Distribution and Phylogeography of the Canyon Treefrog (Hyla arenicolor) in the Rincon Mountains, Arizona

Full text

106 SONORAN HERPETOLOGIST 21 (10) 2008
Distribution and Phylogeography of the Canyon Treefrog (Hyla
arenicolor) in the Rincon Mountains, Arizona
Kevin Bonine1, Taylor Edwards2, Christine Schirmer3, Jamia Fillinger3, and Kris Ratzla3
1University of Arizona, Department of Ecology and Evolutionary Biology, Tucson, Arizona, USA. kebonine@u.arizona.edu (corresponding author)
2University of Arizona, Human Origins Genotyping Laboratory, Tucson, Arizona, USA
3University of Arizona, Arizona Research Laboratories, Tucson, Arizona, USA
Past studies employing population genetics
have shown that many organisms are structured
into phylogenetic units that often correspond
to geographical regions (Avise et al. 1987). This
phylogeography can be the result of many different
biological processes including vicariance, migration,
population expansion, and population bottlenecks
(Knowles et al. 2002). We used mitochondrial DNA
to look for correlated patterns of genetic relatedness
and geographic distribution in Canyon Treefrogs
(Hyla arenicolor) in the Rincon Mountains east of
Tucson, Arizona. By examining the species’ population
genetic structure among drainages, we hoped to infer
historic dispersal patterns among Canyon Treefrog
populations.
Barber (1999a, b) also used mitochondrial DNA
sequence data to help determine range-wide patterns
of phylogeography and gene ow for the Canyon
Treefrog. The phylogenetic analysis of mitochondrial
haplotypes revealed a compelling structure of
United States populations with 50 unique haplotypes
(maternal lineages; Barber 1999a). Phylogenetic
analyses revealed three distinct population regions
in Arizona, each of which had deeply divergent
mitochondrial DNA lineages which corresponded to
nonoverlapping geographical regions (Fig. 1). Barber
suggested that populations of Hyla arenicolor that
inhabit montane habitats of the southwestern desert
are probably geographic isolates. In this study, we
looked for phylogeographic patterns within one of
these montane populations.
We also compared the genetic structure of Canyon
Treefrogs from the Rincon Mountains to that of the
Lowland Leopard Frog (Rana yavapaiensis, Goldberg
et al. 2004), which share their habitat, to determine if
these two sympatric species exhibit similar patterns of
dispersal.
Materials and Methods
We collected Hyla arenicolor adults by hand from
pools in the Rincon Mountains of Arizona (Fig. 2). We
collected outgroup samples from 16 km (9.9 mi) away
at Sabino Canyon in the Santa Catalina Mountains.
Using buccal swabs, we took samples from each frog
in spring and summer 2006. To avoid sampling the
same individual more than once, we used implantable
elastomeric alphanumeric tags (Northwest Marine
Technology, e.g., Daniel et al. 2006). These tags
are preferred over other methods such as toe clips,
especially for a species like the Canyon Treefrog that
spends much of its time on vertical rock surfaces
We conducted all genetic methods and analysis at
the GATC (Genomic Analysis and Technology Core,
Arizona Research Laboratories) at the University of
Arizona. We focused our efforts on mitochondrial
DNA (mtDNA) because, not only does it have a
rapid pace of evolution and extensive intraspecic
polymorphisms (Avise 1987), it is inherited maternally
without recombination, which allows for tracing single
lineages back in time. We chose the cytochrome B
region of the mtDNA using two of the same PCR
primer pairs used in the Barber study (MVZ15-
MVZ34 and MVZ55-MVZ36; 1999a). We sequenced
amplicons in both directions for each primer pair and
concatenated the two regions for analysis.
We identied genetic types (or haplotypes) for all
our samples and conducted several different analyses.
We reconstructed an evolutionary network to generate
hypothesized relationships among haplotypes.
We also looked at the amount of diversity (π) and
polymorphism (θ) in the mtDNA sequences for each
population and for the entire sample set. We tested
for processes that reduce or increase heterozygosity
in a population using the statistic Tajima’s D. Under
Figure 1. Geographic locations and clade anity of Hyla
arenicolor in Arizona (modied from Barber 1999a).
We used
mitochondrial
DNA to look
for correlated
patterns
of genetic
relatedness
and geographic
distribution in
Canyon Treefrogs
(Hyla arenicolor)
in the Rincon
Mountains east of
Tucson, Arizona.

