275–287.
Minin V, Abdo Z, Joyce P, Sullivan J (2003) Performance-based selection of likelihood
models for phylogeny estimation. Systematic Biology, 52, 674–683.
56
Nielsen R, Wakeley J (2001) Distinguishing migration from isolation: a Markov chain
Monte Carlo approach. Genetics, 158, 885–896.
Nielson M, Lohman K, and Sullivan J (2001). Phylogeography of the tailed frog
(Ascaphus truei): implications for the biogeography of the Pacific Northwest.
Evolution, 55, 147–160.
Noss RF, Strittholt JR, Vance-Borland K, Carroll C, Frost P (1999) A conservation plan
for the Klamath-Siskiyou ecoregion. Natural Areas Journal, 19, 392–411.
Nussbaum, RA (1976) Geographic variation and systematics of salamanders of the genus
Dicamptodon Strauch (Ambystomatidae). Miscellaneous Publications of the
Museum of Zoology, University of Michigan, 149, 1–94.
Nussbaum RA, Brodie, Jr ED, Storm RM (1983) Amphibians and Reptiles of the Pacific
Northwest. University of Idaho Press, Moscow, ID.
Posada D, Crandall KA, Templeton AR (2000) GeoDis: a program for the nested c
cladistic analysis of the geographical distribution of genetic haplotypes.
Molecular Ecology, 9, 487–488.
Rogers AR, Harpending H (1992) Population growth makes waves in the distribution of
pairwise genetic differences. Molecular Biology Evolution, 9, 552–569.
Ronquist F, Huelsenbeck JP (2003) MrBayes 3: Bayesian phylogenetic inference under
mixed models. Bioinformatics, 19, 1572–1574.
Sambrook J, Fritsch EF, Maniatis T (1989) Molecular Cloning: A Laboratory Manual,
second edn. Cold Spring Harbor Laboratory Press, New York, NY.
57
Schneider S, Roessli D, Excoffier L (2000) Arlequin: A software for population genetics
data analysis. Ver 2.0. Genetics and Biometry Lab, Dept. of Anthropology,
University of Geneva. http://anthro.unige.ch/arlequin.
Shaffer HB, McKnight ML (1996) The polytypic species revisited: genetic differentiation
and molecular phylogenetics of the tiger salamander (Ambystoma tigrinum)
(Amphibia: Caudata) complex. Evolution, 50, 417–433.
Slatkin M, Maddison WP (1989) A cladistic measure of gene flow inferred from
phylogenies of alleles. Genetics, 123, 603–613.
Soltis DE, Gitzendanner MA, Strenge DD, Soltis PS (1997) Chloroplast DNA
intraspecific phylogeography of plants from the Pacific Northwest of North
America. Plant Systematics and Evolution, 206, 353–373.
Steele CA, Carstens BC, Storfer A, Sullivan J (2005) Testing hypotheses of speciation
timing in Dicamptodon copei and Dicamptodon aterrimus (Caudata:
Dicamptodontidae). Molecular Phylogenetics and Evolution, 36, 90–100.
Swofford, DL (2002) PAUP*. Phylogenetic Analysis Using Parsimony (*and Other
Methods). Version 4. Sinauer Associates, Sunderland, MA.
Taylor EB, Pollard S, Louie D (1999) Mitochondrial DNA variation in bull trout
(Salvelinus confluentus) from northwestern North America: implications for
zoogeography and conservation. Molecular Ecology, 8, 1155–1170.
Templeton AR (2004) Statistical phylogeography: methods of evaluating and minimizing
inference errors. Molecular Ecology, 4, 789–809.
Templeton AR, Boerwinkle E, Sing CF (1987) A cladistic analysis of phenotypic
58
associations with haplotypes inferred from restriction endonuclease mapping. I.
Basic theory and analysis of alcohol dehydrogenase activity in Drosophila.
Genetics, 117, 343–351.
Templeton AR, Crandall KA, Sing CF (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.
Templeton AR, Sing CF (1993) A cladistic analysis of phenotypic associations with
haplotypes inferred from restriction endonuclease mapping. IV. Nested analyses
with cladogram uncertainty and recombination. Genetics, 134, 659–669.
Templeton AR, Routman E, Phillips CA (1995) Separating population structure from
population history: a cladistic analysis of the geographical distribution of
mitochondrial DNA haplotypes in the tiger salamander, Ambysoma tigrinum.
Genetics, 140, 767–782.
Templeton AR (2004) Statistical phylogeography: Methods of evaluating and minimizing
inference errors. Molecular Ecology, 4, 789–809.
Wagner RS, Miller MP, Crisafulli CM, Haig SM (2005) Geographic variation, genetic
structure, and conservation unit designation in the Larch Mountain salamander
(Plethodon larselli). Canadian Journal of Zoology, 83, 396–406.
Waitt, Jr RB, Thorson RM (1983) The Cordilleran ice sheet in Washington, Idaho, and
Montana. In: Late Quaternary Environments of the United States, the Late
Pleistocene (eds Wright, Jr HE, Porter SC), pp. 53–70. University of Minnesota
Press, Minneapolis, MN.
Wake D (1997) Incipient species formation in salamanders of the Ensatina complex.
59
Proceedings of the National Academy of Sciences, 94, 7761–7767.
Watterson GA (1975) On the number of segregating sites in genetical models without
recombination. Theoretical Population Biology, 7, 256–276.
Whittaker RH (1960) Vegetation of the Siskiyou Mountains, Oregon and California.
Ecological Monographs, 30, 279–338.
Wilke T, Duncan N (2004) Phylogeographical patterns in the American Pacific
Northwest: lessons from the arionid slug Prophysaon coeruleum. Molecular
Ecology, 13, 2303–2315.
60
Table 1 Locality information, number of captures, and haplotypes sampled for localities of D. tenebrosus. Locality numbers correspond to those in Fig. 1. Unique haplotype sequences were deposited in GenBank for cyt b (DQ387923–DQ387957) and control region (DQ388392–DQ388426).
Locality n Haplotypes Locality information and museum voucher numbers (if applicable).
1 4 A Tributary at Nooksack Falls, Whatcom Co., WA 2 1 G Mallardy Crk, Snohomish Co., WA 3 4 A 11 Mile Crk, Chelen Co., WA 4 2 A Mine Crk, King Co., WA 5 2 A Mosquito Crk, Kittitas Co., WA 6 5 A Tributary of West Fork of Little Nisqually River, Lewis Co., WA 7 1 A West Fork of Elochoman River, Wahkiakum Co., WA 8 2 E, F Coweeman River, Cowlitz, Co., WA 9 3 A, B Yale Crk, Clark Co., WA 10 5 A, C, D Lewis Crk, Skamania Co., WA 11 1 A McCloskey Crk, Skamania Co., WA 12 3 A Holmes Crk, Skamania Co., WA 13 7 N Saddle Mtn. Crk, Clatsop Co., OR 14 3 N S. Fk. Quartz Crk, Columbia Crk, OR 15 2 A Oneonta Gorge, Multonomah Co., OR; MVZ 187949, 187951 16 6 H, I, J Oak Springs, Wasco Co., OR; MVZ 187944–45 17 3 N, O Kilchis River Park, Tillamook Co., OR; MVZ 192583, 192589–
192590 18 3 K, L Fall Crk, Benton Co., OR; MVZ 187959–61 19 2 T, U Lookout Crk, Lane Co., OR; MVZ 223245–46 20 1 M Smith River Falls, Douglas Co., OR; MVZ 187958 21 2 S, T N. Fk. Willamette River, Lane Co., OR; MVZ 187954–55 22 3 ab, ac Thompson Crk, Josephine Co., OR; MVZ 192606–08 23 2 X Shoat Springs, Jackson Co., OR 24 3 P, Q, R Rowdy Crk, Del Norte Co., CA; MVZ 192601–03 25 4 W, Y, Z,
aa Wingate Crk, Siskiyou Co., CA, MVZ 187933–34; O'Neill Crk, Siskiyou Co., CA, MVZ 187939–40
26 2 V Price Crk, Trinity Co., CA; MVZ 187929, 187931 27 1 ag 2 mi E of Delta, junction Hwy. 5, on Delta Rd.; Shasta Co.,
California, MVZ 192613 28 1 ah Signal Port Crk, Mendocimo Co., CA, MVZ 203397 29 1 af Drive-Thru-Tree at Leggett, Mendocimo Co., CA, MVZ 187978 30 1 ai Hwy 1 between Fort Bragg and Rockport, Mendocimo Co., CA,
MVZ 192579 31 2 ae, ad 1.4 mi S of Little Riv, Mendocimo Co., CA, MVZ 192639–40
61
Table 2 Results of nested clad analysis on haplotypes of the Pacific Giant Salamander (Dicamptodon tenebrosus). Haplotype networks without significant geographical associations are not listed. Clade χ2 statistic Probability Inference chain Inferred pattern*
1-2 94.19 0.003 1-2-3-4-NO RGF with IBD throughout western WA
4-1 39.00 0.000 1-19-NO Allopatric fragmentation between Oak Springs, OR and western WA
4-5 23.33 0.043 1-2-3-4-NO RGF with IBD from northwest CA to southwest OR
5-1 18.00 0.000 1-2-11-YES Range expansion from northwest CA to northwest OR
5-2 44.00 0.000 1-2-3-5-6-13-YES Range expansion from northwest CA to central OR
*RGF = restricted gene flow; IBD = isolation by distance
62
Table 3 Genetic distances within and between northern and southern lineages. Corrected genetic distances are with the HKY+I+G model of sequence evolution. Populations Uncorrected Corrected Within northern 0.00499 0.00592 Within southern 0.00957 0.01188 Between clades 0.01953 0.03163
63
1
2 3
4 5 6
7
9 10
11 12 13
14 15 17
16 18
20 19
21
26 27
29 CA
8
28
30 31
OR
WA
2322
Columbia River
25 24
Fig. 1 Shaded area indicates distribution of D. tenebrosus. Numbers indicate sampled localities and correspond to those in Table 1. The location of the Columbia River refugium is indicated by the thickened section of the river inside the species’ range. The Klamath-Siskiyou refugium is indicated by the lightly shaded area.
64All Populations
Northern Populations
Southern Populations
c.)
* a.)
**
b.) All Populations
**
*
Southern Populations
d.)
**
* 800,000 years
Northern Populations
Southern Populations
e.)
**
*
1.7 million years
20,000 years
Northern Populations
Fig. 2 Population trees representing the five Pleistocene hypotheses tested using coalescent modeling: a) single refugium in the Columbia River Gorge b) single refugium in the Klamath-Siskiyou Mtns c) two refugia separated by a divergence time dating to last glacial maxima at 20,000 years ago d) two refugia separated by a divergence time of 800,000 years ago e) two refugia separated by 1.7 million years ago. Constrained effective populations are indicated with asterisks: * = 31,563 (5.24% of the total Ne), ** = 277,188 (46.05% of the total Ne).
65
Sout
hern
Cla
de
Nor
ther
n C
lade
(22)
(31) (31)
(22)
(28) (27)
(30)
(23) (29)
(25) (25) (25)
(25)
(24) (24)
(26)
(24)
(19, 21) (21)
(19)
(17) (18)
(20)
(17)
(16) (16)
(16) (13, 14)
(2)
(9)
(10) (10)
100 57 73
89 50 36
94 58 56
100 63 57
50 67 39
100 61 50
74 46 19
99 46 46
100 94 67
100 100 100
100 82 70
100 93 88
100 96 92
100 96 88
A (1,3,4,5,6,7,9,10,11,12,15) (8) (8)
0.1 substitutions per site
Fig. 3 Results of Bayesian phylogeny from 1847 bp of cytochrome b and control region. Except for being more resolved and a difference in the relationships of several southern taxa, topology is identical to that of MP and ML (HKY+G+I) analyses. Bootstrap values above branches are Bayesian posterior probabilities; below branches are MP and ML bootstrap values (respectively) from 200 replicates. Numbers in parentheses at the tips indicate population localities in which the haplotype occurred.
66
67
ag af ai
Northern Clade C
4-5
A E D
F B G J
1-2 HI
K
L
M
O
P
Q
S
U T
R
V
W aa
ZY
N
ac
ab
ad
ae ah
5-1
4-1
5-2
Southern Clade
X
Fig. 4 Haplotype network for D. tenebrosus. Lines indicate a connection between haplotypes. Missing haplotypes are shown as black dots. Sampled haplotypes are designated with one or two letters and correspond to those in Appendix 1. One-step clades are shown in white, two-step clades in light gray, three-step clades in medium gray, four-step clades in dark gray and five-step clades in black. Clade numbers are shown for clades with significant association with geography. The thick solid line connecting the northern and southern haplotypes indicates a nonparsimonious connection of 25 steps.
