Elevated Genetic Structure in the Coastal Tailed Frog (Ascaphus truei)

© The American Genetic Association. 2013. All rights reserved. For permissions, please email: [email protected].

202

Elevated Genetic Structure in the Coastal Tailed Frog (Ascaphus truei) in Managed Redwood ForestsAndres AguilAr, robert b. douglAs, eric gordon, JAson bAumsteiger, And mAtthew o. goldsworthy

From the School of Natural Sciences & Sierra Nevada Research Institute, University of California, Merced, 5200 North Lake Road, Merced, CA 95344 (Aguilar and Baumsteiger); the Mendocino Redwood Company, 32601 Holquist Lane/PO Box 489, Fort Bragg, CA 95437 (Douglas and Goldsworthy); and the Mendocino Redwood Company, 850 Kunzler Ranch Road/PO Box 996, Ukiah, CA 95482 (Gordon). Matthew Goldsworthy is now at the National Marine Fisheries Service, Northern California Office, 1655 Heindon Road, Arcata, CA 95521.

Address correspondence to Andres Aguilar at the address above, or e-mail: [email protected].

AbstractLandscape alterations have dramatic impacts on the distribution of genetic variation within and among populations and understanding these effects can guide contemporary and future conservation strategies. We initiated a landscape-scale genetic study of the coastal tailed frog (Ascaphus truei) on commercial timberlands within the southern range of the species in Mendocino County (CA, USA). In total, 294 individuals from 13 populations were analyzed at 9 microsatellite loci. None of the sampled populations departed from mutation–drift equilibrium, indicating recent population bottlenecks were not detected in contemporary samples. Fine-scale analysis indicated sampled populations were structured at the watershed level (mean FST = 0.077 and mean G′ST = 0.425). Landscape analyses suggested wet and moist areas may serve as significant corri-dors for gene flow within watersheds in this region (r2 = 0.32–0.54 for moisture-related features). Results indicate populations of frogs may have persisted at this scale through intense periods of timber harvest, making southern range edge populations of coastal tailed frogs resilient to past land use practices.Key words: conservation, landscape, microsatellites, timberlands

A central role of conservation biology is to understand how anthropogenic habitat alteration affects animal popu-lations so that management solutions can be implemented in time to avoid local extinction (Fahrig and Merriam 1985; Cushman 2006; Lindenmayer and Fischer 2006). Habitat loss and fragmentation, associated with a range of land use prac-tices, often produce a continuum of population responses dependent on the frequency and spatial scale of the disturb-ance events and their interaction with species’ life history, distributional patterns, and regional environmental factors (Fischer and Lindenmayer 2007). Although an extreme and direct result of habitat loss is extinction, less extreme forms of habitat alteration can influence populations over several generations in subtler ways not readily discernible by trad-itional demographic methods; these include smaller effect-ive population size, decreased connectivity, lower genetic diversity, and inbreeding depression (Frankham et al. 2002). Conservation genetics research is being increasingly utilized to assess population metrics to improve the conservation

and management of sensitive species associated with com-mercially valuable industries (Manel et al. 2003; Storfer et al. 2007) and simultaneously address fundamental issues of dis-tributional changes and range limits of those species (Eckert et al. 2008).

The coast redwood (Sequoia sempervirens) forests of California occupy a limited geographic range along the Pacific Coast and are closely associated with maritime summer fog (Barbour et al. 2007). Most of these forests have experienced intense commercial logging for more than a century and consist of second- and third-generation trees. The remaining old-growth and late-seral stands, representing about 5% of the original forest, exist primarily in state and federal parks (Noss 2000). Although timber harvest volumes have declined in recent decades, the redwood region continues to be both a significant source of commercial wood products (Stewart 2007) and an area of high conservation interest (Noss 2000). Redwood forests harbor several sensitive terrestrial and aquatic animal species (Noss 2000), including 3 species of

Journal of Heredity 2013:104(2):202–216doi:10.1093/jhered/ess138Advance Access publication January 16, 2013

Downloaded from https://academic.oup.com/jhered/article-abstract/104/2/202/802926by gueston 14 February 2018

mailto:[email protected]

Aguilar et al. • Tailed Frog Population Genetics

203

co-occurring headwater stream amphibians (Petranka 1998; Adams and Pearl 2005). These forests and their associated amphibian populations not only face pressure from continued timber harvest and human encroachment but also from climate change that reduces the extent and duration of coastal fog (Johnstone and Dawson 2010).

The coastal tailed frog (Ascaphus truei) is a headwa-ter stream amphibian distributed from southern British Columbia to northern California. In California, it is listed as a ‘species of special concern’ and reaches its southern range terminus along a narrow band of mostly privately owned coast redwood forest in southern Mendocino County (Jennings and Hayes 1994). Tailed frogs have long lives, have a lengthy larval period, exhibit delayed reproductive matu-rity, and are highly philopatric to natal streams (Metter 1967; Brown 1975; Daugherty and Sheldon 1982a, 1982b; Wallace and Diller 1998; Burkholder and Diller 2007). They utilize cool, fast-flowing streams containing rocky substrates and low amounts of fine sediment for breeding, egg deposition, and larval development (Welsh and Ollivier 1998; Diller and Wallace 1999; Welsh et al. 2000). Both aquatic and terrestrial life stages have a narrow range of air- and water-temperature tolerance levels (Brattstrom 1963; Brown 1975). All these traits suggest a strong coevolutionary link with old-growth forest systems (Welsh 1990).

Concern over the conservation status of the coastal tailed frog has been raised because much of its range coincides with lands managed for commercial timber production (Adams and Pearl 2005; Kroll 2009), and several studies indicate that commercial logging can impact this and other headwater amphibian populations (Bury 1983; Corn and Bury 1989; Welsh 1990; Welsh and Lind 1991; Dupuis and Steventon 1999; Ashton et al. 2006). Timber harvest and associated activities are known to increase fine sediment deposition in streams, which can directly inhibit larval feeding, reduce periphyton growth, and fill interstitial spaces serving as cover (Corn and Bury 1989; Welsh and Ollivier 1998; Stoddard and Hayes 2005; Jackson et al. 2007). Removal of canopy cover and mature trees also reduces the insulative capacity of the forest to maintain cool and moist terrestrial microclimates, as well as suitable aquatic conditions for cold-water-adapted species (Chen et al. 1998; Welsh et al. 2005a). Such habitat alterations may not only depress the abundance of tailed frogs for a period of time (Ashton et al. 2006), but extreme forms of alterations (e.g., clear-cutting and forest conversion) may interact with topography and regional climate to further isolate or extirpate frog populations across entire watersheds (Welsh et al. 2005b; Welsh and Hodgson 2008).

To date, only a few studies have examined population genetic characteristics of tailed frogs on landscapes experi-encing timber harvest. These studies, conducted in British Columbia and Washington, found larval populations showed reduced genetic diversity and increased bottleneck effects associated with clear-cuts (of ~10-year-old trees) compared with unlogged old-growth forests (Wahbe et al. 2005), and significantly reduced gene flow in areas lacking forest cover (Spear and Storfer 2008). Spear and Storfer (2008) fur-ther noted that lag effects from timber harvest may lead to

erroneous conclusions (i.e., timber harvest leads to higher gene flow) without additional analysis. Although these studies corroborate previous observational studies on the potential for timber harvest to negatively affect tailed frog populations, they may not be representative of tailed frog population genetic characteristics in other parts of the spe-cies’ range where timber harvest occurs, especially areas with different disturbance histories, timber harvest practices, cli-mates, and forest types.

We initiated a population genetic study of coastal tailed frogs on privately owned commercial timberlands in the cen-tral part of the redwood zone. This study area has generated strong conservation interest because it has been intensively harvested for decades, lacks any significant old-growth for-est, and is within the southern extent of the tailed frog’s geographic range. Based on these facts, we hypothesized that 1) extant tailed frog populations exhibit genetic patterns associated with small isolated populations (e.g. bottlenecks); 2) genetic connectivity among populations is primarily asso-ciated with vegetation characteristics affected by timber harvest (forest structure and/or canopy cover); and/or alter-natively 3) if tailed frog populations are not genetically iso-lated, then migration corridors tied to other physiographic features may better explain genetic connectivity.

Materials and MethodsStudy Area

Totally, 4 watersheds were sampled in the North Coast physio-graphic province of Mendocino County, California (Figure 1) based on known tailed frog locations from recent distribu-tion surveys. These watersheds consisted of second- and third-generation coast redwood (S. sempervirens) forests, with Douglas-fir (Pseudotsuga mensziesii) and tanoak (Notholithocarpus densiflorus) forming a significant component. Climate is mari-time and marked by cool, mild summers and wet winters. Mean annual precipitation for study sites during 2003–2009 was 119.9 cm (standard deviation [SD] = 30.4, range = 90.8–139.7 cm), with more than 50% occurring from December to February. Mean weekly average temperature and mean weekly maximum temperature of streams during the summer stress period at the locations where samples were collected were 13.8 °C (SD = 0.47, range = 13.3–15.0 °C) and 14.5 °C (SD = 0.54, range = 13.5–15.9 °C), respectively, in the period 2007–2009. Elevation ranges from sea level to 600 m.