SONORAN HERPETOLOGIST 21 (10) 2008 107
neutrality, in a panmictic, innitely large population, π
and θ should be equal and the Tajima’s D coefcient,
D, is zero. A positive value for D indicates an excess
of heterozygosity, resulting from processes such as
a reduction in population size (bottleneck), long-
term population subdivision, or balancing selection.
A negative value of D indicates a reduction of
heterozygosity that could be the result of an expansion
(increase in number of individuals in the population)
event, positive directional selection, or the presence of
weakly deleterious alleles (Tajima 1989).
We estimated how genetic variation is partitioned
among sample sites within a region and among
individuals within sample sites using AMOVA (analysis
of molecular variance). To assess correlations between
genetic and geographic distances among sample
groups, we performed a Mantel test. Populations for
the Mantel test were dened two ways, by collection
site or by drainage, in an attempt to compare potential
relationships between genetic and geographic distances
that may differ dependent on how the populations
are dened. The collection-site model utilized six
populations while the drainage-based model utilized
seven populations. Each of these sets of populations
were then analyzed twice using both linear and along-
drainage geographic distances.
Results
We sequenced 886 base pairs of cytochrome B
mtDNA, 320 bp using the MVZ 34 primer and 566
bp using the MVZ 36 primer, from 67 specimens
and identied 15 single nucleotide polymorphisms
that gave rise to 9 unique haplotypes distributed
among the seven collection sites (Table 1). Most of
the haplogroups were “closely related”, with only one
or two differences among them, but HAP 2 showed
greater genetic divergence and was seven nucleotide
differences from its closest relation (Fig. 3).
We observed nucleotide diversity, π = 0.00203;
polymorphism, θ = 0.00355. Tajima’s D = -1.24804 (P
> 0.10). The AMOVA showed genetic differentiation
among site-based populations (Fst = 0.07335, P <
0.03). Among drainage-based populations, AMOVA
again showed even higher genetic differentiation (Fst
= 0.14314, P < 0.001). Because Fst varies between
0 and 1 and measures the differentiation among
populations, we can say that for these two mtDNA
regions, an Fst of 0.14 means that 14% of the total
variation is found among sampling sites, or, any given
site contains 86% of the total variation observed.
Mantel tests of each of our four models revealed no
correlations or statistically signicant results.
Discussion
Hyla arenicolor populations in the Rincon Mountains
exhibit genetic structure; however, no geographic
association is apparent in our data set. Some of our
lack of resolution may stem from use of genetic
markers that have changed more slowly than would be
Figure 2. Locations of
Hyla arenicolor genetic
sampling east of Tucson,
Arizona. From left to right,
the dark circles on the map
correspond to Wildhorse
Canyon, Chiminea Canyon
(Pools), Madrona Canyon,
North Rincon Creek,
Rincon Creek, and North
Miller Creek. Samples
were also collected at
Sabino Canyon which is
located several kilometers
northwest of Saguaro
National Park, Rincon
Mountain District.