WA
CA
7
2 3
4 5
6
9 10
11 12
15 16
8
Allopatric fragmentation (North 4-1)
14
17
18 Range expansion (South 5-1)
CA
31
22
24
23
25
Isolation by distance (South 4-5) 26
27
29 30
28
20 19
21
(South 5-2)
OR
13
Isolation by distance (North 1-2)
1
Fig. 5 Results of Nested Clade Analysis overlaid on a map of sampled populations. The haplotype network comprised a northern clade (populations 1-12, 15-16) and southern clade (13-14, 17-31) which could only be connected with a non-parsimonious connection of 25 steps. Clade identities are indicated in parentheses.
68
CHAPTER THREE
Evidence for phylogeographic incongruence of codistributed species based on
small differences in geographic distribution
Abstract
Codistributed species may display either congruent phylogeographic patterns,
suggesting similar responses to a series of shared climatic and geologic events, or
discordant patterns, indicating independent responses. This study compares the
phylogeographic patterns of two similarly distributed salamander species within the
Pacific Northwest of the United States: Cope’s Giant Salamander (Dicamptodon copei)
and Van Dyke’s Salamander (Plethodon vandykei). Previous studies of P. vandykei
support two reciprocally monophyletic lineages corresponding to coastal populations,
located from the Olympic Mtns to the mouth of the Columbia River, and inland
populations within the Cascade Mtns. We hypothesized that D. copei would have a
congruent phylogeographic pattern due to ecological similarities and similar habitat
requirements to P. vandykei. We test this hypothesis by estimating the phylogeny of D.
copei using ~1800 bp of mitochondrial DNA and comparing it to that of P. vandykei.
Sympatric populations of D. copei display a phylogeographic pattern identical to that of
P. vandykei, suggesting similar responses within their shared distribution. Populations of
D. copei occurring outside the range of P. vandykei displayed high levels of genetic
divergence from those sympatric to P. vandykei. Overall, phylogeographic patterns
between the two species were ultimately incongruent due to the high divergence of these
allopatric populations. These results provide an example of codistributed species
69
displaying overall incongruent phylogeographic patterns while simultaneously displaying
congruent patterns within portions of their shared geographic distribution. This pattern
demonstrates that a simple dichotomy of congruent and incongruent phylogeographic
patterns of codistributed species may be too simplistic and that more complex
intermediate patterns can result even from minor differences in species’ ranges.
Introduction
A central objective of comparative phylogeography is to test codistributed species for
concordant phylogeographic patterns (Schneider et al., 1998; Avise, 2000; Argobast and
Kenagy, 2001; Zink, 2002). Studies that reveal concordance among codistributed biota
often provide evidence that a shared series of past events shaped the genetic diversity of
such organisms similarly. Comparative phylogeography enhances our understanding
about the role of climatic, geological, and ecological forces in shaping the geographic
distribution and intraspecific variation of species comprising an ecosystem. While a
variety of studies have demonstrated phylogeographic congruence among codistributed
taxa (Avise, 1992; Schneider et al.; 1998, Riddle et al.; 2000), a comparable number have
also revealed incongruence (Zink, 1996; Taberlet et al., 1998; Hewitt, 1999). Discovery
of incongruent phylogenies among codistributed species suggests independent responses
to past events due to different ecologies, life histories, or post-glacial expansion routes
(Bowen and Avise, 1990; Taberlet et al., 1998; Michaux et al., 2005; Rocha et al., 2005).
Phylogeographic incongruence among codistributed species suggests that evolution of
biotic communities is often neither a synchronized nor a concerted event (Hewitt, 1999;
Sullivan et al., 2000; Brunsfeld et al., 2001; Carstens et al., 2005b). While congruent and
70
incongruent patterns have been demonstrated among codistributed species,
phylogeographic patterns of codistributed species may be incongruent while
simultaneously displaying significant patterns of shared responses to past climatic or
geologic events (Sullivan et al. 2000). This type of scenario reflects how phylogeographic
patterns result from the combination of a changing environment shared by codistributed
organisms and species-specific responses due to unique ecologies and life history traits.
Within the Pacific Northwest of the United States exists a taxonomically rich
assemblage of organisms endemic to the temperate rainforests of the region (Brunsfeld et
al. 2001). Considerable effort has gone into constructing a regional perspective on the
phylogeographic patterns of these endemic organisms (Brunsfeld et al. 2001, Soltis et al.
1997, Carstens et al. 2005b). The codistributed amphibian assemblage within the mesic
forest ecosystem of the Pacific Northwest provides an ideal opportunity to test for
concerted responses to past climatic and geologic events (Carstens et al., 2005b). This
assemblage includes distantly related amphibian species that share ecological and habitat
requirements and includes diverse species such as tailed frogs (Ascaphus truei, A.
montanus.), Pacific Giant Salamanders (Dicamptodon spp.) and plethodontid
salamanders (Plethodon vandykei, P. idahoensis) (Carstens et al., 2005b). Previous
studies on this assemblage have demonstrated a concordant response to the uplift of the
Cascade Mountains ~2 mya, resulting in reciprocally monophyletic lineages
corresponding to coastal populations and interior populations within the northern Rocky
Mountains (Nielson et al., 2001; Carstens et al., 2004; Steele et al., 2005; Carstens et al.,
2005a). The populations of these amphibians found within the Rocky Mountains share
similar geographic distributions and intraspecific studies reveal a common pattern of
71
shallow phylogenetic structuring in these species, suggesting recent colonization events
(Nielson et al., 2001; Carstens et al., 2004; Carstens et al., 2005a). Coastal lineages of
this same amphibian assemblage also share similar geographic distributions. However,
detailed comparative studies have not yet been conducted on these coastal populations to
test for concerted responses to past climatic or geologic events.
Studies of codistributed amphibian assemblages in the Pacific Northwest have
primarily focused on broad scale phylogeography within a species and the deep genetic
divergence between coastal and inland lineages. In contrast, this study adds a new
dimension by focusing on the comparative phylogeography of species that share small
fragmented distributions restricted to coastal temperate rainforest. Two of the
codistributed amphibians within this mesic forest ecosystem are the Cope’s Giant
Salamander (Dicamptodon copei) and the Van Dyke’s Salamander (Plethodon vandykei).
These species are endemic to the Pacific Northwest of the United States, have similar
habitat requirements, and have similarly fragmented distributions. The geographic
distribution of each species is split into three mountainous regions within the Pacific
Northwest: Olympic Mountains, Willapa Hills, and Cascades Mountains (Fig. 1). The
Cope’s Giant Salamander is a neotenic species and usually remains in an aquatic form
throughout it life (Nussbaum, 1976), while the terrestrial Van Dyke’s Salamander is
strongly associated with moist streamside splash zones (Brodie, 1970). The combination
of a similarly fragmented distribution and shared habitat requirements makes these
organisms ideal for testing hypotheses of concerted or independent responses to past
climatic and geologic events. Because other mesic forest amphibians show similar
responses to past geologic events in the Pacific Northwest, (Carstens et al. 2005b) it is
72
reasonable to predict that these two species should also have concordant phylogeographic
topologies.
Results from previous studies on P. vandykei provide a clear phylogeographic
hypothesis which is used to test the phylogeogrphic topology of D. copei. Both
electrophoretic (Howard et al., 1993) and morphological (Wilson and Larsen, 1999)
studies consistently revealed two reciprocally monophyletic lineages corresponding to
coastal populations, located in the Olympic Peninsula and the Willapa Hills, and inland
populations within the Cascade Mtns (Fig. 1). Populations within these two regions are
thought to have been isolated since the late Pleistocene (Wilson and Larsen, 1999) and
are separated by lowland areas of glacial and alluvial deposits that appear to limit
dispersal (Wilson et al., 1995). Both studies also reveal that populations within the
Olympic Peninsula are indistinguishable from those in the Willapa Hills, indicating
recent expansion of P. vandykei into the Olympic Mtns. To test the hypothesis that D.
copei has a similar phylogeographic distribution, mitochondrial DNA is used to estimate
a phylogeny, intraspecific relationships, and elucidate past demographic patterns within
D. copei. The resulting phylogeny is tested for concordance with that of P. vandykei
using a variety of phylogenetic comparison tests.
Material and methods
Sample collection and DNA amplification
Tissue samples of 80 individuals from 24 localities throughout the range of D. copei
were obtained (Fig. 1). DNA was extracted using standard phenol/chloroform extractions
(Sambrook et al., 1989). Approximately 1100 bp of the cytochrome b gene (cyt b) was
73
obtained using the primer sets from Carstens et al. (2005a). Approximately 750 bp of the
mitochondrial control region (CR) was also amplified using a modified 007 primer (5´—
GCACCCAAAGCCAAAATTTTCA—3´) and the 651 primer from Shaffer and McKnight
(1996). Amplicons were purified using centrifugal filters (Millipore; Bedford, MA) and
sequencing reactions were performed using BigDye Kit version 3.1 (Applied Biosystems;
Foster City, CA) with 20-40 ng of PCR product in 10 ul reaction volumes. Sequencing
reactions for cyt b and CR were performed in both 5´ and 3´ directions, purified with a
70% isopropyl wash, and run on either an ABI 377 or ABI 3730 automated sequencer.
Homologous sequences of the three remaining members of the genus (D. aterrimus, D.
ensatus, and D. copei) were used as outgroups (Steele et al., 2005). Sequences were
aligned and edited with Sequencher 4.1 (Gene Codes; Ann Arbor, MI). Sequences
generated from this study are deposited in Genbank (Appendix 1).
Phylogeny reconstruction
Sequence data were analyzed using maximum parsimony (MP), maximum likelihood
(ML), and Bayesian analyses. Redundant haplotypes were removed and DT-MODSEL
(Minin et al. 2003) was used to select a model of evolution. The CR and cyt b data sets
were tested for congruence with a partition homogeneity test in PAUP* ver.4.0b10
(Swofford, 2002) and resulted in a non-significant value of P = 0.39. The two datasets
were subsequently combined for all analyses. Maximum parsimony and ML analyses
were performed with PAUP* using a heuristic search with TBR and 10 random-addition
replicates. For the MP analysis, all sites were weighted equally and gaps treated as
missing data. The HKY+I+G model of evolution was the best fit for the cyt b and CR
74
data as well as the combined data. The ML analysis was performed under this model
where I = 0.7096, G = 0.8779, transition/transversion ratio = 3.0749, and the following
equilibrium base frequencies: A = 0.3252, C = 0.1761 G = 0.1474, T = 0.3513. Branch
support of MP and ML analyses were assessed from 200 non-parametric bootstrap
replicates.
To estimate Bayesian posterior probabilities of nodes, MR.BAYES 3.1 (Huelsenbeck
and Ronquist, 2001; Ronquist and Huelsenbeck, 2003) was used to conduct 5x106
generations of a Bayesian run under an HKY+I+G model with the default flat priors and
sampling every 100th generation. Although DT-MODSEL selected the HKY+I+G model of
evolution for both the cyt b and CR data sets, the values for transition/transversion ratios,
proportion of variable sites, and among site rate heterogeneity differed for each gene.
Therefore, the cyt b and CR sequences were partitioned and the data sets unlinked,
thereby allowing these parameters to vary across the two data sets during analysis. Two
independent runs were performed simultaneously on the data with each run using one
cold and three heated chains. Examination of the posterior probability distributions
suggested the Markov chain reached stationarity within 200,000 generations but the first
25% of the samples (1,750,000 generations) was discarded as ‘burn in’ to ensure
stationarity. The average standard deviation of split frequencies between the two
independent runs at completion was 0.0036, indicating convergence of the two runs on a
stationary distribution.
Testing Topologies
75
Phylogenetic congruence of trees from constrained and unconstrained ML analyses
was tested using four different methods: Shimodaira-Hasegawa (SH) test (Shimodaira
and Hasegawa, 1999), Approximately Unbiased (AU) test (Shimodaira, 2002),
parametric bootstrap (Huelsenbeck et al., 1996), and Bayesian posterior probabilities
(Huelsenbeck et al., 2002). These tests are commonly used in topological comparisons of
phylogenies but differ in their intrinsic statistical qualities. Testing for phylogenetic
concordance using this suite of tests allows one to more easily determine the degree of
confidence to place on the resulting p-values.