Forest structure patterns on the landscape have been heavily influenced by commercial timber harvests during the past 120 years. These timberlands have experienced at least 2 harvest entries and have been shaped by a regimen of clear-cutting and repeated burning that removed most of the large valuable trees (primary forest) and left a significant portion of conifer-dominated land to be overtaken by pioneering tanoak. Commercial harvests that occurred before the passage of the Z’Berg–Nejedly Forest Practice Act of 1973 provided scarce protection to watershed conditions important to fishes and other aquatic species. During this time, roads were built adjacent to watercourses and entire


Journal of Heredity 2013:104(2)

204

watersheds were harvested with very little forest retention. Since 1998, however, the timber volume harvested has been dramatically reduced and the method of harvest has transitioned to greater reliance on uneven-aged techniques (i.e., group and single tree selection), with an increased focus on riparian stand management and restoration to improve in-stream habitat connectivity for aquatic species.

Sampling and DNA Extraction

Tailed frog larvae were sampled during the months of March–June in 2007, 2008, and 2009. Small tail clips (~2 mm2) were taken from each individual and placed immediately in 95% ethanol. DNA was extracted from a collection of 294 lar-val tail clips using a DNeasy extraction kit (Qiagen, Inc., Valencia, CA). A random sample of these extractions was then verified using a Nanodrop ND-1000 spectrophotom-eter (Thermo Scientific, Wilmington, DE) to ensure high-quality quantifiable DNA.

Polymerase chain reaction (PCR) was performed on extracted samples using microsatellites developed by Spear et al. (2008). Thirteen microsatellites were screened in these populations for variability within and among popula-tions. Further, 3 multiplexes, consisting of either 4 or 5

microsatellites, were amplified using a Qiagen Multiplex Kit (Qiagen, Inc., Valencia, CA) following manufacturer’s specifi-cations. Cycling conditions consisted of 95 °C for 15 min, fol-lowed by 27 cycles at 94 °C for 30 s, a locus-specific annealing temperature (see Spear et al. 2008) for 90 s, and finishing with an extension step at 72 °C for 60 s. An additional extension of 60° for 30 min was added to ensure reaction completion.

Before amplification, each microsatellite marker was individually labeled with a fluorescent dye found in the G5 set from ABI (Applied Biosystems, Inc., Foster City, CA). Amplified products were processed through an ABI 3130 × l Sequencer/Fragment Analyzer to allow for visualization of individual alleles. Allele size was determined with ABI Genemapper version 3.7 software using the LIZ500 size standard to allow for uniform binning and range identification (Table 1). Final genotypic data were exported into Convert version 1.31 (Glaubitz 2004) for conversion to the most common genetic analysis programs.

Population Genetic Analysis

To account for possible family structure in our samples, we first used the method of Wang (2004) and Wang and Santure (2009) to identify full sibs within each of our sites with the

Figure 1. Map showing the 4 sampled watersheds in Mendocino County, CA (United States).



205

program Colony (Jones and Wang 2010). Full sibs were removed before any further analysis as they may distort allele frequencies and create inflated levels of Hardy–Weinberg and linkage disequilibrium. Genepop’007 (Rousset 2008) was used to estimate the observed and expected heterozygosity. The methods of Guo and Thompson (1992) and Raymond and Rousset (1995) were, respectively, used to determine within-population deviations from Hardy–Weinberg equi-librium (HWE) and to estimate linkage disequilibrium (LD) across loci. Due to HWE deviations, the remaining analy-ses were performed with only 9 loci (see Results). Within-site allelic richness was estimated using the approach of Kalinowski (2005).

Population Structure

We tested for deviations in mutation–drift equilibrium using heterozygosity excess (Luikart and Cornuet 1998) and the method of Garza and Williamson (2001). The Wilcoxon test (Piry et al. 1999<cb> was used to determine whether recent population bottlenecks have led to increases in expected het-erozygosity, relative to a population with a constant effective population size (Ne) at drift–migration equilibrium, across sampled sites. In total, 1000 simulations were performed (2-phase model, 10% variance in multistep mutations, and 90%, single-step mutations) to determine mutation–drift equilibrium expectations. Deviations from equilibrium con-ditions using M-ratios were assessed through 10 000 simula-tions assuming θ (= 4 Neμ) values of 0.2, 0.5, 1.0, 2.0, and 5.0. These values assume a mutation rate of 5 × 10–4 and a pre-bottleneck Ne of 100, 250, 500, 1000, and 2500 for each value of θ. Additionally, an average size of non-single-step mutations of 3.5 and 22% of non-single-step mutations were assumed (see Peery et al. 2012, for review of param-eters). Analyses were performed with the software packages BottleneCk version1.2 (Piry et al. 1999< and CritiCal_m (Garza and Williamson 2001).

Pairwise measures of genetic differentiation were estimated as FST (Weir and Cockerham 1984), G′ST (Hedrick 2005),

Jost’s D (Jost 2008) and Dps (Bowcock et al. 1994). G′ST is a correction of standard FST measures (Weir and Cockerham 1984), which standardizes FST by the maximum possible value between 2 populations. A data set to estimate the maximal FST among population pairs was created in reCode version 1.0 (Meirmans 2006). Values for the original and recoded data sets were run through Fstat version 2.9.3 (Goudet 1995) to generate pairwise FST values. Original values were divided by the recoded FST values to obtain G′ST. Jost’s D was estimated using the online software smodG version 1.2.5 (Crawford 2010) and Dps was estimated with msa version 4.05 (Dieringer and Schlötterer 2003). Analysis of molecular variance (AMOVA) was used to assess the degree of population structure at dif-ferent hierarchical levels using 2 different sets of groupings in our AMOVAs. The first included all populations and the grouped sites were based on watershed (Cottaneva, Navarro, Elk {Upper and Lower}, and Alder—see Figure 1), whereas the second analysis was similar except for the exclusion of the Cottaneva Creek sites. This latter analysis was done to exam-ine patterns of isolation by distance at smaller spatial scales. Significance of the AMOVAs was determined with 10 000 per-mutations using the program arlequin version 3.5 (Excoffier and Lischer 2010). The method of Wilson and Rannala (2003) was used to determine whether directional gene flow is occur-ring among the sampled sites. Bayesass+ version 1.3 was run for 6 × 107 iterations, by sampling every 2 × 103 iterations and discarding information from the first 1 × 107 iterations as burn in. Delta values for F, P, and m were adjusted to ensure that the acceptance ratios fell within the recommended 40–60% range suggested by the Bayesass+ users’ manual. The analysis was run 3 independent times to assure convergence.

Self-assignments were performed using the Rannala and Mountain (1997) method in GeneClass version 2.0 (Piry et al. 2004). The program struCture version 2.2 (Pritchard et al. 2000; Hubisz et al. 2009) was used to characterize the genetic ancestry of sampled populations. We ran struCture with the following parameters: 50 000 burn-in steps, 150 000 recorded Markov chain Monte Carlo (MCMC) steps, and use of the admixture model with no previous information on

Table 1 Least-cost path variables used in landscape genetic analysis (see Materials and Methods), path criteria used for each variable, the resolution for each layer and the source of the data layers. Specific hypotheses that were tested with each variable are given

Least-Cost Path Variables Least-Cost Path Criteria Resolution Source Hypothesis

Coastal Redwood vegetation (1986)

Cost ratio: 10:5:1 (other vegetation:small conifer: medium:large conifer)

40 acres California Department of Forestry and Fire Protection

Paths selected have larger conifers

Canopy covera Maximize canopy 30 × 30 m 2001 National Land Cover Database

Paths selected have higher canopy cover

Heat load index (HLI)b Minimize HLI on path 10 × 10 m DEM Paths selected have lower heat load

Composite topographic index (CTI)c

Maximize CTI 10 × 10 m DEM Path selected have higher wetness

Stream Cost ratio: 10:1 (terrestrial: stream)

30 × 30 m — Path selected follow streams

aHomer et al. 2007.bMcCune and Keon 2002; McCune 2007.cYang et al. 2005; Sorensen et al. 2006.



206

population of origin. A total of 10 independent runs were performed varying K from 1 to 8. The approach of Evanno et al. (2005) was used to find the optimal K over all runs. Outputs were exported to Clumpp version 1.1.2 (Jakobsson and Rosenberg 2007) and combined using a greedy algorithm with 10 000 random permutations. Results were exported to distruCt version 1.1 (Rosenberg 2004) for viewing.