108 SONORAN HERPETOLOGIST 21 (10) 2008
necessary to capture very recent movement patterns
among populations.
Changes in the Tucson basin in the past 100
years may play a role in our observed results. The
populations we sampled may have been more broadly
connected to other Hyla arenicolor populations in
greater eastern Pima County; but they are no longer
connected because of human-mediated landscape
changes. A much larger population (sensu broader
geographic area) recently (<100 yrs) reduced to
isolated foothill populations in the different mountain
ranges would be consistent with our results. Such
a recent event would likely not have had sufcient
time to be revealed in mtDNA differences across the
landscape. Our negative Tajima’s D value suggests
recent expansion (increase in number of individuals)
across all of our study populations.
Hyla arenicolor populations probably experience
large booms and busts consistent with variation in
water availability. Typical monsoon weather patterns
in the Tucson area drives some of this variability and
may also assist in dispersal of the Canyon Treefrog;
higher water levels and relatively moist environmental
conditions allow them to return to habitat patches to
recolonize as water returns. Thus, each canyon site
historically had equal opportunity to be colonized by
all possible genetic types (haplotypes) in the source
population during productive years. During a bust
year, those genetic types each have equal probability of
being locally extirpated, thus leaving a distribution of
haplotypes unrelated to geographic location. Each time
this pattern occurs, it is like shufing a genetic deck of
cards. This would explain why more distantly related
haplotypes might be found at the same site (e.g., HAP
2 and HAP 3), as opposed to them having evolved at
each specic site. Thus, there is genetic structure, but
it is not tied to geography at the resolution level of our
analysis.
Compared to the Lowland Leopard Frog (Rana
yavapaiensis), a threatened species living in the same
habitat in the Rincon Mountains, we nd that while
there is no strong pattern of geographic structure
in the Canyon Treefrog, Lowland Leopard Frog
populations exhibit greater genetic isolation due to
geographical barriers such as ridges and likely depend
more on drainage systems for gene ow (Goldberg et
al. 2004). Goldberg et al. (2004), using nuclear DNA
microsatellite data, reported extensive phylogenetic
and geographic structure among Rana yavapaiensis
populations in the Rincon Mountains and suggested
that the Lowland Leopard Frog underwent population
bottlenecks in the mid-20th century leading to the
observed isolated local populations.
If at smaller scales Canyon Treefrogs are less
constrained to drainages, they may have an advantage
for recolonizing unoccupied habitat when it
becomes available as compared to species like the
Lowland Leopard Frog. Moreover, Canyon Treefrogs
appear able to clear chytrid fungus (Batrachochytrium
dendrobatidis [Bd]) within a summer (Kim Baker,
unpublished results, 2007), making them much less
susceptible to the population crashes observed in
many Rana species that may be linked to Bd infection.
If Canyon Treefrogs can disperse while still infected
they may be important reservoirs and vectors in the
spread of Bd and would complicate conservation
efforts directed at Lowland Leopard Frogs and other
susceptible species.
Future analyses of Hyla arenicolor using nuclear
DNA microsatellites should allow us to observe
genetic changes more consistent with the timescale
over which these changes are thought to have
occurred. These analyses will allow for a more
informed discussion of Hyla arenicolor dispersal
patterns in the Rincon Mountains of Arizona.
Haplogroup
Sample Site Hap
1
Hap
2
Hap
3
Hap
4
Hap
5
Hap
6
Hap
7
Hap
8
Hap
9Total
Sabino Canyon 11111 5
Madrona Drainage
Chiminea Pools 2 1 2 1 6
Madrona Canyon 3 2 1 1 2 9
Wild Horse-Lower Pools 52 4 11
Rincon Drainage
Rincon Creek 16 4 1 1 22
Rincon North 6 1 2 9
North Miller Creek 3 2 5
TOTAL 34 3 10 2 10 3 2 1 2 67
Table 1. Distribution of 9 haplogroups for 67 Hyla arenicolor specimens from 7 study populations (6 in Saguaro
National Park, Rincon Mountain District, and 1 from Sabino Canyon, Santa Catalina Mountains, Arizona).
If at smaller
scales Canyon
Treefrogs are
less constrained
to drainages,
they may have
an advantage
for recolonizing
unoccupied
habitat when it
becomes available
as compared to
species like the
Lowland Leopard
Frog.

SONORAN HERPETOLOGIST 21 (10) 2008 109
Acknowledgements
This work was supported in part by a Tucson
Herpetological Society Lowe Research Fund Award
to C. Schirmer. We also thank Arizona Research
Laboratories for their continued support of
conservation genetic research.
Literature Cited
Avise, J.C., Arnold J.A., Ball R.M., et al. 1987.
Intraspecic phylogeography; the mitochondrial
DNA bridge between population genetics and
systematics. Annual Review of Ecology and
Systematics 18:489-522.
Barber, P.H. 1999a. Phylogeography of the Canyon
Treefrog, Hyla arenicolor (Cope) based on
mitochondrial DNA sequence data. Molecular
Ecology 8:547-562.
Barber, P.H. 1999b. Patterns of gene ow and
population genetic structure in the Canyon
Treefrog, Hyla arenicolor (Cope). Molecular Ecology
8:563-576.
Daniel, J.A., K.A. Baker, and K.E. Bonine. 2006.
Retention rates of surface and implantable in
marking methods in the Mediterranean House
Gecko (Hemidactylus turcicus), with notes on capture
methods and rates of skin shedding. Herpetological
Review 37:319-321.
Goldberg, C., D.E. Swann, and J.E. Wallace. 2004.
Genetic structure of Lowland Leopard Frog
(Rana yavapaiensis) populations in and near Saguaro
National Park, Arizona. Report for SNP and
Western National Parks Association (Grant 03-11).
Knowles, L.L., and W.P. Maddison. 2002. Statistical
phylogeography. Molecular Ecology 11:2623-2635.
Tajima, F. 1989. Statistical method for testing
the neutral mutation hypothesis by DNA
polymorphism. Genetics 123:585-595.
Figure 3. Evolutionary net-
work or “cladogram” for 9
Hyla arenicolor haplotypes
from 7 study sites east of
Tucson, Arizona. Unbro-
ken linear connections
between haplotypes rep-
resent a single nucleotide
(A,T,C, or G) dierence out
of 886. Thus, HAP 6 and
HAP 9 were dierent at
one site, HAP 6 and HAP 7
were 2 dierent at 2 sites,
and HAP 5 and HAP 2 had
7 nucleotide dierences
between them.