Shimodaira-Hasegawa Test
The SH test (Shimodaira and Hasegawa, 1999) is a modified version of the Kishino-
Hasegawa test (Kishino and Hasegawa, 1989) and is often preferred because of its ability
to compare an a posteriori topology (e.g., a ML topology derived from the dataset) to a
topology of interest (Goldman et al., 2000). Even though the SH test is capable of
simultaneously testing among many alternative topologies, we used the minimum number
of two topologies. This allows for a more direct comparison of the results with other
topological tests which can only test between two topologies at a time (Buckley 2002).
The best constrained and unconstrained ML trees were compared in PAUP* using 1000
bootstrap replicates and the RELL resampling criteria.
Approximately Unbiased Test
While the SH test is generally considered an appropriate test for comparing tree
topologies, it has been noted that it may be too conservative of a test (i.e. less likely to
76
reject alternative topologies under consideration) (Shimodaira, 2002; Buckley 2002). For
this reason, the AU test was developed to reduce the potential bias of the SH test
(Shimodaira, 2002). We conducted the AU test in Consel (Shimodaira and Hasegawa,
2001) using the site-wise log-likelihood values from the best ML tree obtained from the
data and the ML tree constrained a topology consistent with P. vandykei.
Parametric Bootstrap
Phylogenetic concordance was also tested using a parametric bootstrap (Goldman,
1993; Huelsenbeck and Bull, 1996). The model of sequence evolution selected by DT-
ModSel (HKY+I+G) was used to simulate 100 datasets on the constrained topology using
Seq-Gen (Rambaut and Grassly, 1997). Constrained and unconstrained ML searches were
conducted on each simulated dataset in PAUP* and the null distribution of the test
statistic was generated by calculating the difference of log likelihood scores from each
dataset (δ = ln L constrained – ln L unconstrained). This same difference in log likelihood scores
of the observed sequence data is used as the test statistic to evaluate phylogenetic
concordance between the constrained and unconstrained ML trees.
Bayesian posterior probabilities
While the parametric bootstrap assesses topological uncertainty by generating a null
distribution of the test statistic using simulated data under the chosen model of evolution,
Bayesian hypothesis testing generates a distribution of trees with high posterior
probabilities given the data, prior probabilities, and model of evolution. Two independent
Bayesian runs were performed simultaneously on the data for 5 x 106 generations with
77
topologies sampled every 100th generation. After discarding the first 25% of samples as
‘burn in’ the remaining 37,500 topologies from each run were imported into PAUP*.
This posterior distribution of topologies was then filtered with the constrained topology.
The proportion of trees in the distribution consistent with the constrained topology is the
Bayesian conditional probability that the constrained topology is correct (Huelsenbeck et
al., 2002).
Nested clade analysis
To test for significant association of haplotypes with geography, we performed a
Nested Clade Analysis (NCA) (Templeton et al., 1987, 1995). A minimum spanning
network was constructed using TCS 1.18 (Clement et al., 2000) and haplotypes were
nested using rules of Templeton et al. (1987) and Templeton and Sing (1993).
Geographical localities for each population were calculated using latitude and longitude.
GEODIS 2.4 (Posada et al., 2000) was used to test for significant association of haplotypes
and geography. Although there is some controversy surrounding the validity of NCA (see
Knowles and Maddison 2002?), we followed the inference key that was revised to deal
with this criticism in Templeton (2004) for clades with significant geographical
associations.
Results
Summary of DNA sequences
We sequenced 1830 nucleotides of mitochondrial DNA; 1135 bases of partial cyt b
sequence and 695 bases of partial CR sequence. We found 28 distinct haplotypes from 80
78
individuals; 14 haplotypes were found in multiple individuals and 14 haplotypes were
represented by single individuals. The most frequently sampled haplotype, designated as
‘Sol Duc 1’ in the phylogeny (Fig. 2), was found in 16 of the 80 individuals (20.0%) and
was present in 6 of the 24 localities, all of which are located in the Olympic Peninsula.
Phylogenetic analyses
There were 112 parsimony-informative sites in the complete data set and the MP
analysis found eight equally parsimonious MP trees with a tree length of 272 steps.
Topology of bootstrapped MP trees and the single best ML tree (-ln 4071.6256) with
branch support over 50% were identical to the Bayesian topology (Fig. 2).
The phylogeny of D. copei reveals a well-supported lineage corresponding to
populations in a small geographic area along the southern edge of the Columbia River
Valley (Fig. 2). There is also support for a sister relationship between the coastal
populations and inland populations occurring in the Cascade Mtns north of the Columbia
River. Other populations in the Cascade Mtns that occur south of the Columbia River
form a separate clade that is sister to the coastal and north Cascadian lineages. The
phylogenetic pattern is suggestive of a sister relationship between monophyletic coastal
and Cascadian lineages, but only when considering populations occurring within the
range of Plethodon vandykei. Populations of D. copei that occur outside the range of P.
vandykei, namely Cascade populations south of the Columbia River, display a high
degree of divergence from the remainder of the D. copei populations (Table 1).
Phylogenetic Concordance
79
The combined results of the SH, AU, parametric bootstrap and Bayesian hypothesis
tests confirm that the topology of the D. copei phylogeny is not concordant with the P.
vandykei phylogeny (Table 2). All tests resulted in significant p-values, and the most
conservative SH test, resulted in a significant p-value (0.045), while the parametric
bootstrap and Bayesian hypothesis test easily rejected concordance (p<0.001). These
results are consistent with comparative studies which indicate a tendency for the SH test
to be conservative while Bayesian posterior probabilities and parametric bootstrap tests
readily reject phylogenetic concordance (Buckley, 2002). The AU test, which was
developed to reduce conservative bias in the SH test (Shimodaira, 2002), still had a
highly significant p-value (0.007) that was intermediate between the SH test and
Bayesian and parametric bootstrap tests.
Nested clade analysis
The minimum spanning haplotype network consisted of three main clades
corresponding to populations within the Cascade Mtns, along the Pacific Coast, and the
Columbia Valley (Fig. 3). Some loops are present in the network but were resolved using
nesting procedures from Templeton et al. (1992) and Templeton and Sing (1993). These
loops ultimately do not affect nesting design or conclusions inferred from the analysis.
Seven nested haplotype networks had a significant association with geography and
conclusive inferences (Table 2). The overall pattern of the spanning network is consistent
with phylogenetic results and indicates that genetic structure of populations north of the
Columbia River share a genetic pattern similar to that of Plethodon vandykei while
populations south of the Columbia River represent divergent lineages. The major
80
inferrences from the nested clade analysis are northward expansion of Cascade
populations, colonization of the Olympic Peninsula from the Willapa Hills, and restricted
overall gene flow among the fragmented populations (Table 3, Fig. 4).
Discussion
The comparison of phylogeographic patterns between the Cope’s Giant
Salamander (D. copei) and the Van Dyke’s salamander (P. vandykei) demonstrates an
overall pattern of incongruence, while sympatric populations simultaneously exhibit
identical and congruent patterns. This result provides evidence that similarly distributed
organisms can demonstrate concordant phylogenies within their shared distribution, but
that allopatric populations may display significant levels of phylogenetic signal and
effectively obscure any congruent phylogeographic pattern.
Phylogeography of the Cope’s Giant Salamander
The deepest phylogenetic divergence among D. copei populations is the separation of
populations in the Columbia River Valley from the remainder of all other populations.
These divergent populations are geographically restricted to several short tributaries that
drain directly into the Columbia River and are not joined to the large interconnected
network of headwater streams that run throughout the region. Because D. copei rarely
metamorphose and remain primarily in an aquatic phase, the lack of connection with
other watersheds seems to have prevented stream-based dispersal into and out of this
population. The Columbia River Valley has been identified as a Pleistocene refugium for
a variety of fishes (Brown et al., 1992; Bickham et al., 1995; Taylor et al., 1999;
81
McCusker et al., 2000; Haas and McPhail, 2001) and for other salamander species
(Wagner et al., 2005; Steele and Storfer, in press). These divergent populations of D.
copei appear to have been restricted into several streams within this glacial refugium and
have subsequently remained isolated within the Columbia River Valley.
The Columbia River appears to be a fairly strong barrier to gene flow for this species
as there is no geographic overlap of haplotypes found north and south of the Columbia
River despite the species’ distribution encompassing both sides of the river (Fig. 1). The
Fox Creek population (locality #11), which occurs south of the Columbia River in the
Willapa Hills region of Oregon, is not as phylogenetically distinct as populations
occurring south of the Columbia in the Cascades; however, this population appears to
have been separated long enough to accumulate a high number of mutations between
haplotypes therein and the nearest northern haplotype (Fig. 3).
The Columbia River seems to be a barrier of varying degrees of penetrability for
different amphibian species. Similar to D. copei, the Larch Mountain salamander
(Plethodon larselli) has recently expanded northwards across the river into its current
range (Wagner et al., 2005). However, the river separates highly divergent northern and
southern lineages of the Pacific Giant Salamander (D. tenebrosus) (Steele and Storfer, in
press) and appears to have prevented further northward expansion of the Oregon Slender
Salamander (Batracoseps wrightii) (Miller et al., 2005).
Dicamptodon copei and Plethodon vandykei: Same but different
Despite D. copei and P. vandykei being distantly related salamander species, it was
expected that, due to similarity in habitat requirements, they would have responded
82
concordantly to past geologic and climatic events. In addition, their similarly fragmented
geographic distributions further suggested a similarity in phylogenetic topologies.
Phylogenetic topologies of the two species were indeed similar, but only when
considering populations in sympatry. The removal of allopatric populations from the
dataset results in a phylogenetic topology identical to that of P. vandykei (not shown) and
sympatric populations of D. copei display a sister relationship between coastal
populations and populations in the north Cascade Mtns. Long term separation between
coastal and Cascade populations of P. vandykei has been inferred by molecular and
morphological evidence (Howard et al., 1993; Wilson and Larsen, 1999). This separation,
as first mentioned by Wilson and Larsen (1999), is in agreement with the fossil pollen
record (Baker, 1983; Barnosky et al., 1987) and indicates an uninhabitable xeric
environment in the lowlands separating the coastal and Cascade populations of P.
vandykei during late Pleistocene through much of the Holocene. Presumably, these dry
lowlands had a comparable effect on sympatric D. copei populations, resulting in
phylogeographic patterns similar to that of P. vandykei. Additionally, both species
appeared to have responded similarly to the availability of a post-glacial environment in
the Olympic Peninsula (Crandell, 1965; Easterbrook, 1976) by expanding there from the
Willapa Hills. The presence of a widespread haplotype in the Olympic Peninsula (Sol
Duc 1) indicates recent expansion into this region by D. copei. This expansion is also
corroborated by the results of the nested clade analysis (Clade 2-7; Table 3). Plethodon
vandykei also seems to have recently expanded into the Olympic Peninsula because the
smallest morphological differences within this species occur between populations from
the Olympic Peninsula and the Willapa Hills (Wilson and Larsen, 1999); these same
83
populations are also shown to be electrophoretically indistinguishable in allozyme
genotypes (Howard et al., 1993). Thus, in the areas where the two species have
overlapping ranges, they have responded concordantly to past climatic and geologic
events, resulting in similar phylogeographic topologies.
Phylogenetic concordance among sympatric populations of D. copei and P. vandykei
encompasses the extent of any phylogeographic similarity. The range of D. copei within
the Cascade Mtns is slightly larger than that of P. vandykei and extends ~100 km
southward across the Columbia River into the Cascade Mtns of Oregon. These allopatric
populations of D. copei tend to be distinct from populations north of the Columbia River
(Fig. 2). Topological incongruence between the P. vandykei phylogeny and the complete
D. copei phylogeny is driven by the occurrence of these genetically divergent D. copei
populations. Although the geographic distribution of D. copei is only slightly larger than
the range of P. vandykei, it encompasses a geographical barrier (i.e. Columbia River)
capable of producing significant phylogenetic signal. While there is evidence of shared
responses to past climatic or geologic events by some populations, the phylogenetic
topologies of all populations for the two species display significant discordance.
Comparative phylogeography
A phylogeographic comparison of P. vandykei and D. copei provides an example of
two codistributed species that are dissimilar in their phylogenetic topologies but
nonetheless show some concordance in their past responses to a changing environment. A
variety of comparative phylogeographic studies have demonstrated concordant responses
of codistributed taxa to either past climatic events (Avise, 1992; Avise 1996) or geologic
84
events (Nielson et al., 2001; Carstens et al., 2004; Steele et al., 2005), while other studies
reveal discordant topologies due to independent responses to past climatic and geologic
events (Sullivan et al., 2000; Carstens et al., 2005b; Donovan et al., 2000). As more
comparative studies are completed, it will likely become clear that the two alternative
hypotheses of concordant and independent responses for codistributed taxa represent a
false dichotomy (Sullivan et al., 2000). The two alternative hypotheses of concerted and
independent responses of codistributed taxa are not always mutually exclusive and
topologies may simultaneously display characteristics predicted by both hypotheses
(Sullivan et al., 2000). Thus, codistributed species are likely to have a combination of
concordant and dissimilar patterns in their phylogenies indicating some degree of a
shared history but not complete phylogeographic concordance.