Landscape Analysis

We examined the relationship between several landscape variables and genetic connectivity by analyzing genetic data with geographic information system (GIS) data in a model selection framework. Following the development of a priori hypotheses and selection of landscape variables, we created least-cost paths (LCPs) in a GIS representing potential gene flow routes between sites (Spear and Storfer 2010). Totally, 6 LCPs were generated between 10 sites in 3 nearby watersheds (Lower Navarro, Elk, and Lower Alder), yielding 45 unique routes (3 sites in Cottaneva Creek watershed were excluded due to the >40-km separation distance from the other sites). LCPs were based on 5 variables hypothesized to directly or indirectly influence temperature and moisture regimes criti-cal to tailed frog life-history patterns: composite topographic index (CTI; a measure of “wetness”), heat load index (HLI; a measure of incident solar radiation), presence or absence of streams, canopy cover, and forest structure (Table 1).

Because we were limited by a small number of sam-pling locations, we developed simple paths to evaluate the relative importance of these landscape variables on genetic connectivity. Thus, 5 LCPs were based on single variables, and a sixth LCP was derived by combining 2 variable rasters minimizing HLI and maximizing CTI. HLI and CTI were derived from a Digital Elevation Model (DEM) in a GIS. CTI was calculated using the DEM and the D-infinity mul-tiflow direction algorithm for estimating flow accumulation from upslope contributing area (Tarboton 1997; Tarboton 2003; Stichbury et al. 2011) and is positively related to mois-ture. The D-infinity flow accumulation raster was estimated with TauDEM 5.0, and CTI was calculated on an ArcInfo Workstation (Tarboton and Mohammed 2010). Stream path was based on a raster indicating the presence or absence of a perennial stream bearing aquatic life. A cost ratio of 1:10 (stream: terrestrial) was used for this path following Spear and Storfer (2010). Canopy cover percentage was derived from the United States Geological Survey National Land Cover Database 2001 (version 1.0; Homer et al. 2007) and was max-imized over the LCP. Finally, the forest structure variable ras-ter was derived from the 1986 Coastal Redwood Vegetation layer developed by Fox (1998). We chose this coarse-scale layer because it represented a lag time of 20 years before our sampling and might better explain contemporary genetic pat-terns (Spear and Storfer 2008). For this path, 13 unique for-est structure classes spanning the study area were condensed into 3 distinct attribute categories in the Coastal Redwood Vegetation layer based on species, size class, and density and these were assigned a cost ratio of 10:5:1 (other vegetation: small conifer: medium-to-large conifer).

A cost raster was developed for each of the 5 variable rasters, where a cell was allocated a weighted value between 1 and 10 (1 equals the least cost or most favorable conditions, and 10 equals the greatest cost) to estimate the cost to tra-verse a given cell for that variable (Table 1). Canopy cover was the exception as it was given a weighted value between 1 and 100 (the inverse of canopy cover percentage). A cost-distance function calculates the LCP traveling through the most favorable cells for a given variable from a source cell (the location of a sample site) to all other sampled sites. This resulted in 2, often nonidentical, paths for each pairwise com-parison, effectively doubling the number of paths for each LCP. Because gene flow directionality between sites was not determinable, we partitioned the LCP data set into 2 groups of 45 pairwise comparisons and analyzed each group sep-arately. Independent variables such as topographic distance, average canopy cover, average forest structure, average CTI, and average HLI were calculated for each LCP between sites. LCPs were created using ARCGIS version 10 and the LCP’s topographic distance was calculated using Surface Tools extension (Jenness 2008) for ARCVIEW 3.2, version 1.6b.

Linear regression was used to examine relationships between several independent variables for a given LCP and each genetic distance measure (dependent variable). A second-order Akaike Information Criterion (AICc = n × log {(RSS/n) + 2K[n/(n − K − 1)]}), correcting for small sam-ple size, was calculated for each regression model and used to determine the most parsimonious model in the set represent-ing our a priori hypotheses (Burnham and Anderson 2002). The AICc was calculated using the maximum likelihood esti-mate (MLE) approximated from the linear regression out-put (formula: MLE = RSS/n, where RSS = residual sum of squares and n = sample size; Burnham and Anderson 2002). The number of parameters (K) used in each model was deter-mined by the number of estimated regression parameters, plus intercept and variance. LCP models were compared with traditional isolation-by-distance (IBD) models and a null model to determine whether specific landscape variables were more informative predictors of genetic connectivity than a resistance-free landscape. Landscape variables were retained in the model only if their estimated coefficient’s 95% confidence intervals were nonoverlapping with zero.

Regression diagnostics were performed on each linear regression model to ensure that assumptions of normally distributed residuals and constant variance were met before calculating the MLEs used in model selection. Similarly, because our data were distributed in geographic space, it was also necessary to ensure that linear regression was the appropriate statistical method to model landscape variables and genetic distance. Failing to account for spatial autocorrelation in statistical models may result in biased parameter estimates and spurious model fit and, thus, lead to poor inference (Legendre 1993). To assess whether spatial autocorrelation in the residuals was present in our data, we calculated Global Moran’s I for each candidate model and determined whether it was significantly different from the random result (P < 0.05). We employed the inverse-distance weighting method and based our matrix on single (straight-line) midpoint values between site



207

pairs (see Spear and Storfer 2008, 2010). Finally, after ruling out the presence of any spatial dependencies in our models, we employed the nonparametric bootstrap to assess model uncertainty, resampling the original data set with replacement (N = 45) and determining the top model in each run of 10 000 iterations. All statistics were performed with Stata/SE version 11.2 for Windows. Stata modules spwmatrix and anketest were used to generate the spatial weights matrix and calculate Global Moran’s I, respectively (Jeanty 2010a, 2010b).

ResultsPopulation Genetic Analysis

A total of 294 individual tailed frogs were genotyped at 12 loci; however, 3 loci were removed due to HWE deviations (see the following sections; Table 2). Moreover, 33 individuals were removed due to the identification of full-sib relation-ships. The HWE analysis produced 25 out of 156 significant tests, whereas the LD analysis produced 14 out of 858 signifi-cant tests (both taking into account Bonferroni corrections for multiple comparisons). A majority of significant departures from HWE were found at 3 loci (A31, A17, and A3–14 tests), a problem rectified by removing these loci from subsequent analyses. Moreover, 1 final locus (A1) did not consistently amplify across samples, leaving a total of 9 microsatellite loci. Allelic richness did not vary substantially across sample sites (Table 2), though the sites from the Cottaneva Creek drainage had slightly lower levels of allelic richness when compared with other collection sites. Only 1 site showed significant departure from mutation–drift equilibrium (CA1008) with the heterozygosity excess test. Further, 2 sites contained signifi-cantly lower M-ratio values (Table 2). These were dependent on which pre-bottleneck θ was assumed.


Levels of pairwise genetic differentiation were significant in all but 2 comparisons (Supplementary Material). Lower levels of pairwise differentiation were evident from within-water-shed comparisons. Mean overall FST and G′ST were 0.077 and 0.425, respectively. The AMOVA indicated that, for both sets of groupings, substantial genetic structure existed at all hier-archical levels examined (Table 3). A lower amount of genetic variation is partitioned among groups when the Cottaneva Creek sites are removed (Table 3: Group 2).

Individual-based assignments successfully assigned 96.1% of individuals back to their original site of collection (Table 4). When misassignments did occur, individuals were always assigned to a site within the same watershed (Table 4). The Bayesian analysis of migration rates revealed that significant migration events only occurred at the within-drainage level, with single-source populations found in the Cottaneva Creek, Navarro River, and Elk River catchments (Table 4). The Evanno method suggested that K = 2 was the most appropriate level of clustering (not shown). However, inspection of the likelihood values for all values of K indicated a plateau at K = 4 (Supplementary Material). We, therefore, also report results of the STRUCTURE analysis for K = 3

and K = 4 (Figure 2). There was shared ancestry between the lowermost site in the Navarro watershed (WL1003) and the Cottaneva Creek sites and the uppermost sites in the Navarro watershed (WL1008 and WL1009) and the sites from Elk and Alder Creeks across all reported K values (Figure 2).

Landscape Analysis

Landscape modeling, using different measures of genetic distance, consistently found that LCP models based on HLI + CTI or stream path ranked higher than models based on IBD, canopy, forest structure, CTI, and HLI. These 2 models had substantial support and were indistinguishable from one another (ΔAICc values: 0–2; Burnham and Anderson 2002) in analyses when FST and G′ST were used, whereas HLI + CTI was the top overall model when Jost’s D and Dps were used (Table 5). If we include models with ΔAICc values 2–3 in the candidate set, FST had 7 highly supported models, G′ST 2, Jost’s D 4, and Dps 3 (Table 5). The LCP based on HLI + CTI ranked high across all genetic distances; Stream and IBDEuclidean distance ranked high for 3 genetic distances; and IBDTopographic distance ranked high for 2 genetic distances. All remaining models had moderate levels of support (ΔAICc values 3–7) and all LCP and IBD models were more inform-ative than the null model (Table 5). Topographic distance was the only independent variable retained in the regression analysis because its 95% confidence interval did not include zero. All distance measures were positively correlated with genetic distance (data not shown), and 23–54% of the vari-ation was explained by any single model for all 4 genetic distance measures (Table 5). Model rankings based on boot-strap selection frequencies (πi) for 10 000 bootstrap samples mirrored AICc rankings for the top models based on mois-ture variables. These frequencies represented a confidence set ranging from 80–93% for all measures of genetic dis-tance (Table 5).

DiscussionAmphibians are often touted as strong indicators of envi-ronmental change and stress due to various physiological and ecological characteristics (Welsh and Ollivier 1998; Welsh and Hodgson 2008). Although long-distance dispersal events are rare for amphibian species (Funk et al. 2005b), a majority of studies often find geographically limited gene flow and dispersal and a strong role for topography in mitigating gene flow (Funk et al. 2005a; Wang 2009; Savage et al. 2010). Our study demonstrates that tailed frogs in the southernmost portion of their range exhibit significant fine-scale popula-tion structure at the watershed level and that gene flow is cor-related with environmental features that relate to the biology of the species. We also found that despite a history of land use alterations in the region, a majority of the sites did not show signs of a recent population bottleneck. These results suggest that although habitat alteration in the study region has been pronounced, populations of coastal tailed frogs in the Mendocino coast region appear to have remained viable despite widespread disturbance.


http://jhered.oxfordjournals.org/lookup/suppl/doi:10.1093/jhered/ess138/-/DC1



208

Tabl

e 2

Site

cod

es, s

ampl

e siz

es (N

full =

tota

l col

lect

ed a

t eac

h sit

e; N

trim

= to

tal r

emai

ning

afte

r rem

oval

of

high

ly re

late

d in

divi

dual

s [se

e M

ater

ials

and

Met

hods

for d

etai

ls]),

estim

ates

of

obse

rved

and

exp

ecte

d he

tero

zygo

sitie

s (H

O a

nd H

E) f

or e

ach

of th

e sa

mpl

e sit

es. M

ean

alle

lic ri

chne

ss (A

R) a

nd re

sults

from

het

eroz

ygos

ity e

xces

s and

M-R

atio

test

s are

also

pre

sent

ed

Wat

ersh

ed

Cot

tane

va C

reek

Nav

arro

Riv

erEl

k C

reek

Ald

er C

reek

Site

RC10

04RC

1005

RC10

06W

L100

3W

L100

5W

L100

7W

L100

8W

L100

9C

L100

3C

E10

01C

E10

02.5

CE

1009

CA

1008

Nfu

ll35

3131

2314

3014

1630

831

1516

Ntri

m31

2727

2114

2714

1625

828

1515

Locu

s

A12

H

E0.

228

0.07

30.

037

0.13

80.

000

0.03

70.

000

0.38

70.

000

0.00

00.

070

0.00

00.

497

H

O0.

194

0.07

40.

037

0.04

80.

000

0.03

70.

000

0.00

00.

000

0.00

00.

000

0.00

00.

000

A

2

HE

0.81

80.

751

0.74

80.

897

0.86

20.

831

0.88

40.

911

0.80

50.

672

0.88

80.

805

0.84

1

HO

0.61

30.

815

0.76

90.

714

0.84

60.

667

0.85

70.

875

0.73

10.

625

0.82

10.

600

0.46

7

A4

H

E0.

844

0.86

70.

844

0.93

70.

905

0.92

90.

939

0.88

30.

909

0.87

50.

940

0.93

30.

926

H

O0.

903

0.92

60.

926

0.95

21.

000

0.96

31.

000

0.93

80.

885

0.87

50.

929

0.93

30.

800

A

15

HE

0.84

10.

877

0.83

50.

911

0.70

40.

903

0.87

00.

891

0.88

60.

656

0.86

40.

908

0.86

4

HO

0.74

20.

926

0.81

50.

952

0.71

40.

963

0.78

60.

938

0.80

80.

625

0.89

30.

933

0.80

0

A24

H

E0.

913

0.89

90.

919

0.93

10.

921

0.91

20.

921

0.88

10.

888

0.89

10.

888

0.93

30.

887

H

O0.

935

0.88

90.

926

0.90

50.

857

0.92

60.

929

0.56

30.

692

1.00

00.

750

0.93

30.

867

A

26

HE

0.88

20.

908

0.93

10.

937

0.94

70.

904

0.95

00.

933

0.80

00.

875

0.91

80.

924

0.92

6

HO

0.90

30.

926

0.92

60.

857

1.00

00.

852

1.00

00.

938

0.80

81.

000

0.96

40.

867

0.93

3

A29

HE

0.85

00.

822

0.83

70.

937

0.90

20.

924

0.91

00.

869

0.93

30.

875

0.92

80.

939

0.93

4

HO

0.74

20.

667

0.77

80.

952

1.00

00.

852

0.71

40.

688

0.91

71.

000

0.96

30.

929

0.92

9

A13

H

E0.

508

0.44

00.

444

0.67

40.

600

0.60

50.

429

0.71

20.

324

0.47

60.

168

0.19

10.

434

H

O0.

613

0.55

60.

444

0.71

40.

692

0.74

10.

500

0.50

00.

231

0.62

50.

179

0.20

00.

333

A

14

HE

0.91

40.

908

0.92

20.

906

0.91

10.

904

0.95

20.

935

0.89

80.

883

0.93

40.

939

0.93

1

HO

1.00

00.

963

0.88

90.

810

0.53

80.

889

0.92

90.

875

0.92

30.

875

0.88

91.

000

0.66

7

Mea

n

HE

0.75

50.

727

0.72

40.

808

0.75

00.

772

0.76

20.

823

0.71

60.

735

0.73

30.

730

0.80

5

HO

0.73

80.

749

0.72

30.

767

0.73

90.

765

0.74

60.

701

0.66

60.

736

0.71

00.

711

0.64

4

AR

7.44

7.42

7.74

10.0

98.

949.

2410

.04

9.2

8.59

7.55

9.63

9.9

9.06

Wilc

oxon

nsns

nsns

nsns

nsns

nsns

nsns

P <

0.0

5M

-rat

io0.

764

0.81

90.

843

0.77

60.

705*

0.79

70.

817

0.77

80.

765

0.65

3**

0.81

50.

747

0.70

5

ns, not significant at α = 0.05; *, significant with θ = 0.2; **, significant with θ ≤ 2.0.



209

Population Genetic Analysis

Equitable amounts of allelic richness and heterozygosity were observed across sites, with slightly lower amounts in Cottaneva Creek. This equitable amount of variation, coupled with the lack of any consistent bottleneck signatures across sites, indicates that populations studied here have comparable demographic histories. The lack of any reduction in variation across sites is unusual given the history of timber harvest in the region. Spear and Storfer (2008) found a limited incidence of recent bottlenecks for coastal tailed frogs in forested regions from Washington State, whereas Spear and Storfer (2010) did not find any bottleneck signatures in Rocky Mountain tailed frog (Ascaphus montanus) populations from harvested sites in north-central Idaho. In either case, they attributed the lack of bottleneck signatures, for both the Heterozygosity Excess test and the M-Ratio test, to a demographic rebound from historical reductions in effective population size and/or increased migration. Given that our study sites have been intensively harvested during the past century (Noss 2000; Stewart 2007) and that harvest volumes have declined and practices changed over the past decade (Douglas R, personal observation), it is conceivable our study populations have recovered from initial declines, thus obscuring any bottleneck signatures. Likewise, elevated within-watershed migration rates may also account for the lack of recent reductions in effective population size throughout our study sites (see following section and Busch et al. 2007). In all cases, it appears that most coastal and Rocky Mountain tailed frog populations are capable of genetically rebounding from large-scale environmental perturbations.


The major unit that governs genetic structure in this study appears to be the watershed. AMOVA results indicated a sig-nificant degree of genetic variation partitioned at the water-shed level, which is also reflected in the struCture analysis. Local dynamics appear to be important in structuring popula-tions of tailed frogs in the southern portion of their range. Assignment “tests” indicate a high degree of assignment back to the initial site of collection, with any misassignments occur-ring at the within-watershed scale. Likewise, the Bayesian analysis for recent gene flow only found significant instances within watersheds. The struCture analysis predominately

grouped samples by watershed, with the exception of indi-viduals from WL1008 and WL1009 (Navarro River), who displayed mixed ancestry with samples from the Elk/Alder Creek watershed. Interestingly, the Bayesian analysis of recent migration did not detect any migratory events between these sites. This discrepancy could be due to historical connec-tions and contemporary isolation between the 2 watersheds. Moreover, a single intervening watershed does exist between the Navarro River and Elk Creek, Greenwood Creek, though we did not sample this region. Further sampling in this area would allow differentiation between the relative roles of gene flow versus historical structure. Regardless, this observation is restricted to a single case and does not discount the high amount of population structure observed in our study.