References
Argobast, B.S., Kenagy, G.J., 2001. Comparative phylogeography as an integrative
approach to historical biogeography. J. Biogeog. 28, 819–825.
Avise, J.C., 1992. Molecular population structure and the biogeographic history of a
regional fauna: a case history with lessons for conservation biology. Oikos 63, 62–76.
Avise, J.C., 1996. Toward a regional conservation genetics perspective: phylogeography
of faunas in the southeastern United States. In: Avise, J.C., Hamrick, J.L. (Eds.),
Conservation Genetics: Case Histories from Nature. Chapman and Hall, New York,
pp. 431–70.
Avise, J.C., 2000. Phylogeography: the history and formation of species. Harvard
University Press, Cambridge.
85
Baker, R.G., 1983. Holocene vegetational history of the western United States. In:
Wright, Jr., H.E. (Ed.), Late Quaternary environments of the United States. Vol. 2.
University of Minnesota Press, Minneapolis, pp. 109–127.
Barnosky, C.W., 1981. A record of late Quaternary vegetation from Davis Lake, southern
Puget Lowland, Washington: Quatern. Res. 16, 221–239.
Bickham, J.W., Wood, C.C., Patton, J.C., 1995. Biogeographic implications of
cytochrome b sequences and allozymes in sockeye (Oncorhynchus nerka). J.
Hered. 86, 140–144.
Bowen, B.W., Avise, J.C., 1990. The genetic structure of Atlantic and Gulf of Mexico
populations of sea bass, menhaden, and sturgeon: the influence of zoogeographic
factors and life history patterns. Mar. Biol. 107, 371–381.
Brunsfeld, S.J., Sullivan, J., Soltis, D.E., Soltis, P.S., 2001. Comparative
phylogeography of northwestern North America: a synthesis. In: Silvertown, J.,
Antonovics, J. (Eds.), Integrating ecology and evolution in a spatial context.
Blackwell Publishing, Williston, VT, pp. 319–339.
Brodie, Jr., E.D., 1970. Western salamanders of the genus Plethodon: systematics and
geographic variation. Herpetologica 26, 468–516.
Brown, J.R., Beckenbach, A.T., Smith, M.J., 1992. Influence of Pleistocene glaciations
and human interaction upon mitochondrial DNA diversity in white sturgeon
(Acipenser transmontanus) populations. Can. J. Fish. Aquat. Sci. 49, 358–367.
Buckley, T.R., 2002. Model misspecification and probabilistic tests of topology:
evidence from empirical data sets. Syst. Biol. 51, 509–523.
Carstens, B.C., Stevenson, A.L., Degenhardt, J.D., Sullivan, J., 2004. Testing nested
86
phylogenetic and phylogeographic hypotheses in the Plethodon vandykei species
group. Syst. Biol. 53, 781–792.
Carstens, B.C., Degenhardt, J.D., Stevenson, A.L., Sullivan, J., 2005a. Accounting for
coalescent stochasticity in testing phylogeographic hypotheses: modeling
Pleistocene population structure in the Idaho Giant Salamander Dicamptodon
aterrimus. Mol. Ecol. 14, 255–265.
Carstens, B.C., Brunsfeld, S.J., Demboski, J.R., Good, J.D., Sullivan, J., 2005b.
Investigating the evolutionary history of the Pacific Northwest mesic forest
ecosystem: hypothesis testing within a comparative phylogeographic framework.
Evolution 59, 1639–1652.
Clement, M., Posada, D., Crandall, K., 2000. TCS: a computer program to estimate gene
genealogies. Mol. Ecol. 9, 1657–1660.
Crandell, D., 1965. The glacial history of western Washington and Oregon. In: Wright,
Jr., H.E., Frey, D.G. (Eds.), The Quaternary of the United States, Princeton
University Press, Princeton, pp. 341–353.
Donovan, M.F., Semlitsch, R.D., Routman, E.J., 2000. Biogeography of the southeastern
United States: a comparison of salamander phylogeographic studies. Evolution
54, 1449–1456.
Easterbrook, D.J., 1976. Quaternary geology of the Pacific Northwest. In: Mahaney,
W.C. (Ed.), Quaternary stratigraphy of North America. Dowden, Hutchinson and
Ross, Stroudsburg, PA, pp 441-462.
Goldman, N.J., 1993. Statistical tests of models of DNA substitution. J. Mol. Evol.
36, 182–198.
87
Goldman, N., Anderson, J.P., Rodrigo, A.G., 2000. Likelihood-based tests of
topologies in phylogenetics. Syst. Biol. 49, 652–670.
Haas, G.R., McPhail, J.D., 2001. The post-Wisconsinan glacial biogeography of bull
trout (Salvelinus confluentus): a multivariate morphometric approach for
conservation biology and management. Can. J. Fish. Aquat. Sci. 58, 2189–2203.
Hewitt, G.M., 1999. Post-glacial recolonisation of European biota. Biol. J. Linn. Soc. 58,
87–112.
Howard, J.H., Seeb, L.W., Wallace, R., 1993. Genetic-variation and population
divergence in the Plethodon vandykei species group (Caudata, Plethodontidae).
Herpetologica 49, 238–247.
Huelsenbeck, J.P., Bull, J.J., 1996. A likelihood ratio test to detect conflicting
phylogenetic signal. Syst. Biol. 45, 92–98.
Huelsenbeck, J.P., Hillis, D.M., Jones, R., 1996. Parametric bootstrapping in
molecular phylogenetics: applications and performance. In: Ferraris, J.D.,
Palumbi, S.R. (Eds.), Molecular zoology: advances, strategies, and protocols.
Wiley-Liss, New York, pp. 19–45.
Huelsenbeck, J.P., Lagret, B., Miller, R.E., Ronquist, F., 2002. Potential applications and
pitfalls of Bayesian inference of phylogeny. Syst. Biol. 51, 673–688.
Huelsenbeck, J.P., Ronquist, F., 2001. MRBAYES: Bayesian inference of phylogenetic
trees. Bioinformatics 17, 754–755.
Kishino, H., Hasegawa, M., 1989. Evaluation of the maximum-likelihood estimate of the
evolutionary tree topologies from DNA sequence data, and the branching order in
Hominoidea. J. Mol. Evol. 29, 170–179.
88
McCusker, M.R., Parkinson, E., Taylor, E.B., 2000. Mitochondrial DNA variation in
rainbow trout (Oncorhynchus mykiss) across its native range: testing
biogeographical hypotheses and their relevance to conservation. Mol. Ecol. 9,
2089–2108.
Michaux, J.R., Libois, R., Filippucci, M.G., 2005. So close and so different: comparative
phylogeography of two small mammal species, the yellow-necked fieldmouse
(Apodemus flavicollis) and the woodmouse (Apodemus sylvaticus) in the western
Palearctic region. Heredity 94, 52–63.
Miller, M.P., Haig, S.M., Wagner, R.S., 2005. Conflicting patterns of genetic structure
produced by nuclear and mitochondrial markers in the Oregon slender salamander
(Batrachoseps wrighti): implications for conservation efforts and species
management. Conserv. Genet. 6, 275–287.
Minin, V., Abdo, Z., Joyce, P., Sullivan, J., 2003. Performance-based selection of
likelihood models for phylogeny estimation. Syst. Biol. 52, 674–683.
Nielson, M.K., Lohman, K., Sullivan, J., 2001. Phylogeography of the tailed frog
(Ascaphus truei): implications for biogeography of the Pacific Northwest.
Evolution 55:147–160.
Nussbaum, R.A., 1976. Geographic variation and systematics of salamanders of the
genus Dicamptodon Strauch (Ambystomatidae). Miscellaneous Publications No.
149, Museum of Zoology, University of Michigan, Ann Arbor.
Posada, D., Crandall, K.A., Templeton, A.R., 2000. GeoDis: a program for the nested
cladistic analysis of the geographical distribution of genetic haplotypes. Mol.
Ecol. 9, 487–488.
89
Rambaut, A.E., Grassly, N.C., 1997. SEQ-GEN: an application for the Monte Carlo
simulation of DNA sequence evolution along phylogenetic trees. Comp. Appl.
Biosci. 13:235–238.
Riddle, B.R., Hafner, D.J., Alexander, L.F., Jaeger, J.R., 2000. Cryptic vicariance in the
historical assembly of a Baja California peninsular desert biota. Proc. Natl. Acad.
Sci. USA 97, 14438–14443.
Rocha, L.A., Robertson, D.R., Roman, J., Bowen, B.W., 2005. Ecological speciation in
tropical reef fishes. Proc. R. Soc. Lond. B 272, 573–579.
Ronquist, F., Huelsenbeck, J.P., 2003. MrBayes 3: Bayesian phylogenetic inference
under mixed models. Bioinformatics 19, 1572–1574.
Sambrook, J., Fritsch, E.F., Maniatis, T., 1989. Molecular cloning: a laboratory manual,
second ed. Cold Spring Harbor Laboratory Press, New York.
Schneider, C.J., Cunningham, M., Moritz, C., 1998. Comparative phylogeography and
the history of endemic vertebrates in the wet tropics rainforests of Australia. Mol.
Ecol. 7, 487–498.
Shaffer, H.B., McKnight, M.L., 1996. The polytypic species revisited: genetic
differentiation and molecular phylogenetics of the tiger salamander (Ambystoma
tigrinum) (Amphibia: Caudata) complex. Evolution 50, 417–433.
Shimodaira, H., Hasegawa, M., 2001. CONSEL: for assessing the confidence of
phylogenetic tree selection. Bioinformatics 17, 1246–1247.
Shimodaira, H., 2002. An approximately unbiased test of phylogenetic tree selection.
Syst. Biol. 51, 492–508.
Shimodaira, H., Hasegawa, M., 1999. Multiple comparisons of log-likelihoods with
90
applications to phylogenetic inference. Mol. Biol. Evol. 16, 1114–1116.
Soltis, D. E., M. A. Gitzendanner, D. D. Strenge, and P. E. Soltis. 1997. Chloroplast
DNA intraspecific phylogeography of plants from the Pacific Northwest of North
America. Plant Syst. Evol. 206:353–373.
Steele, C.A., Carstens, B.C., Storfer, A., Sullivan, J., 2005. Testing hypotheses of
speciation timing in Dicamptodon copei and Dicamptodon aterrimus (Caudata:
Dicamptodontidae). Mol. Phylogenet. Evol. 36, 90–100.
Sullivan, J., Arellano, E., Rogers, D.S., 2000. Comparative phylogeography of
mesoamerican highland rodents: concerted versus independent responses to past
climatic fluctuations. Am. Nat. 155, 755–768.
Swofford, D.L., 2002. PAUP*. Phylogenetic Analysis Using Parsimony (*and other
methods). Ver. 4. Sinauer Associates, Sunderland, MA.
Taberlet, P., Fumagalli, L., Wust-Saucy, A.G., Cosson, J.F., 1998. Comparative
and postglacial colonization routes in Europe. Mol. Ecol. 7, 453–464.
Taylor, E.B., Pollard, S., Louie, D., 1999. Mitochondrial DNA variation in bull trout
(Salvelinus confluentus) from northwestern North America: implications for
zoogeography and conservation. Mol. Ecol. 8, 1155–1170.
Templeton, A.R., 2004. Statistical phylogeography: methods of evaluating and
minimizing inference errors. Mol. Ecol. 4, 789–809.
Templeton, A.R., Sing, C.F., 1993. A cladistic analysis of phenotypic associations with
haplotypes inferred from restriction endonuclease mapping. IV. Nested analyses
with cladogram uncertainty and recombination. Genetics 134, 659–669.
Templeton, A.R., Boerwinkle, E., Sing, C.F., 1987. A cladistic analysis of phenotypic
91
associations with haplotypes inferred from restriction endonuclease mapping. I.
Basic theory and analysis of alcohol dehydrogenase activity in Drosophila.
Genetics 117, 343–351.
Templeton, A.R., Crandall, K.A., Sing, C.F., 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.
Templeton, A.R., Routman, E., Phillips, C.A., 1995. Separating population structure
from population history: a cladistic analysis of the geographical distribution of
mitochondrial DNA haplotypes in the tiger salamander, Ambysoma tigrinum.