High site fidelity is a common pattern documented in tailed frogs in our study region (Burkholder and Diller 2007), which would account for the significant population struc-ture observed at the watershed level. In coastal tailed frogs from the Olympic Peninsula, relatively high amounts of gene flow were found across 2 study regions (5–20 km; Spear and Storfer 2008). A similar result was also found for Rocky Mountain tailed frogs (Spear and Storfer 2010). In contrast, our southernmost sites (Navarro, Elk, and Alder) are within the southern extent of the geographic range of the coastal tailed frog; yet, we find a relatively higher degree of dif-ferentiation when compared with these other studies. This degree of site fidelity could account for resilience to past dis-turbances, which would have a much greater effect on spe-cies with higher dispersal capabilities, or could be a result of adaptation to habitat fragmentation or warmer regional cli-mate. This also suggests larger-scale environmental changes (like those from regional and global climate change) would have a much greater effect on this species.

Landscape Analysis

Recent genetic analyses of tailed frog populations reveal that a combination of physiographic, vegetative, and climate vari-ables may influence gene flow over different landscapes that exhibit a variety of natural and anthropogenic disturbances (Spear and Storfer 2008, 2010; Spear et al. 2012). Our results indicate that tailed frog gene flow on commercially man-aged coast redwood forests near their southern range termi-nus likely occurs along riparian corridors and other areas of

Table 3 Results of the AMOVA analysis for the 2 grouping scenarios. Scenario ‘1’ grouped all sampled sites by watershed and scenario ‘2’ grouped all sites by watershed after the removal of the Cottaneva Creek sites

Group Source of Variation Sum of Squares Variance Component Percentage Variation P values

1 Among groups 136.724 0.2990 8.04 <0.0001Among sites within groups 52.017 0.0688 1.85 <0.0001Within sites 1836.593 3.3514 90.11 <0.0001Total 2025.334 3.7193 — —

2 Among groups 32.210 0.0917 2.58 0.0038Among sites within groups 38.091 0.0748 2.11 <0.0001Within sites 1209.013 3.3866 95.31 0.0002Total 1279.314 3.5531 — —



210

Tabl

e 4

Results of the individual-based assignments and Bayesian analysis for recent gene flow. Numbers on the diagonal indicate self-assignment back to the original source population.

Mean estimates of gene flow, followed by 95% confidence intervals (in parentheses) are given only for cases where elevated gene flow was found and are presented after individual

assig

nmen

ts. A

ll va

lues

are

gre

ater

than

wou

ld b

e ex

pect

ed fo

r 13

popu

latio

ns. N

umbe

rs in

bol

d in

dica

te m

isass

ignm

ents

Sour

ce

Ass

igne

d to

/Mig

ratio

n In

to

RC

1004

RC

1005

RC

1006

WL1

003

WL1

005

WL1

007

WL1

008

WL1

009

CL1

003

CE1

001

CE1

0025

CE1

009

CA

1008

RC10

0427

4 /

0.27

2(0

.202

, 0.3

23)

/ 0.

273

(0.1

98, 0

.323

)—

——

——

——

——

—

RC10

05 1

26—

——

——

——

——

——

RC10

06—

225

——

——

——

——

——

WL1

003

——

—21

——

——

——

——

—W

L100

5—

——

—14

——

——

——

——

WL1

007

——

——

—26

1—

——

——

—W

L100

8—

——

——

—14

——

——

——

WL1

009

——

—/

0.22

4(0

.135

, 0.3

09)

— 1

—15

——

——

—

CL1

003

——

——

——

——

26—

/ 0.

257

(0.1

73, 0

.318

)—

—

CE

1001

——

——

——

——

—8

——

—C

E10

025

——

——

——

——

——

27 1

—C

E10

09—

——

——

——

——

——

15—

CA

1008

——

——

——

——

——

——

15



211

high moisture with low solarization (streams and wet areas on north-facing slopes). Although these results run coun-ter to our initial hypothesis that forest structure and canopy cover would be more informative predictors of gene flow, we are unable to dismiss these alternative hypotheses as fully uninformative as there is a range of factors that could have influenced this outcome. Furthermore, we also found that univariate regression models relating topographic distance to genetic distance best explained variation over all the LCPs without any contribution from additional landscape covari-ates. This contrasts with previous studies that routinely iden-tified several landscape variables as informative predictors of gene flow—a result that may be attributed to regional differ-ences in physiography and forest management.

Previous studies of tailed frog genetics on managed land-scapes appear to have greater topographic relief than our

study area, in addition to larger blocks of distinct even-aged stands and nonforest areas due to clear-cutting (Spear and Storfer 2008, 2010; Spear et al. 2012). In contrast, our study area was located in a low-elevation, coastal redwood for-est that had a past history of large-scale clear-cutting. Even though we used canopy cover and forest structure layers in our GIS analyses to represent past landscape conditions, the different management regimes may have had the effect of homogenizing the landscape with respect to these variables. This could make it difficult to distinguish their effect on gene flow relative to a landscape with a detectable history of intensive even-aged management, and this possibly explains why they consistently ranked lower than moisture-related and resistance-free surfaces in the model selection analyses (Anderson et al. 2010). Future studies should pursue the use of data that more accurately represent fine-scale landscape

Figure 2. Results of the STRUCTURE analysis for K= 2, 3, and 4. Proportion of ancestry to each cluster is indicated in pie charts for each sample site.



212

features conducive for tailed frog migration corridors, unlike the scale and resolution of the forest structure and canopy cover GIS layers used here.

Overall, these results suggest that tailed frog migration corridors in this region occur along areas that provide cool and moist habitat, a finding consistent with this species’ known biology. Tailed frogs have low tolerance for elevated air and water temperatures (Brown 1975; Welsh 1990; Welsh et al. 2005a), especially in the larval stage (Altig and Brodie 1972), and are subject to desiccation at all stages of their life cycle. Although several studies have documented that post-metamorphic tailed frogs move through upland habitats in harvested areas (Matsuda and Richardson 2005), others sug-gest they exhibit low dispersal ability, remain close to the

stream environment, and move through riparian corridors (Daugherty and Sheldon 1982b; Burkholder and Diller 2007). Spear and Storfer (2010) found that stream paths were asso-ciated with genetic connectivity for Rocky Mountain tailed frogs and suggested this pattern was a result of both regional climate and habitat disturbance. A similar pattern with stream path was also detected in coastal tailed frog populations in managed areas lacking significant forest cover near Mount St. Helens (Spear et al. 2012). Reduced annual precipitation and increased summertime temperatures associated with more southerly latitudes likely impose further physiological stress on tailed frog populations and may explain regional differences in their behavior, habitat use patterns, and ulti-mately, their southern range distribution. We propose that

Table 5 Model selection results relating genetic distance to topographic distance for least-cost path (LCP) and isolation by distance (IBD) models (N = 45). Bootstrap selection frequencies (πi) are based on the number of times a model was selected as the top model out of 10 000 bootstrap samples of the original data set. Bold lines indicate modes with ΔAICc <3

Genetic Distance

Model Rank Model

No. ofParameters AICc ΔAICc

AkaikeWeights, wi r2

BootstrapSelectionFrequency, πi

FST 1 LCPHLI + CTI 3 –374.43 0.00 0.30 0.35 0.412 LCPStream 3 –373.59 0.84 0.20 0.32 0.343 LCPCTI 3 –372.99 1.44 0.15 0.33 0.184 IBDEuclidean

a 3 –372.55 1.88 0.12 0.32 0.035 IBDTopographic

b 3 –371.89 2.54 0.08 0.31 0.016 LCPForest 3 –371.55 2.88 0.07 0.30 0.007 LCPCanopy 3 –371.50 2.93 0.07 0.30 0.018 LCPHLI 3 –366.84 7.59 0.01 0.23 0.029 null3 2 –357.62 16.81 0.00 — 0.00

G′ST 1 LCPStream 3 –253.949 0.00 0.45 0.42 0.522 LCPHLI + CTI 3 –253.233 0.72 0.31 0.41 0.353 IBDTopographic

a 3 –250.012 3.94 0.06 0.37 0.014 IBDEuclidean

b 3 –249.333 4.62 0.04 0.36 0.005 LCPHLI 3 –249.216 4.73 0.04 0.36 0.116 LCPForest 3 –249.208 4.74 0.04 0.36 0.007 LCPCanopy 3 –248.372 5.58 0.03 0.34 0.008 LCPCTI 3 –247.703 6.25 0.02 0.33 0.009 nullc 2 –231.386 22.56 0.00 — 0.00