Genetics 140, 767–782.
Wagner, R.S., Miller M.P., Crisafulli, C.M., Haig, S.M., 2005. Geographic variation,
genetic structure, and conservation unit designation in the Larch Mountain
Salamander (Plethodon larselli). Can. J. Zool. 83, 396–406.
Wilson, Jr., A.G., Larsen, Jr., J.H., McAllister, K.R., 1995. Distribution of Van Dyke's
salamander (Plethodon vandykei Van Denburgh). Am. Midl. Nat., 134:288–393.
Wilson, A.G., Larsen, J.H., 1999. Morphometric analysis of salamanders of the
Plethodon vandykei species group. Am. Midl. Nat. 141, 266–276.
Zink, R.M., 1996. Comparative phylogeography in North American birds. Evolution 50,
308–317.
Zink, R.M., 2002. Methods in comparative phylogeography, and their application to
studying evolution in the North American aridlands. Integr. Comp. Biol. 42, 953–
959.
92
Table 1
Genetic distances among main lineages of Dicamptodon copei. Shown below the
diagonal are genetic distances corrected under the HKY+I+G model of sequence
evolution in units of substitutions per site. Uncorrected percent sequence divergences are
shown above the diagonal.
Columbia Valley S. Cascade Mtns N. Cascade Mtns Coastal Columbia Valley — 0.01422 0.01464 0.01521 S. Cascade Mtns 0.01558 — 0.00923 0.1177 N. Cascade Mtns 0.01605 0.00983 — 0.01 Coastal 0.01683 0.01272 0.01064 —
93
Table 2 Results for tests of phylogenetic concordance between D. copei and P. vandykei.
Ln Likelihood scores P-value Constrained Unconstrained δ SH test AU test P-boot Bayesian PP 4079.7863 4071.6256 8.1607 0.045 0.007 < 0.001 < 0.001
94
Table 3
Results and inferences of nested clade analysis. Haplotype networks without significant
geographical associations or significant networks with inconclusive inferences are not
listed.
Clade χ2 statistic Probability Inference chain Inferred pattern*
1-1 13.36 0.012 1-2-11-12-No Contiguous northward range expansion of northern most populations in Cascade Mtns
1-5 1.0 0.000 1-2-3-5-6-7-Yes RGF with some long distance dispersal within Olympic Mtns
2-7 27.00 0.001 1-2-3-5-6-7-8-Yes Past gene flow from Willapa Hills into Olympic Mtns followed by extinction of intermediate populations
4-3 16.00 0.017 1-2-11-17-4-No RGF within Cascade populations north of Columbia River
5-1 66.07 0.000 1-2-3-5-6-13-Yes Long distance colonization of Cascade populations across Columbia River coupled with subsequent fragmentation
5-2 24.15 0.016 1-2-11-Yes Southward range expansion across Columbia River of populations in Willapa Hills
6-1
1.0 0.000 1-2-3-5-6-7-Yes RGF with some long distance dispersal among Cascade, Coastal, and Columbia populations
*RGF = restricted gene flow
95
(Canada) (USA)
Dicamptodon copei
BC
Pacific Ocean
Olympic Mtns and Willapa Hills
Cascade Mtns
Plethodon vandykei
23
15
12
24
18 19
21 20 22
16 17
14
13
11
10
9
8 7
6
5
4
3 2
1
WA
OR
Fig. 1. Shaded area indicates distribution of Dicamptodon copei within the Pacific Northwest of the United States. Inset map indicates range of the codistributed salamander species Plethodon vandykei and phylogenetic relationships among the regions as indicated by morphological (Wilson and Larsen, 1999) and allozyme studies (Howard et al., 1993). Darkly shaded areas indicate regions of D. copei’s range where P. vandykei does not occur. Numbers indicate sampled localities. Localities in Olympic Mtns correspond to numbers 1–8, Willapa Hills comprise samples 9–11, and Cascade Mtns correspond to localities 12–24.
96
Fig. 2. Phylogeny of Dicamptodon copei based on 1135 bp of cytb b and 695 bp
of control region. Bayesian topology is presented but is identical to bootstraped MP and ML topologies. Values above branches are Bayesian posterior probabilities; below branches are ML and MP bootstrap values (respectively) from 200 replicates. Four main lineages are indicated corresponding to: Cascade Mtns north of Columbia River (localities 12–19), Cascade Mtns south of Columbia River (localities 23–24), Coastal lineages in Olympic Peninsula and Willapa Hills (localities 1–11), and populations restricted to several tributaries in the Columbia River Valley (localities 20–22).
97
Columbia River Valley
B
S1
T2
E
WS
T4
Mc
Mc3
Ma
L2L4
(4-3)
Cascade Mtns (5-1)
SD1(2-7)
E3
F1 F2
E2
T1
St10
E
(1-5)
Coastal (5-2)
Y4Tr4Tr2
BV Tr1
Y2
Y3
M4(1-1)
Fig. 3. Minimum spanning haplotype network for Dicamptodon copei. Lines indicate a connection between haplotypes. Missing haplotypes are shown as black dots. Sampled haplotypes are designated with abbreviations and correspond to those in the Appendix. One-step clades are shown in white, two-step clades in light gray, three-step clades in medium gray, four-step clades in dark gray and five-step clades in black. Clade numbers are shown only for clades with significant association with geography.
98
3.) 4.)
2.) 1.)
Fig. 4. Historical demographic patterns for Dicamptodon copei as inferred by the nested clade analysis. A black dot represents the location of the divergent Columbia River populations. 1.) Populations colonize areas north of Columbia River, 2.) Restricted western gene flow established coastal populations, 3.) Coastal populations expand north into Olympic Peninsula and south across Columbia River, 4.) Cascade populations expand northward into current distribution of species.
99
Appendix 1. Locality information of Dicamptodon copei samples used in this study. Unique haplotype sequences are deposited in GenBank and accession numbers refer to cyt b and control region, respectively.
Locality Number
Number Sequenced
Name of haplotypes sampled (abbreviated haplotype name)
Locality information
1 2 Sol Duc 1 (SD1) Lake Crk, Clallam Co, WA 2 3 Sol Duc 1 (SD1) Hyas Crk, Clallam Co, WA 3 3 Sol Duc 1 (SD1) Sol Duc Crk, Clallam Co, WA 4 2 Sol Duc 1 (SD1) Tower Crk, Jefferson Co, WA 5 4 Sol Duc 1 (SD1) Sam’s River, Jefferson Co, WA 6 2 Sol Duc 1 (SD1) July Crk, Grays Harbor Co, WA 7 3 Elk 2 (E2), Elk 3 (E3) Elk Crk, Mason Co, WA 8 5 Elk 3 (E3) Cabin Crk, Mason Co, WA 9 3 Stillman 10 (St10) Sillman Basin, Lewis Co, WA
10 2 Elochoman 1 (E) W Fk Elochoman Crk, Wahkiakum Co, WA 11 2 Fox 1 (F1), Fox 2 (F2) Fox Crk, Clatsop Co, OR 12 3 Mona 4 (M4) Trib Nisqually River, Lewis Co, WA 13 1 East 1 (E) East Crk, Lewis Co, WA 14 2 Little 4 (L4) Jefferson Crk, Skamania Co, WA 15 8 Little 2 (L2), Little 4 (L4) Little Crk, Skamania Co, WA 16 2 White Salmon 1 (WS) White Salmon Crk, Skamania Co, WA 17 5 Trout 1 (T1), Trout 2 (T2),
Trout 4 (T4) Trout Crk, Skamania Co, WA
18 4 Mabee 1 (Ma1), McClowsky 1 (Mc1)
Trib Washougal River, Skamania Co, WA
19 4 Mabee 1 (Ma1), McClowsky 1 (Mc1), McClowsky 3 (Mc3)
McClowsky Crk, Skamanina Co, WA
20 5 Young 2 (Y2), Young 3 (Y3), Young 4 (Y4)
Young Crk, Multnomah Co, OR
21 1 Bridal Veil 1 (BV) Bridal Veil Crk, Hood River Co, OR 22 6 Trib Young 1 (Tr1), Trib Young
2 (Tr2), Trib Young 4 (Tr4) Unnamed tributary of Young Crk, Multnomah Co, OR
23 5 Still 1 (S1), Still4 Sill Crk, Clakamas Co, OR. 24 3 Boulder 1 (B) Boulder Crk, Wasco Co, OR.
100
CHAPTER FOUR
SCALING UP FROM LIFE HISTORY DYNAMICS TO PHYLOGEOGRAPHIC
PATTERNS: A COMPARATIVE STUDY OF TWO SYMPATRIC
SALAMANDER TAXA
Abstract
There are conceptual relationships between life history variation, dispersal ability,
genetic connectivity and phylogeographic distributions. Yet empirical studies that link local life
history variation to understanding variation in species’ geographic distributions are rare.
Organisms with life histories that include high dispersal potential tend to have little genetic
population structure while the opposite is true for organisms with lower dispersal ability.
Although it has rarely been tested, the predictions about local life history variation should scale
up to predictions about species’ ranges and phylogeographic patterns. That is, species with
higher local dispersal should have larger geographic distributions with less phylogeographic
structure than those with lower dispersal. We test these predictions in a model system using two
closely related taxa of stream-breeding giant salamanders in the Pacific Northwest. Cope’s giant
salamander (Dicamptodon copei) rarely metamorphoses and dispersal and gene flow should be
limited along stream corridors. In contrast, Pacific giant salamanders (D. tenebrosus) generally
metamorphose into terrestrial adults and should have overland as well as stream based dispersal
and gene flow. We use neutral microsatellite markers to test the predictions that Pacific giant
salamanders have higher dispersal and gene flow than Cope’s giant salamanders in several
watersheds in the Cascade Mountains of Washington State where the two species are sympatric.
101
Results indicate that the metamorphosing species (D. tenebrosus) displayed a lack of genetic
population structuring, no pattern of isolation by distance, and a low overall FST value while the
non-metamorphosing species (D. copei) displayed a large degree of genetic population structure,
significant isolation by distance and significantly higher overall FST value. This pattern help
explain the phylogeographic distributions of the two species. Pacific giant salamanders have a
broader and more contiguous distribution than Cope’s giant salamanders and show post-
Pleistocene dispersal and colonization. In contrast, Cope’s giant salamanders have a more limited
geographic range, where they are restricted to three geographically distinct mountain ranges and
display more phylogeographic structure. These results support the conceptual notion that an
understanding of life history variation on a local scale can lead to a better understanding of the
causation of species’ distributions in general.
Key words: life history variation, dispersal, comparative gene flow, population structure,
phylogeography, Dicamptodon copei, Dicamptodon tenebrosus.
INTRODUCTION
Variation in life history traits have clear affects on evolutionary patterns and processes
within populations (Newman 1992, Roff 1992, Stearns 1992). One central life history
characteristic is dispersal ability, which is capable of structuring populations genetically
(Bohonak 1999). Variation in dispersal ability should also affect species’ distributions and
phylogeographic patterns, whereby high dispersal should lead to greater gene flow on the local
scale, reduced phylogeographic structure and lead to broader species’ geographic distributions
102
than species with limited dispersal. A variety of studies have linked dispersal ability to genetic
patterns at the population level (Hellberg 1996, King and Lawson 2001, Dawson et al. 2002), but
to out knowledge no study has linked patterns of life history variation to gene flow and then
linked patterns of gene flow to broader phylogeographic patterns.
Variation in dispersal ability should be associated with phylogeographic patterns because
intraspecific evolution is often influenced by differential rates of gene flow among populations
(Lomolino et al. 2006). Specific predictions about the relationship between dispersal ability and
degree of population structuring already exist such that organisms with high dispersal ability
tend to display increased gene flow among populations resulting in lower population
differentiation than organisms with lower dispersal abilities (Bohonak 1999). If patterns of
genetic structuring at the population level are scaled up to the species-level, it should be possible
to make predictions about phylogeographic patterns. We predict that low dispersal organisms
will display not only low levels of gene flow and higher levels of population structuring but also
limited geographic distribution and high levels of regional phylogeographic structuring due to
greater susceptibility to vicariance events. In contrast we predict that species with greater
dispersal abilities to have higher levels of gene flow resulting in lower levels of population
structuring, more continuous geographic distributions, and lower levels of phylogeographic
structuring facilitated recent range expansion.
In order to properly evaluate these predictions researchers need to simultaneously
compare the genetic population structure of high and low dispersal organisms. Ideally, such
studies should be performed in a common environment on closely related species that have clear
differences in dispersal capabilities. Failure to control for evolutionary history by comparing
103
sympatric species that are not closely related can confound life history traits associated with
phylogeny, or comparing species that are phylogenetically similar but allopatric can confound
life history differences with dissimilar habitats or environmental histories. Because of these
restrictions, it can be difficult to find study organisms that meet these criteria and as a result, few
studies have examined comparative patterns of gene flow in organisms in which clear dispersal
differences are known for phylogenetically comparable species within a common environment
(e.g. King and Lawson 2001, Dawson et al. 2002).
Two organisms meet these requirements and provide an ideal opportunity to test
hypotheses of how life history traits affect gene flow and phylogeography. These are two
sympatric species of giant salamander in the genus Dicamptodon: Cope’s giant salamander (D.
copei) and the Pacific giant salamander (D. tenebrosus). These two species of stream-breeding
salamanders not only have partially overlapping distributions in portions of the Pacific
Northwest but can be found in close sympatry, thereby providing an opportunity to test for
contrasting rates of gene flow of each species within a common environment. These two species
also have clear differences in life history traits that should affect their dispersal ability. The
larvae of D. tenebrosus commonly metamorphose into terrestrial adults that may disperse
overland from their natal streams. Therefore, movement between streams may be correlated with
overland distance and a low degree of genetic differentiation among populations, which should
scale up to shallow phylogeographic structure at the species scale. In contrast, D. copei is a
neotenic species and retains larval characteristics (e.g. gills) throughout its life, thereby limiting
overland dispersal between localities (Nussbaum 1970, 1976). Individuals of this species should
be constrained to their natal streams which should lead to low levels of overland gene flow, and
104
result in a high degree of genetic differentiation among streams and strong phylogeographic
structure at the species scale. Because of the aquatic nature of this species, gene flow should be
correlated with stream distance between localities rather than overland distances.
Metamorphosed individuals of D. copei are rarely found and only three terrestrial specimens
have been reported (Leonard et al. 1993). In these rare cases of transformation, terrestrial D.
copei would also be capable of dispersing overland from their natal streams but at a rate much
lower than that of D. tenebrosus. However, no metamorphosis of D. copei has been reported in
this study area.
While the two species are not sister taxa phylogenetically, D. copei is more closely
related to D. tenebrosus than other species in the genus that display metamorphosis (Steele et al.
2005, Chapter 1 herein). Additionally, range-wide phylogeographic studies have been conducted
on these two organisms and information is available on the level of phylogeographic structuring
within each (Steele and Storfer 2006; Steele and Storfer submitted), making it possible to test the
prediction of correlation between variation in life history traits, population level structuring and
species-level phylogeography.
This study uses these two species of salamanders to test the predictions inverse
relationship between dispersal ability and genetic structuring at the population level as well as
across a species’ range, to determine the number of distinct populations for each species in the
study area, and to test for isolation-by-distance along overland and aquatic dispersal routes.
MATERIALS AND METHODS
105
Sample collection and DNA amplification — Tissue samples from D. copei and D.
tenebrosus were obtained from 11 localities in the Cascade Mountains of Washington State
where they are known to occur in sympatry (Fig. 1). Sites were selected such that pairwise
distances between localities represented a range of distances, thereby allowing predictions about
the relationship between dispersal ability and genetic population structure to be examined.
Localities 1 through 10 occur within one river drainage while locality 11 is in a separate
drainage. Samples of D. copei comprised neotenic adults and larvae while samples of D.
tenebrosus primarily comprised aquatic larvae but included some metamorphosed adults. DNA
was extracted from tail clips using Qiagen DNeasy kits.
We developed 15 microsatellite markers (Appendix 1 and 2) with Ecogenics GMbH for
the Cope’s giant salamander which averaged approximately 16 alleles per locus (Steele et al., in
prep. Nine of these same loci cross-amplified and were polymorphic for D. tenebrosus with an
average of approximately 9 alleles per locus. PCR conditions for microsatellite amplification
followed locus specific settings (Steele et al., submitted). Batches of samples were run with
negative controls and an identical positive control across all runs to ensure consistency in scoring
alleles. Forward PCR primers were fluorescently labeled with one of four different dye colors to
allow for multiplexing on an ABI 3730 automated sequencer (Applied Biosystems Inc., Foster
City, CA) with a LIZ 500 bp size standard. Microsatellite alleles were scored using Genemapper
v3.7 (Applied Biosystems).
Genetic analyses — The program GENEPOP version 3.4 (Raymond and Rousset 1995) was
used to asses genetic variability within and among sampled localities, to calculate number of
alleles per locus, FIS at each locus in all populations, observed and expected heterozygosities,
106
deviation from Hardy-Weinberg equilibrium (HWE), and linkage disequilibrium between loci. A
global estimate of FST and a 95% confidence interval was estimated for each species from Weir
and Cockerham’s θ (1984) using FSTAT version 2.9.3 (Goudet 2001). Pairwise FST values were
calculated using in the program ARLEQUIN version 3.01 (Excoffier et al. 2005).
The program STRUCTURE (Pritchard et al. 2000) was used to infer genetic population structure by
using a Bayesian clustering algorithm to assign individuals to k populations based on their
multilocus genotypes. STRUCTURE assumes that loci within each sample are not in linkage
disequilibrium and in Hardy-Weinberg equilibrium. Loci that did not meet theses requirements
were removed from the analysis to meet this assumption. The program was run for 1.2x105
iterations with the first 2x104 iterations discarded as burn-in for each probable value of k.
Stationarity of the Markov chain before sampling was confirmed by viewing graphs of ln
likelihood values plotted against iterations. Variance in ln likelihood values from 5 repetitions
run on each value of K was used to calculate the parameter ∆k (Evanno et al. 2005). This
parameter is used for determining the number of genetically homogeneous clusters at the highest
level of hierarchical population structuring Evanno et al. (2005). Because the ∆k parameter
represents the uppermost level of population structuring, there can be substantial sub-structuring
of individuals within these initial groupings (Evanno et al. 2005). We iteratively examined
clusters for further sub-structuring until calculation of ∆k revealed no further population
structuring.
To test for an isolation-by-distance correlation between genetic distance and geographic
distance we used Mantel tests conducted in ARLEQUIN (v3.01, Excoffier et al. 2005). A matrix of
pairwise genetic distances using FST values was compared to a matrix of pairwise geographic
107
distances calculated either as straight-line topographic distance between localities or minimum
stream distance between localities. Topographic distance between localities is measure of
straight distance between localities that also accounts for changes in elevation. Minimum stream
distance was measured as the shortest stream distance between localities and does not allow for
any overland travel. The software ArcGis 8.2 (ESRI) was used to measure pairwise topographic
distances from a digital elevation model of the area and minimum stream distances from digital
stream map of the area. Significance of correlations was determined through 100,000 random
matrix permutations.
RESULTS
Hardy-Weinberg and linkage disequilibrium — Two loci (D07 and D20) were
significantly out of HWE in D. copei after correcting for multiple comparisons (Appendix 1) and
were removed from the dataset. Locus D17 was in linkage disequilibrium with two other loci
(D17 X D14 and D17 X D08) and also was removed from all analyses. Two loci (D04 and D05)
were significantly out of HWE in D. tenebrosus after correcting for multiple comparisons
(Appendix 2) and were removed from the dataset. No loci were in linkage disequilibrium in D.
tenebrosus after correcting for multiple comparisons. Removal of these loci resulted in a genetic
dataset that satisfied the assumptions of the population assignment program STRUCTURE
(Pritchard et al. 2000).
Population structuring — Global values of θ and 95% confidence intervals indicate a
significantly higher degree of population structuring in D. copei (θ = 0.079 ± 0.013) than in D.
tenebrosus (θ = 0.031 ± 0.008). Pairwise FST values for D. copei were all significantly different
108
from zero (P < 0.001) and ranged from small (0.0106) to moderate (0.1789) levels of divergence.
Pairwise FST values for D. tenebrosus were considerably lower, had a smaller range of -0.0130 to
0.1034, and included some values not significantly different form zero (Table 1).
Graphical results from the program STRUCTURE revealed a large degree of population
substructure across the study area for D. copei (Fig. 2) and no genetic population structure for D.
tenebrosus (Fig. 3). Iterative examination of population clusters assigned all sampled localities
for D. copei as distinct genetic clusters, except for localities 2, 3, 4 and 6 which are in close
proximity of each other (Fig. 1). Analysis of individuals from these localities resulted in the
inability to assign individuals to more than one genetic cluster suggesting genetic admixture at a
spatial scale of approximately 5 km for this low dispersal species. In contrast, individuals of D.
tenebrosus could not be assigned to more than one genetic cluster even at the highest level of
hierarchical population structure, indicating genetic homogeneity across the entire study area, a
maximum overland distance of 21.1 km, for this high dispersal species.
Isolation by distance — Results of Mantel tests indicated no significant correlations of
genetic distances for D. tenebrosus with topographic (r = 0.049, P = 0.43) and stream distances
(r = -0.025, P = 0.57). For D. copei, genetic distances were strongly correlated with both stream
(r = 0.678, P < 0.01) and topographic distances (r = 0.462, P < 0.01).
DISCUSSION
Based on differences in life history traits that affect dispersal ability, we predicted
contrasting patterns of population structuring for two species of congeneric salamanders within a
common environment. Results are in agreement with these predictions and demonstrate how
109
differences in life history may subsequently influence intraspecific population genetic structure.
These results also are consistent with the pattern seen between closely related metamorphosing
and non-metamorphosing populations of ambystomatid salamanders (Shaffer 1984). Such
comparative studies are sometimes confounded by contrasting allopatric species or
phylogenetically distant species (see Dawson et al. 2002 and references therein). By controlling
for these variables we provide evidence of differing levels of population level gene flow due to
differential dispersal capabilities. The genetic structuring patterns observed at the population
level help explain the phylogeographic patterns for each species. The low dispersal salamander,
D. copei, not only displays high levels of genetic structuring at a local scale, but also displays
strong phylogeographic structure across it distribution (Chapter 3) while the high dispersal
species, D. tenebrosus, displays low levels of population level genetic structuring and also
shallow phylogeographic structuring and evidence of recent post-glacial range expansion
(Chapter 2). This relationship between dispersal ability, population level genetic structuring, and
phylogeographic patterns supports the null hypothesis tested that life history traits can influence
the distribution and a species’ distribution and its genetic patterns within that distribution.
Linking Dispersal and Population Structure — The low frequency of terrestrial adults in
D. copei likely limits dispersal among localities and results in the substantial degree of genetic
population structure that we found. An overall estimate of genetic differentiation within the study
area indicates a moderate level of genetic differentiation (θ =0.079). All pairwise FST values were
also significantly different from zero (Table 1) indicating strong population structuring in this
species. However, 12 polymorphic loci may provide enough statistical power to differentiate
even small differences from zero. Individuals in the study area could be assigned into 8 distinct
110
genetic groups with many of these groups associated with a single sampled locality (Figure 2).
The determination of these genetic groupings is not based on FST values but rather on the
assignment of multilocus genotypes into clusters that minimize deviation from Hardy-Weinberg
equilibrium and linkage disequilibrium, thereby providing additional evidence that this species
high structured genetically. This species also displays a significant pattern of isolation-by-
distance as indicated by significant correlation of FST values with both stream and topographic
distances. The significant correlation of genetic differentiation with both measures of physical
distance may be due in part to the large pairwise genetic differences that, when analyzed in
Mantel tests, results in a significant correlations regardless of how physical distance is measured
between the sites. Overall, the high degree of genetic population structure in this species is
consistent with the prediction that low dispersal organisms should display high degrees of
genetic population structure (Bohonak 1999).
In contrast, regular metamorphosis of D. tenebrosus into terrestrial adults likely
diminishes genetic structuring of populations. However, our analyses did not reveal that gene
flow was significantly correlated with either topographic or stream distance. This result is likely
due to the fact that D. tenebrosus displayed no genetic structure over the study area, so
significant isolation-by-distance patterns could not be detected at the geographic scale of the
study. This species displays a significantly lower overall estimate of population structuring (θ =
0.031) than the low dispersing D. copei. Some pairwise FST values were not significantly
different from zero (Table 1) indicating high gene flow in the species. However, only seven loci
were used for this species, which likely reduced the statistical power of detecting small
differences from zero. Nevertheless, our inability to assign individuals into discrete genetic
111
clusters suggests high a degree of genetic admixture within the study area. The pattern of
population structuring seen within this species is also concordant with the paradigm that high
dispersal organisms displaying low degrees of genetic population structure.