Jost’s D 1 LCPHLI + CTI 3 –250.153 0.00 0.44 0.40 0.672 LCPStream 3 –247.554 2.60 0.12 0.36 0.153 IBDEuclidean

a 3 –247.547 2.61 0.12 0.36 0.054 LCPForest 3 –247.385 2.77 0.11 0.36 0.025 IBDTopographic

b 3 –246.507 3.65 0.07 0.35 0.006 LCPCanopy 3 –246.114 4.04 0.06 0.34 0.017 LCPCTI 3 –246.02 4.13 0.06 0.34 0.008 LCPHLI 3 –244.85 5.30 0.03 0.32 0.109 null 2 –229.221 20.93 0.00 — 0.00

Dps 1 LCPHLI+CTI 3 –280.506 0.00 0.53 0.54 0.682 IBDTopographic

a 3 –278.188 2.32 0.16 0.51 0.103 IBDEuclidean

b 3 –277.534 2.97 0.12 0.50 0.044 LCPStream 3 –276.702 3.80 0.08 0.49 0.125 LCPCanopy 3 –276.128 4.38 0.06 0.49 0.016 LCPForest 3 –274.772 5.73 0.03 0.47 0.007 LCPCTI 3 –273.605 6.90 0.02 0.46 0.008 LCPHLI 3 –272.034 8.47 0.01 0.44 0.049 null 2 –247.822 32.68 0.00 — 0.00

aEuclidean distance path accounting for topography (slope).bMinimum straight-line distance between sites.cAll-constants model.



213

tailed frogs in the extreme southern part of their range are more likely to be confined to stream corridors, seeps, springs, and other wet areas because refugia are limited in upland habitats and suitable conditions for overland movement vary seasonally. Watercourse buffers and wet area protections are therefore likely to be highly beneficial to the tailed frog as the maintenance of suitable water temperatures has been a key determinant in their ability to persist in a region with a long history of timber harvesting now currently dominated by young redwood forests.

Study Limitations

Landscape genetic studies have the potential to shape conser-vation policy and land management decisions, yet research-ers have cautioned that modeling techniques can be sensitive to numerous factors that can lead to erroneous conclu-sions (Anderson et al. 2010; Cushman and Landguth 2010; Landguth et al. 2010; Spear et al. 2010). Our analysis may suf-fer from 1 or more of these issues. First, the number of tailed frog sites sampled in this study may be too small to accom-modate the modeling of landscape variables in a way that dis-tinguishes competing a priori hypotheses. The considerable unexplained variation in all of our models may be due to the small sample size but could also indicate that there were addi-tional variables or higher resolution GIS features that should have been included in the analysis. Second, our analysis showed that FST contained the least amount of power to dis-tinguish among numerous landscape models when compared with other genetic distance measures. It has been argued that FST may not adequately reflect genetic differentiation for highly variable markers like microsatellites (Hedrick 2005; Jost 2008; Meirmans and Hedrick 2011; Whitlock 2011). The genetic distances presented here have different underlying assumptions about genetic differentiation, and their inclusion in the model selection analyses was able to identify a single model (HLI + CTI) with strong support across distances. However, it is unclear if the inability to distinguish among numerous models using FST is a shortcoming of this particu-lar metric, our sampling design, or the scale and quality of the GIS layers used in the analysis. Therefore, we advocate using these results as working hypotheses requiring further refine-ment and evaluation. Future work in the region should focus on increasing sampling intensity and consider additional vari-ables (GIS data and more complex cost surfaces) to better evaluate these landscape hypotheses.

Conclusions and Conservation ImplicationsUnderstanding population structure at range edges is of high interest to the conservation of the entire species as range-edge populations may contain highly divergent populations, distinct pools of genetic variation, and may suffer increased susceptibility to local extinction in the face of global climate change (Hampe and Petit 2005). Recent work has suggested that summer fog coverage may decline in the coast redwood

region (Johnstone and Dawson 2010), which could impose a further physiological stress on southern range-edge amphib-ian populations that depend on the cool, maritime climate. In all, given the high amount of population structure and potential negative impacts from warming scenarios predicted by climate change models, it appears that southern edge tailed frog populations may be highly susceptible to localized extinction, even in the face of presumed resiliency to past habitat alteration and loss. Conservation efforts should con-sider the potential impacts of future climate change scenarios and how best to mitigate their negative effects.

Previous work on the relationship between amphibian population genetic parameters and timber harvest prac-tices in western North America has ultimately been aimed at determining the detrimental effects of timber harvest. We were unable to assess the impacts of fragmentation on tailed frog differentiation because suitable reference sites (e.g., old-growth or late-seral forest) are absent from this region. However, we found no evidence for recent reduc-tions in effective population size for any of our sites. Spear and Storfer (2008) found evidence for long-distance gene flow and movement over forested habitats for tailed frogs in coastal Washington (United States). Our results differ in that recent migration events were only found within water-sheds, but we did find evidence for migration corridors along streams and moist areas. This contradiction may underscore the difference in regional climates and forest types between coastal northern California and Washington (United States).

Clearly this work, and that of others (Welsh et al.), sug-gests that conservation of headwater stream amphibians in the redwood region of northern California is a highly com-plex and important issue. We suggest this region should be of high conservation importance given the narrow physiological tolerances of amphibians, the general feature of high genetic structure, and a history of large-scale landscape alteration. Conservation and management strategies should not only aim to establish riparian buffers in harvested regions but also identify restoration projects that will hasten the development of suitable habitats in degraded areas conducive for tailed frog breeding and overland movement. Likewise, maintaining areas of upslope habitats, which provide linkages between headwater streams, would also be beneficial for improving connectivity within and between watersheds (Olsen and Burnett 2009). Furthermore, detailed estimates of direct dis-persal, coupled with a more intensive landscape approach, will provide a more accurate assessment of movement cor-ridor requirements for this species.

Supplementary MaterialSupplementary material can be found at http://www.jhered.oxfordjournals.org/.

FundingUnited States Department of Agriculture(2007-35101-18171) to A.A.





214

AcknowledgmentsWe would like to thank D. Ulrich, C. Morris, K. Vodopals, and J. Heras for assistance in the collection of samples. D. Cooper and A. Perez assisted in the generation of microsatellite data. All samples were collected with per-mission from the California Department of Fish and Game (SC-006943) under the purview of an animal care protocol from UC Merced (UCM07-01/190). We are grateful to the Mendocino Redwood Company for access to the field sites. This manuscript was improved by comments from 5 anony-mous reviewers and A.J. Crawford.

ReferencesAdams MJ, Pearl CA. 2005. Ascaphus truei. In: Lannoo M, editor. Amphibian declines: the conservation status of United States species. Berkeley (CA): University of California Press. p. 382–384.

Altig R, Brodie ED. 1972. Laboratory behavior of Ascaphus truei tadpoles. J Herpetol. 6:21–24.

Anderson CD, Epperson BK, Fortin MJ, Holderegger R, James PM, Rosenberg MS, Scribner KT, Spear S. 2010. Considering spatial and temporal scale in landscape-genetic studies of gene flow. Mol Ecol. 19: 3565–3575.

Ashton DT, Marks SB, Welsh HH. 2006. Evidence of continued effects from timber harvesting on lotic amphibians in redwood forests of north-western California. Forest Ecol Manag. 221:183–193.

Barbour MC, Keeler-Wolf T, Schoenherr AA. 2007. Terrestrial vegetation of California. Berkeley (CA): University of California Press.

Bowcock AM, Ruiz-Linares A, Tomfohrde J, Minch E, Kidd JR, Cavalli-Sforza LL. 1994. High resolution of human evolutionary trees with polymorphic microsatellites. Nature. 368:455–457.

Brattstrom BH. 1963. A preliminary review of the thermal requirements of amphibians. Ecol. 44:238–255.

Brown HA. 1975. Temperature and development of the tailed frog, Ascaphus truei. Comp Biochem Physiol A Comp Physiol. 50:397–405.

Burkholder LL, Diller LV. 2007. Life history of post-metamorphic coastal tailed frogs (Ascaphus truei) in northwestern California. J Herpetol. 41:251–262.

Burnham KP, Anderson DR. 2002. Model selection and multimodel inference: a practical information-theoretic approach. 2nd ed. New York: Springer-Verlag.

Bury RB. 1983. Differences in amphibian populations in logged and old growth redwood forest. Northwest Sci. 57:167–178.

Busch JD, Waser PM, Dewoody JA. 2007. Recent demographic bottlenecks are not accompanied by a genetic signature in banner-tailed kangaroo rats (Dipodomys spectabilis). Mol Ecol. 16:2450–2462.