Linking Population Structure and Phylogeographic Patterns — The different dispersal
abilities of these two salamanders, and the consequent patterns of population structuring, can be
directly linked to the geographic distributions and phylogeographic patterns observed within
each species. In D. copei, low dispersal potential likely is responsible for its small and
fragmented geographic range within the Pacific Northwest. The species is restricted to three
disjunct mountain ranges (Cascade Mtns, Olympic Mtns, and Willapa Hills) (Chapter 3, page 96)
in Washington state and Oregon, and there are few known localities between these regions
(Petranka 1998). Because D. copei can be found in close sympatry with D. tenebrosus where the
two co-occur, it is presumed that D. copei could utilize the same stream habitats as the more
expansive D. tenebrosus. However, low gene flow likely prevents rapid colonization of streams
outside its current distribution and precludes a distribution of the same extent as its congener. As
a result, the low dispersal ability in this species is likely responsible for its small geographic
distribution. Low dispersal also means greater susceptibility to the genetic structuring of
vicariant events. This species displays high levels of phylogeographic structure across its small
range and has four well supported lineages corresponding to small sections of its (Chapter 3,
page 97; Steele and Storfer, submitted). One such lineage is restricted to several tributaries
draining into the Columbia River, exemplifying the consequences of low dispersal ability on
phylogeographic structure. The lack of connection to other watersheds appears to have
sufficiently hindered any gene flow with this population and resulted in monophyly. Other
112
lineages are restricted to portions of its range on either side of the Columbia River in the Cascade
Mountains. The remaining coastal lineage is separated from the Cascade populations by
approximately 60 km of lowlands that experienced a xerification during the Pleistocene (Baker,
1983; Barnosky et al., 1987). Phylogeographic analyses reveal no shared haplotypes among these
four regions despite their close, and sometimes parapatric, proximity. This strong
phylogeographic pattern across a small geographic distribution is consistent with the prediction
of strong genetic structuring in a low dispersal species when patterns are examined at a larger
geographic scale.
In contrast, the high dispersal ability of D. tenebrosus likely facilitated its large
continuous range in the Pacific Northwest. This species is found from the US/Canada border to
northern California and has the largest distribution in the genus. Phylogeographic analyses reveal
just two well supported lineages forming a distinct north-south split across its range (Steele and
Storfer 2006, Chapter 2 herein). Northern populations, which span from the Columbia River to
the Fraser River in British Columbia, are characterized by low levels of haplotype diversity and
shallow phylogeographic structure. This pattern likely was caused by post-glacial range
expansion from a small Pleistocene refugium located in or near the Columbia River valley
(Steele and Storfer 2006). The high dispersal ability of this species made possible the northward
expansion of approximately 400 km to the Fraser River, which appears to limit further dispersal.
The southern populations have higher haplotype diversity, perhaps indicative of a larger
Pleistocene population, but correlation of haplotypes with geography also reveal recent range
expansion (Steele and Storfer 2006). The species expanded approximately 450 km northward
from a southern Pleistocene refugium along the Oregon/California border until it reached the
113
Columbia River, which also appears to limit further dispersal. A general historical scenario for
D. tenebrosus is that Pleistocene glaciation restricted a large ancestral range into two refugia
from which it recently expanded into its current distribution. Long range dispersal from the two
refugia was likely facilitated by high dispersal ability. The relatively few phylogeographic
lineages for an organism with a large geographic distribution is also consistent with the
prediction of limited genetic structuring in a high dispersal species when patterns are examined
at a larger geographic scale.
CONCLUSION
Dispersal ability has been recognized as a driving force in shaping genetic population
structure in a variety of organisms (Whiteley et al. 2004, King and Lawson 2001, Dawson et al.
2002, Doherty et al. 1995). A general relationship is that populations of high dispersal species
have less genetic structure than low dispersal species (Bohonak 1999). Scaling up to the species
level, it should follow, then, that low dispersal organisms should inherently have limited
geographic distributions and high levels of regional phylogeographic structuring due to the
absence of homogenizing gene flow. High dispersal species should have larger and more
continuous geographic distributions and comparatively lower levels of phylogeographic
structuring. The results of this comparative study are consistent with these predictions and show
how variation in the life history trait of dispersal ability in these closely related sympatric species
of salamander can be tied to not only patterns of population structuring, but also
phylogeographic patterns.
114
Literature Cited
Bohonak, A.J. 1999. Dispersal, Gene Flow, and Population Structure. The Quarterly
Review of Biology. 74:21-45.
Baker, R.G., 1983. Holocene vegetational history of the western United States. In:
Wright, Jr., H.E. (Ed.), Late Quaternary environments of the United States. Vol. 2.
University of Minnesota Press, Minneapolis, pp. 109–127.
Barnosky, C.W., 1981. A record of late Quaternary vegetation from Davis Lake, southern
Puget Lowland, Washington: Quatern. Res. 16, 221–239.
Dawson MN, Louie KD, Barlow M, Jacobs DK and Swift CC. 2002. Comparative
phylogeography of sympatric sister species, Clevelandia ios and Eucyclogobius
newberryi (Teleostei, Gobiidae), across the California Transition Zone. Molecular
Ecology 11: 1065–1075.
Doherty, PJ, Planes S, Mather P. 1995. Gene Flow and Larval Duration in Seven Species
of Fish from the Great Barrier Reef. Ecology 76:2373-2391.
Evanno, G., S. Regnaut, J. Goudet. 2005. Detecting the number of clusters of individuals
using the software structure: a simulation study. Mol. Ecol. 14:2611-2620.
Excoffier, L. G. Laval, and S. Schneider (2005) Arlequin ver. 3.0: An integrated software
package for population genetics data analysis. Evolutionary Bioinformatics Online 1:47-
50.
Goudet, J. 2001. FSTAT, a program to estimate and test gene diversities and fixation
indices (version 2.9.3). Available from http://www.unil.ch/izea/softwares/fstat.html.
Updated from Goudet (1995).
115
Hellberg, ME. 1996. Dependence of gene flow on geographic distance in two solitary
corals with different larval dispersal capabilities. Evolution 50: 1167-1175.
King RB, Lawson R. 2001. Patterns of population subdivision and gene flow in three
sympatric natricine snakes. Copeia, 2001, 602–614
Leonard, W.P., H.A. Brown, L.L.C. Jones, K.R. McAllister, and R.M. Storm. 1993.
Amphibians of Washington and Oregon. Seattle Audubon Society, Seattle.
Lomolino, M.V., B.R. Riddle, and J.H. Brown. 2006 Biogeography. 3rd edition. Sinauer
Associates, Inc.
Newman. 1992. Adaptive plasticity in amphibian metamorphosis. Bioscience 42:671-
678.
Nussbaum, R.A. 1970. Dicamptodon copei, n. sp., from the Pacific Northwest, U.S.A.
(Amphibia: Caudata: Ambystomatidae). Copeia. 1970: 506–514.
Nussbaum, R.A., 1976. Geographic variation and systematics of salamanders of the
genus Dicamptodon Strauch (Ambystomatidae). Miscellaneous Publications No.
Petranka, J. W. 1998. Salamander of the United States and Canada. Smithsonian
Institution Press, Washington, DC.
Pritchard, JK, Stephens, M, Donnelly, P. 2000. Inference of population structure from
multilocus genotype data. Genetics 155:945-959.
Raymond M., Rousset.1995 genepop (version1.2): population genetics software for exact
test and ecumenicism. Journal of Heredity 86:248-249.
Roff D. A. 1992. The evolution of life histories: theory and analysis. Chapman and Hall,
New York.
116
117
Shaffer, HB. 1984. Evolution in a paedomorphic lineage. 1. An electrophoretic analysis
of the Mexican ambystomatid salamanders. Evolution 38:1194-1206.
Stearns S. C. 1992. The evolution of life histories. Oxford Univ. Press, New York.
Steele CA, Carstens BC, Storfer A, and Sullivan J. 2005 .Testing hypotheses of
speciation timing in Dicamptodon copei and Dicamptodon aterrimus (Caudata:
Dicamptodontidae). Mol Phylo and Evol.
Steele, CA. and Storfer A. 2006. Coalescent-based hypothesis testing supports multiple
Pleistocene refugia in the Pacific Northwest for the Pacific Giant Salamander
(Dicamptodon tenebrosus). Molecular Ecology.
Steele, CA. and Storfer A. Evidence for phylogeographic incongruence of codistributed
species based on small differences in geographic distribution. Mol Ecol and Phylogent.
Submitted.
Steele, CA, Giordano A, Storfer A. Polymorphic microsatellite loci for the salamander
genus Dicamptodon. Submitted.
Weir, B.S., Cockerham, C.C., 1984. Estimating F-statistics for the analysis of population
structure. Evolution 38:1358–1370.
Whiteley, A.R., Spruell, P., Allendorf, F.W. 2004. Ecological and life history
characteristics predict population divergence of salmondis in the same landscape. Mol
Ecol 13:3675-3688.
8
11
1 2 3 4 5 6 7 8 9 10 111 — 0.0285 0.0338 0.0718 0.0471 0.0217 0.0796 0.056 0.0714 0.0769 0.1722 0.0037 — 0.0102
0.0469 0.0351 0.0180
0.0425
0.0385 0.0498 0.0418 0.13133 0.0574 0.0361 — 0.0302 0.0396 0.013 0.052 0.0442 0.0522 0.0642 0.15074 0.0411 -0.0211 0.0577 — 0.0623
0.0315
0.0763 0.0982 0.0969 0.1061 0.1958
5 0.0345 -0.018 0.0448 0.0126 — 0.045 0.0521 0.0744 0.0766 0.0772 0.18136 0.0336 -0.0059 0.0229
0.0382 0.0186 — 0.0522 0.0555 0.0706 0.0704 0.1585
7 0.0124 -0.0101 0.013 0.0294 0.0146 0.0123 — 0.0542 0.0582 0.0643 0.16768 0.0296 0.0042 0.0464 0.0328 0.0102 0.0178 0.0173 — 0.0357 0.0665 0.16089 0.0394 -0.0284 0.0801 0.0189 0.0145 0.035 0.0282 0.0298 — 0.0728
0.1952
10 0.0267 -0.0079 0.0184 0.0061 0.004 0.0158 0.0185 0.0214 0.0224 — 0.1143 11 0.0363 0.0006 0.0666 0.0156 0.0217 0.0415 0.0468 0.0398 0.0419 0.0248 —
Table 1. Below diagonal are pairwise FST values for D. tenebrosus, above the diagonal are values for D. copei. Values significantly different from zero are indicated in bold. Locality numbers (1—11) correspond to those in Figure 1.
118
Pacific Ocean
(Canada) (USA)
BC
OR
WA
5 Km Oregon
Washington
Columbia River
Fig. 1. Map showing location of study area. Inset map shows sampled localities.
119
1.00
0.80
0.60
0.40
0.20
b.)
1 2 3 4 5 6 7 8 9 10 11
1.00
0.80
0.60
0.40
0.20
1.00
0.80
0.60
0.40
0.20 2 3 4 6
1 2 3 4 5 6
1.00
0.80
0.60
0.40
0.20
c.)
7 8 9
1.00
0.80
0.60
0.40
0.20
1 2 3 4 5 6 7 8 9 10 11
1.00
0.80
0.60
0.40
0.20
a.)
d.)
Fig. 2. Graphical output from the program STRUCTURE for D. copei, a species with low dispersal ability, showing a high degree of population structuring. Number of clusters in each analysis is based on calculation of ∆K. Each column represents a sampled individual. Colored proportions of columns represent probability of assignment to different clusters. Values along X-axis represent locality number, values along Y-axis represent assignment probability to different clusters. (a.) initial clustering of all 11 localities into two groups: localities 1-6 and localities 10-11. (b.) subsequent analysis of initial clusters reveals substructure in localities 1-6 and localities 7-9. Localities 10 and 11 cluster independently. (c.) analysis at the third level of population structuring reveals localities 1 and 5 to be distinct while localities 2,3,4, and 6 display genetic admixture. Localities 7-9 form independent clusters. Analysis of localities 2,3,4 and 6 confirm a lack of further substructure indicated by no clear value for delta K and a graphical display showing genetic admixture at minimum value of K = 2.
120
121
Fig. 3. Graphical output for D. tenebrosus, a species with high dispersal ability, showing no population structuring in sampled localities in the study area.
1.00
0.80
0.60
0.40
0.20
0.00
1 2 3 4 5 6 7 8 9 10 11
Appendix 1. Summary stats for D. copei. Loci with * indicate it was removed from analysis due to linkage disequilibrium or out of HWE.
Locality 1 2 3 4 5 6 7 8 9 10 11 Total
No. of samples 30 24 29 10 22 16 30 29 28 29 27 274
Locus D04 No. of alleles 12 9 10 8 10 8 8 10 10 11 10 25 Fis -0.02
0.01 0.05 0.17 0.05 -0.01 0.07 0.06 0.05 -0.08 -0.03 He 25.63
20.24
24.24
7.12
19.00
13.81
25.73
23.34
23.11
25.11
16.47
Ho 26 20 23 6 18 14 24 22 22 27 17HWE
0.72
0.68
0.10
0.13
0.15
0.59
0.02
0.53
0.15
0.56
0.51
0.11
D08 No. of alleles
5 5 9 6 7 6 6 8 6 10 8 12 Fis 0.00 -0.04 0.01 -0.10
0.13 0.17 0.02 0.08 0.01 0.14 -0.04
He 21.03
18.30
24.11
7.29
17.19
13.13
22.51
20.71
20.15
24.28
18.32 Ho 21 19 24 8 15 11 22 19 20 21 19
HWE
0.14
0.85
0.56
0.30
0.33
0.49
0.86
0.29
0.81
0.05
0.52
0.47
D13
No. of alleles
10 8 12 5 9 9 11 11 8 10 8 16 Fis 0.15 0.09 -0.10 0.23 0.00 0.04 0.03 0.11 0.09 0.12 0.00
He 24.53
16.51
20.94
5.11
18.00
13.55
25.67
22.47
21.89
25.07
20.92 Ho 21 15 23 4 18 13 25 20 20 2 21
HWE
0.06
0.22
0.68
0.05
0.81
0.16
0.38
0.13
0.08
0.86
0.73
0.09
D14
No. of alleles
9 7 8 7 7 7 10 9 6 8 9 12 Fis -0.09 -0.02 0.11 -0.05
0.01 -0.08 -0.02 0.03 0.08 0.00 -0.13
He 25.80 19.62 22.38
8.63
17.21
13.00
25.54
15.47
19.53
20.98
20.45 Ho 28 20 20 9 17 14 26 15 18 21 23
HWE 0.99 0.81 0.81 0.30 0.71 0.95 0.32 0.10 0.23 0.71 0.90 0.87
120
122
D18
No. of alleles
7 8 8 6 9 8 7 8 7 11 6 16 Fis 0.11 -0.02 0.09 0.03 -0.17 -0.02 -0.04 0.09 -0.07 -0.02 0.25
He 23.67
19.64
25.23
8.26
17.95
13.71
24.05
20.80
23.47
25.61
17.32 Ho 21 20 23 8 21 14 25 19 25 26 13
HWE
0.46
0.19
0.24
0.06
0.86
0.96
0.19
0.60
0.29
0.73
0.03
0.18
D22
No. of alleles
8 9 9 7 6 7 7 8 7 8 5 10 Fis -0.17 0.04 0.06 0.27 -0.01 0.05 0.03 0.12 -0.06 0.30 0.29
He 23.93
18.67
23.38
6.71
16.88
13.68
24.73
23.87
19.93
22.62
13.96 Ho 28 18 22 5 17 13 24 21 21 16 10
HWE
0.13
0.44
0.47
0.01
0.17
0.33
0.93
0.37
0.46
0.03
0.03
0.01
D25
No. of alleles
6 7 7 5 4 6 4 6 3 7 5 9 Fis 0.17 -0.10 -0.07 0.12 0.15 -0.10 -0.02 0.05 0.05 0.05 0.01
He 21.66
18.17
19.65
6.74
14.02
11.84
11.78
17.92
8.41
17.89
15.16 Ho 18 20 21 6 12 13 12 17 8 17 15
HWE
0.17
0.39
0.96
0.27
0.35
0.76
1.00
0.62
1.00
0.36
0.50
0.86
D06
No. of alleles
7 7 6 4 8 5 7 6 5 8 8 10 Fis -0.22 -0.02 -0.04 0.04 0.01 -0.16 -0.06 0.07 -0.01 0.07 0.06
He 21.42
15.76
20.24
6.26
17.16
11.28
20.75
22.44
17.82
22.64
18.02 Ho 26 16 21 6 17 13 22 21 18 21 17
HWE
0.14
0.81
0.26
1.00
0.39
0.31
0.92
0.87
0.33
0.60
0.51
0.80
D07*
No. of alleles
10 10 10 6 10 10 9 16 14 11 10 22 Fis 0.22 -0.03 0.07 0.43 -0.07 0.10 0.86 0.40 0.28 0.54 0.34
He 24.21 18.41 21.47 6.84
19.58
13.31
21.09
24.89
24.69
21.47
21.19 Ho 19 19 20 4 21 12 3 15 18 10 14
121
123
HWE
0.01
0.24
0.11
0.02
0.08
0.29
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
D15
No. of alleles
9 8 12 7 10 8 7 10 7 8 7 15 Fis 0.07 0.08 0.03 -0.01
-0.02 0.10 -0.02 -0.11 -0.09 0.02 -0.10
He 25.78
16.21
24.71
7.95
17.58
12.17
22.53
24.44
21.05
22.40
16.34 Ho 24 15 24 8 18 11 23 27 23 22 18
HWE
0.33
0.29
0.64
0.30
0.21
0.03
0.94
0.47
0.90
0.50
1.00
0.51
D23
No. of alleles
7 6 7 5 7 5 7 7 7 10 8 15 Fis -0.11 -0.07 -0.03 0.09 -0.11 -0.08 0.30 0.17 0.24 -0.07 0.00
He 22.67
16.85
18.43
6.53
15.35
10.21
24.25
21.69
22.31
23.46
19.94 Ho 25 18 19 6 17 11 17 18 17 25 20
HWE
0.35
0.46
0.07
0.67
0.68
0.35
0.05
0.21
0.00
0.42
0.59
0.02
D05
No. of alleles
11 9 11 6 8 7 7 7 7 9 10 18 Fis -0.01 -0.15 0.02 -0.21
0.16 -0.08
0.07 0.08 -0.11 0.02 0.11
He 21.84
15.71
24.45
6.71
14.23
6.53
23.49
21.65
18.98
21.47
20.22 Ho 22 18 24 8 12 7 25 20 21 21 18
HWE
0.12
0.16
0.84
0.04
0.32
0.90
0.07
0.59
0.36
0.34
0.08
0.07
D17*
No. of alleles
5 7 9 7 10 7 8 9 8 9 6 13 Fis 0.01 0.05 0.16 -0.17
-0.13 0.03 -0.06 0.14 -0.06 0.08 -0.10
He 18.10
15.76
22.49
7.76
15.09
11.28
23.68
24.28
19.85
23.96
18.29 Ho 18 15 19 9 17 11 25 21 21 22 20
HWE
0.63
0.41
0.23
0.84
0.34
0.35
0.79
0.03
0.90
0.70
0.50
0.59
D20*
No. of alleles
14 13 17 10 11 14 19 19 21 20 8 32 Fis -0.04 0.04 -0.05 -0.08 0.05 0.09 0.03 0.09 -0.05 0.15 0.03
He 24.06 14.55 20.91 7.47 11.56 13.15 26.88 27.44 26.65 24.62 18.53
122
124
Ho 25 14 22 8 11 12 26 25 28 21 18HWE
0.44
0.59
0.92
0.58
0.17
0.03
0.05
<0.01
0.84
0.04
0.86
<0.01
D24
No. of alleles
5 4 6 5 4 5 4 5 4 10 9 15 Fis -0.05 -0.17 -0.14 0.12 -0.04 -0.06 -0.03 -0.14 -0.15 -0.03 0.02
He 18.12
13.71
20.16
6.76
12.54
10.41
19.51
17.64
19.13
22.34
20.38 Ho 19 16 23 6 13 11 20 20 22 23 20
HWE 0.52 0.84 0.54 0.08 0.81 0.77 0.40 0.83 0.74 0.84 0.52 0.93
123
125
Appendix 2. Summary stats for D. tenebrosus. Loci with * indicate it was removed from analysis due to linkage disequilibrium or out of HWE.
Locality 1 2 3 4 5 6 7 8 9 10 11 Total
No. of samples 18 8 24 30 29 30 29 31 9 20 22 250
Locus
D04* No. of alleles 4 3 5 5 4 4 5 7 3 8 3 10
Fis 0.08
0.39 0.77 0.06 0.24 0.24 0.53 0.26 0.45 0.13 0.48 He 7.56
3.18
8.37
9.51 14.37 14.44
18.87
18.91
3.53
11.48 13.39 Ho 7 2 2 9 11 11 9 14 2 10 7
HWE 0.86 0.52 <0.01 0.61 0.07 0.13 <0.01 0.06 0.12 0.24 <0.01 <0.01
D13
No. of alleles
9 6 9 8 7 9 11 10 8 10 7 17
Fis 0.00 -0.25 0.04 0.12 0.03 -0.02 -0.02 -0.03 0.02 -0.08 -0.03He 14.97
4.91
16.62
21.55
21.55
24.46
26.37
25.18
8.12
16.70
17.47 Ho 15 6 16 19 21 25 27 26 8 18 18
HWE 0.74 0.95 0.57 0.24 0.32 0.61 0.63 0.56 0.04 0.34 0.75 0.65
D14
No. of alleles
8 4 5 6 6 5 6 8 4 6 7 11
Fis 0.05 -0.09 -0.02 -0.29 -0.15 0.04 -0.24 -0.03 0.06 -0.21 -0.07He 14.69
4.64
13.72
17.88
18.29
20.90
22.60
18.38
6.35
12.52
16.90 Ho 14 5 14 23 21 20 28 19 6 15 18
HWE 0.46 1.00 0.51 0.08 0.31 0.11 0.16 0.13 0.44 0.08 0.37 0.09
D25 No. of alleles 3 1
4 1 4 3 3 3 4 5 2 7
124
126
Fis -0.06 NA -0.17 NA -0.08 -0.15 -0.07 0.37 0.30 0.10 -0.08He 3.77
NA 7.73 NA 5.57 8.75 5.61 4.72 4.24
7.73 3.72
Ho 4 NA 9 NA 6 10 6 3 3 7 4HWE
1.00
NA 1.00
NA 1.00
1.00
1.00
0.17
0.53
0.03
1.00
0.87
D18
No. of alleles 7 5 6 8 8 6 6 7 5 8 5 9
Fis -0.06 -0.24 -0.10 -0.06 -0.10 -0.10 0.00 -0.19 -0.09 -0.13 0.18He 14.21
5.77
18.31
24.51
23.75
22.75
20.95
22.78
7.35
15.14
16.91 Ho 15 7 20 26 26 25 21 27 8 17 14
HWE 0.34 0.82 0.01 0.03 0.65 0.20 0.24 0.10 0.11 0.94 0.07 0.01
D06
No. of alleles
3 3 4 3 4 4 5 6 3 4 5 8
Fis -0.28 -0.25 0.07 0.10 0.22 0.28 0.11 0.14 0.02 0.11 -0.01He 7.89 4.08
11.77
15.54
16.51
9.71
12.33
16.22
4.06
11.18
14.79 Ho 10 5 11 14 13 7 11 14 4 10 15
HWE
0.73
0.63
0.87
0.57
0.34
0.00
0.36
0.07
0.25
0.11
0.29
0.04
D24
No. of alleles
2 2 3 3 3 3 2 3 2 2 2 5
Fis NA 1.00 -0.09 -0.08 0.25 -0.12 -0.04 -0.01 -0.07 -0.12 -0.06He 1.00
1.87
5.53
5.58 9.25 8.08 2.89
1.90
1.88
4.49 2.85 Ho 1 0 6 6 7 9 3 2 2 5 3
HWE NA 0.07 1.00 1.00 0.32 1.00 1.00 1.00 1.00 1.00 1.00 0.99
D05*
No. of alleles
4 2 5 7 3 6 3 5 6 5 7
10 Fis 0.33 1.00 0.51 0.80 1.00 0.28 0.83 0.69 0.83 0.90 0.71
He 5.80 3.27 5.96 14.84
8.31 13.81
5.59
12.61
5.36
9.52 13.52 Ho 4 0 3 3 0 10 1 1 1 1 4
125
127
HWE
0.01
0.02
0.02
<0.01
<0.01
0.10
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
D17
No. of alleles
6 3 4 6 5 6 3 4 2 5 4 7
Fis 0.17 -0.67 -0.16 0.02 -0.17 -0.13 0.39 0.15 -0.33 0.40 -0.28HeHo
12.0010
3.225
11.2313
19.3419
20.5124
20.4423
13.008
16.5114
3.825
11.457
13.3817
HWE 0.12 0.17 0.87 0.36 0.27 0.33 0.02 0.15 1.00 0.02 0.40 0.03
126
128