Chen J, Saunders SC, Crow TR, Naiman RJ, Brosofske KD, Mroz GD, Brookshire BL, Franklin JF. 1998. Microclimate in forest ecosystem and landscape ecology. Bioscience. 49:288–297.

Corn PS, Bury RB. 1989. Logging in western Oregon: responses of head-water habitats and stream amphibians. Forest Ecol Manag. 29:39–57.

Crawford NG. 2010. SMOGD: software for the measurement of genetic diversity. Mol Ecol Res. 10:556–557.

Cushman SA. 2006. Effects of habitat loss and fragmentation on amphib-ians: a review and prospectus. Biol Cons. 128:231–240.

Cushman SA, Landguth EL. 2010. Spurious correlations and inference in landscape genetics. Mol Ecol. 19:3592–3602.

Daugherty CH, Sheldon AL. 1982a. Age-determination, growth, and life history of a Montana population of the tailed frog (Ascaphus truei). Herpetologica. 38:461–468.

Daugherty CH, Sheldon AL. 1982b. Age-specific movement patterns of the tailed frog Ascaphus truei. Herpetologica. 38:468–474.

Dieringer D, Schlötterer C. 2003. Microsatellite Analyer (MSA): a platform independent analysis tool for large microsatellite data sets. Mol Ecol Notes. 3:167–169.

Diller LV, Wallace RL. 1999. Distribution and habitat of Ascaphus truei in streams on managed, young growth forests in North Coastal California. J Herpetol. 33:71–79.

Dupuis L, Steventon D. 1999. Riparian management and the tailed frog in northern coastal forests. Forest Ecol Manag. 124:35–43.

Eckert CG, Samis KE, Lougheed SC. 2008. Genetic variation across species’ geographical ranges: the central-marginal hypothesis and beyond. Mol Ecol. 17:1170–1188.

Evanno G, Regnaut S, Goudet J. 2005. Detecting the number of clusters of individuals using the software STRUCTURE: a simulation study. Mol Ecol. 14:2611–2620.

Excoffier L, Lischer HE. 2010. Arlequin suite ver 3.5: a new series of pro-grams to perform population genetics analyses under Linux and Windows. Mol Ecol Resour. 10:564–567.

Fahrig L., Merriam G. 1985. Habitat patch connectivity and population survival. Ecology. 66:1762–1768.

Fischer J, Lindenmayer DB. 2007. Landscape modification and habitat fragmentation: a synthesis. Glob Ecol Biogeog. 16:265–280.

Fox L. 1989. A classification, map and volume estimate for the coast red-wood forest in California [Internet]. Sacramento (CA): The Forest and Rangeland Resources Assessment Program, California Department of Forestry and Fire Protection. [cited 2012 Dec 20]. Available from: http://frap.cdf.ca.gov/data/frapgisdata/select.asp

Frankham R, Ballou JD, Briscoe DA. 2002. Introduction to conservation genetics. Cambridge (MA): Cambridge University Press.

Funk WC, Blouin MS, Corn PS, Maxell BA, Pilliod DS, Amish S, Allendorf FW. 2005a. Population structure of Columbia spotted frogs (Rana luteiventris) is strongly affected by the landscape. Mol Ecol. 14:483–496.

Funk WC, Greene AE, Corn PS, Allendorf FW. 2005b. High dispersal in a frog species suggests that it is vulnerable to habitat fragmentation. Biol Lett. 1:13–16.

Garza JC, Williamson EG. 2001. Detection of reduction in population size using data from microsatellite loci. Mol Ecol. 10:305–318.

Glaubitz JC. 2004. CONVERT: a user-friendly program to reformat diploid genotypic data for commonly used population genetic software packages. Mol Ecol Res. 4:309–310.

Goudet J. 1995. FSTAT (Version 2.9.3): a computer program to calculate F-statistics. J Hered. 86:485-486.

Guo SW, Thompson EA. 1992. Performing the exact test of Hardy-Weinberg proportion for multiple alleles. Biometrics. 48:361–372.

Hampe A, Petit RJ. 2005. Conserving biodiversity under climate change: the rear edge matters. Ecol Lett. 8:461–467.

Hedrick PW. 2005. A standardized genetic differentiation measure. Evolution. 59:1633–1638.

Homer C, Dewitz J, Fry J, Coan M, Hossain N, Larson C, Herold N, McKerrow A, VanDriel JN, Wickham J. 2007. Completion of the 2001 National Land Cover Database for the conterminous United States. Photogramm Eng Remote Sens. 73:337–341.

Hubisz MJ, Falush D, Stephens M, Pritchard JK. 2009. Inferring weak popu-lation structure with the assistance of sample group information. Mol Ecol Resour. 9:1322–1332.

Jackson CR, Batzer DP, Cross SS, Haggerty SM, Sturm CA. 2007. Headwater streams and timber harvest: channel, macroinvertebrate, and amphibian response and recovery. For Sci. 53:356–370.

Jakobsson M, Rosenberg NA. 2007. CLUMPP: a cluster matching and per-mutation program for dealing with label switching and multimodality in ana-lysis of population structure. Bioinformatics. 23:1801–1806.


http://frap.cdf.ca.gov/data/frapgisdata/select.asp

http://frap.cdf.ca.gov/data/frapgisdata/select.asp


215

Jeanty PW. 2010a. ANKETEST: Stata module to perform diagnostic tests for spatial autocorrelation in the residuals of OLS, SAR, IV, and IV-SAR models, Statistical Software Components S457113. Chestnut Hill (MA): Boston College Department of Economics.

Jeanty PW. 2010b. SPWMATRIX: Stata module to generate, import, and export spatial weights, Statistical Software Components S457111. Chestnut Hill (MA): Boston College Department of Economics.

Jenness J. 2008. Surface Tools (surf_tools.avx) extension for ArcView 3.x. [Internet]. Version 1.6b. Flagstaff (AZ): Jenness Enterprises. [cited 2012 Dec 20]. Available from: http://www.jennessent.com/arcview/surface_tools.htm

Jennings MR, Hayes MP. 1994. Amphibian and reptile species of spe-cies concern in California. Final Report No. 8023. Rancho Cordova (CA): California Department of Fish and Game.

Johnstone JA, Dawson TE. 2010. Climatic context and ecological implica-tions of summer fog decline in the coast redwood region. Proc Natl Acad Sci USA. 107:4533–4538.

Jones OR, Wang J. 2010. COLONY: a program for parentage and sibship inference from multilocus genotype data. Mol Ecol Resour. 10:551–555.

Jost L. 2008. G(ST) and its relatives do not measure differentiation. Mol Ecol. 17:4015–4026.

Kalinowski ST. 2005. HP-RARE 1.0: a computer program for performing rarefaction on measures of allelic richness. Mol Ecol Notes. 5:187–189.

Kroll AJ. 2009. Sources of uncertainty in stream-associated amphibian ecol-ogy and responses to forest management in the Pacific Northwest, USA: a review. Forest Ecol Manag. 257:1188–1199.

Landguth EL, Cushman SA, Schwartz MK, McKelvey KS, Murphy M, Luikart G. 2010. Quantifying the lag time to detect barriers in landscape genetics. Mol Ecol. 19:4179–4191.

Legendre P. 1993. Spatial autocorrelation: trouble or new paradigm? Ecology. 74:1659–1673.

Lindenmayer DB, Fischer J. 2006. Habitat fragmentation and landscape change: an ecological and conservation synthesis. Washington (DC): Island Press.

Luikart G, Cornuet J-M. 1998. Empirical evaluation of a test for identify-ing recently bottlenecked populations from allele frequency data. Cons Biol. 12:228–237.

Manel S, Schwartz MK, Luikart G, Taberlet P. 2003. Landscape genetics: combining landscape ecology and population genetics. Trends Ecol Evol. 18:189–197.

Matsuda BM, Richardson JS. 2005. Movement patterns and relative abun-dance of coastal tailed frogs in clearcuts and mature forest stands. Can J Forest Res. 35:1131–1138.

McCune B. 2007. Improved estimates of incident radiation and heat load using non-parametric regression against topographic variables. J Veg Sci. 18:751–754.

McCune B, Keon D. 2002. Equations for potential annual direct incident radiation and heat load. J Veg Sci. 13:603–606.

Meirmans PG. 2006. Using the AMOVA framework to estimate a standard-ized genetic differentiation measure. Evolution. 60:2399–2402.

Meirmans PG, Hedrick PW. 2011. Assessing population structure: F(ST) and related measures. Mol Ecol Resour. 11:5–18.

Metter DE. 1967. Variation in the ribbed frog Ascaphus truei Stejneger. Copeia. 3:634–649.

Noss RF, editor. 2000. The redwood forest: history, ecology, and conserva-tion of the coast redwoods. Washington (DC): Island Press.

Olsen DH, Burnett KM. 2009. Design and management of linkage areas across headwater drainages to conserve biodiversity in forest ecosystems. Forest Ecol Manag. 258(Suppl):S117–S126.

Peery MZ, Kirby R, Reid BN, Stoelting R, Doucet-Bëer E, Robinson S, Vásquez-Carrillo C, Pauli JN, Palsbøll PJ. 2012. Reliability of genetic bottle-neck tests for detecting recent population declines. Mol Ecol. 21:3403–3418.

Petranka JW. 1998. Salamanders of the United States and Canada. Washington (DC): Smithsonian Institution Press.

Piry S, Alapetite A, Cornuet JM, Paetkau D, Baudouin L, Estoup A. 2004. GENECLASS2: a software for genetic assignment and first-generation migrant detection. J Hered. 95:536–539.

Piry S, Luikart G, Cornuet J-M. 1999. BOTTLENECK: a computer pro-gram for detecting recent reductions in the effective population size using allele frequency data. J Hered. 90:502–503.

Pritchard JK, Stephens M, Donnelly P. 2000. Inference of population struc-ture using multilocus genotype data. Genetics. 155:945–959.

Rannala B, Mountain JL. 1997. Detecting immigration by using multilocus genotypes. Proc Natl Acad Sci USA. 94:9197–9201.

Raymond M, Rousset F. 1995. An exact test for population differentiation. Evolution. 49:1280–1283.

Rosenberg NA. 2004. DISTRUCT: a program for the graphical display of population structure. Mol Ecol Notes. 4:137–138.

Rousset F. 2008. genepop’007: a complete re-implementation of the genepop software for Windows and Linux. Mol Ecol Resour. 8:103–106.

Savage WK, Fremier AK, Shaffer HB. 2010. Landscape genetics of alpine Sierra Nevada salamanders reveal extreme population subdivision in space and time. Mol Ecol. 19:3301–3314.

Sorensen R, Zinko U, Seibert J. 2006. On the calculation of the topographic wetness index: evaluation of different methods based on field observations. Hydrol Earth Syst Sc. 10:101–112.

Spear SF, Balkenhol N, Fortin MJ, McRae BH, Scribner K. 2010. Use of resistance surfaces for landscape genetic studies: considerations for param-eterization and analysis. Mol Ecol. 19:3576–3591.

Spear SF, Baumsteiger J, Storfer A. 2008. Newly developed polymorphic microsatellite markers for frogs of the genus Ascaphus. Mol Ecol Resour. 8:936–938.

Spear SF, Crisafulli CM, Storfer A. 2012. Genetic structure among coastal tailed frog populations at Mount St. Helens is moderated by post-distur-bance management. Ecol Appl. 22:856–869.

Spear SF, Storfer A. 2008. Landscape genetic structure of coastal tailed frogs (Ascaphus truei) in protected vs. managed forests. Mol Ecol. 17:4642–4656.

Spear SF, Storfer A. 2010. Anthropogenic and natural disturbance lead to differing patterns of gene flow in the Rocky Mountain tailed frog, Ascaphus montanus. Biol Cons. 143:778–786.

Stewart W. 2007. New economies of the redwood region in the 21st century. In: Standiford RB, Giusti GA, Valachovic Y, Zielinski WJ, technical coor-dinators. Proceedings of the Redwood Region Forest Science Symposium: what does the future hold? 2004 Mar 15–17; Rohnert Park, CA. Gen. Tech. Rep. PSW-GTR-194. Albany (CA): Pacific Southwest Research Station, USDA Forest Service. p. 393–401.

Stichbury G, Brabyn L, Green TGA, Cary C. 2011 Spatial modeling of wet-ness for the Antarctic Dry Valleys. Pol Res. 30:6330.

Stoddard MA, Hayes JP. 2005. The influence of forest management on headwater stream amphibians at multiple spatial scales. Ecol Appl. 15:811–832.

Storfer A, Murphy MA, Evans JS, Goldberg CS, Robinson S, Spear SF, Dezzani R, Delmelle E, Vierling L, Waits LP. 2007. Putting the “landscape” in landscape genetics. Heredity (Edinb). 98:128–142.

Tarboton DG. 1997. A new method for the determination of flow direc-tions and upslope areas in grid digital elevation models. Water Resour Res. 33:309–319.

Tarboton DG. 2003. Terrain analysis using digital elevation models in hydrology. Proceedings of the 23rd ESRI International Users Conference, 2003 Jul 7–11; San Diego, CA. Redlands (CA). [cited 2012 Dec 20]. Available from: http://proceedings.esri.com/library/userconf/proc03/p1038.pdf

Tarboton DG, Mohammed IN. 2010. TauDEM 5.0 [Internet]. [cited 2012 December 20]. Available from: http://hydrology.usu.edu/taudem/taudem5.0/


http://www.jennessent.com/arcview/surface_tools.htm

http://www.jennessent.com/arcview/surface_tools.htm

http://proceedings.esri.com/library/userconf/proc03/p1038.pdf

http://hydrology.usu.edu/taudem/taudem5.0/


216

Wahbe TR, Ritland C, Bunnell FL, Ritland K. 2005. Population genetic structure of tailed frogs (Ascaphus truei) in clearcut and old-growth stream habitats in southwest British Columbia. Can J Zool. 83:1460–1468.

Wallace RL, Diller LV. 1998. Length of the larval cycle of Ascaphus truei in coastal streams of the redwood region, Northern California. J Herpetol. 32:404–409.

Wang IJ. 2009. Fine-scale population structure in a desert amphibian: land-scape genetics of the black toad (Bufo exsul). Mol Ecol. 18:3847–3856.

Wang J. 2004. Sibship reconstruction from genetic data with typing errors. Genetics. 166:1963–1979.

Wang J, Santure AW. 2009. Parentage and sibship inference from multilocus genotype data under polygamy. Genetics. 181:1579–1594.

Weir BS, Cockerham CC. 1984. Estimating F-statistics for the analysis of population structure. Evolution. 38:1358–1370.

Welsh HH. 1990. Relictual amphibians and old-growth forests. Cons Biol. 4:309–319.

Welsh HH, Hodgson GR. 2008. Amphibians as metrics of critical biological thresholds in forested headwater streams of the Pacific Northwest, U.S.A. Fresh Biol. 53:1470–1488.

Welsh HH, Hodgson GR, Karraker NE. 2005a. Influences of the vegetation mosaic on riparian and stream environments in a mixed forest-grassland land-scape in “Mediterranean” northwestern California. Ecography. 28:537–551.

Welsh HH, Hodgson GR, Lind AJ. 2005b. Ecogeography of the herpeto-fauna of a northern California watershed: linking species patterns to land-scape processes. Ecography. 28:521–536.

Welsh HH, Lind AJ. 1991. The structure of the herpetofaunal assem-blage of the Douglas-fir/harwood forests of northwestern California and

southwestern Oregon. In: Ruggiero LF, Aubry KB, Carey AB, Huff MH, technical coordinators. Wildlife and vegetation of unmanaged Douglas-fir forests. Gen. Tech. Rep. PNW-GTR-285. Portland (OR): Pacific Northwest Research Station, USDA Forest Service. p. 394–413.

Welsh HH, Ollivier LM. 1998. Stream amphibians as indicators of eco-system stress: a case study from California’s redwoods. Ecol Appl. 8: 1118–1132.

Welsh HH, Roelofs TD, Frissell CA. 2000. Aquatic ecosystems of the red-wood region. In: Noss RF, editor. The redwood forest: history, ecology, and conservation of the coast redwoods. Washington (DC): Island Press. p. 165–199.

Whitlock MC. 2011. G′ST and D do not replace FST. Mol Ecol. 20:1083–1091.

Wilson GA, Rannala B. 2003. Bayesian inference of recent migration rates using multilocus genotypes. Genetics. 163:1177–1191.

Yang X, Chapman GA, Young MA, Gray JM. 2005. Using compound topo-graphic index to delineate soil landscape facets from digital elevation models for comprehensive coastal assessment. In: Zerger A, Argent RM, editors. MODSIM 2005 International Congress on Modelling and Simulation: advances and applications for management and decision making; 2005 Dec 12–15; Melbourne, Victoria. Canberra (Australia): Modelling and Simulation Society of Australia and New Zealand. p. 1511–1517. [cited 2012 Dec 20]. Available from: http://www.mssanz.org.au/modsim05/papers/yang_x.pdf

Received February 9, 2012; Revised December 5, 2012; Accepted December 6, 2012

Corresponding Editor: Andrew Crawford


http://www.mssanz.org.au/modsim05/papers/yang_x.pdf

Elevated Genetic Structure in the Coastal Tailed Frog (Ascaphus truei)

Documents

Transcript of Elevated Genetic Structure in the Coastal Tailed Frog (Ascaphus truei)