Multilocus Phylogeography of the Flightless Darkling Beetle

Biological Journal of the Linnean Society, 2010, 99, 424–444. With 5 ﬁgures

Multilocus phylogeography of the ﬂightless darkling beetle Nyctoporis carinata (Coleoptera: Tenebrionidae) in the California Floristic Province: deciphering an evolutionary mosaic

MAXI POLIHRONAKIS* and MICHAEL S. CATERINO

Department of Invertebrate Zoology, Santa Barbara Museum of Natural History, 2559 Puesta del Sol Road, Santa Barbara, CA 93105, USA

Received 16 July 2009; accepted for publication 9 September 2009bij_1360 424..444

The California Floristic Province (CFP) is considered a global biodiversity hotspot because of its confluence of high species diversity across a wide range of threatened habitats. To understand how biodiversity hotspots such as the CFP maintain and generate diversity, we conducted a phylogeographic analysis of the flightless darkling beetle, Nyctoporis carinata, using multiple genetic markers. Analyses of both nuclear and mitochondrial loci revealed an east–west genetic break through the Transverse Ranges and high genetic diversity and isolation of the southern Sierra Nevada Mountains. Overall, the results obtained suggest that this species has a deep evolutionary history whose current distribution resulted from migration out of a glacial refugium in the southern Sierra Nevada via the Transverse Ranges. This finding is discussed in light of similar genetic patterns found in other taxa to develop a foundation for understanding the biodiversity patterns of this dynamic area. © 2010 The Linnean Society of London, Biological Journal of the Linnean Society, 2010, 99, 424–444.

ADDITIONAL KEYWORDS: cytochrome oxidase I (COI/CoxI) – guftagu – southern Sierra Nevada.

INTRODUCTION will develop a better understanding of how these regions accumulate and maintain diversity (Avise, The California Floristic Province (CFP) is simulta- 2000). The dynamic geographic history and high neously a region of high biodiversity and one in which habitat diversity of the CFP are frequently invoked as development and other land uses severely threaten causes of increased species diversification and have parts of the native biota. As such, it has been recog- been used to explain phylogenetic patterns across nized as a globally significant biodiversity hotspot many plant and vertebrate taxa (Calsbeek, Thompson by Conservation International in an effort to raise & Richardson, 2003). Recent comparative phylogeo- awareness and establish its needs as a top conserva- graphic analyses of the CFP have revealed relatively tion priority (Conservation International, 2009). The concordant spatial barriers and high genetic diversity CFP comprises approximately 70% of the state of related to the orogeny of the Transverse Ranges, California and extends north across the state bound- Tehachapi Mountains, and the Sierra Nevada (Cals- ary into Oregon and south into the Baja California beek et al., 2003; Feldman & Spicer, 2006; Rissler peninsula. Understanding the range of diversity et al., 2006; Chatzimanolis & Caterino, 2007a), and patterns of species in this region is essential for have used this information to identify diversification preservation of its delicate ecological systems. By ‘hotspots’ within the larger CFP ‘hotspot’ (Davis et al., reconstructing the evolutionary history of species and 2008). Although these generalized patterns provide a integrating this information with geographic data, we useful comparative framework for phylogeographic investigation of the CFP, fine-scale analyses of *Corresponding author. E-mail: [email protected] patterns within species continue to provide the

424 © 2010 The Linnean Society of London, Biological Journal of the Linnean Society, 2010, 99, 424–444 PHYLOGEOGRAPHY OF N. CARINATA 425 foundation for our understanding of how biodiversity isms exhibit a north–south genetic break attributed patterns arise in this region because concordance of to the Transverse Ranges (Maldonado, Vila & genetic barriers across broad spatial scales among Wayne, 2001; Calsbeek et al., 2003; Sgariglia & taxa does not necessarily imply parallel patterns of Burns, 2003; Forister, Fordyce & Shapiro, 2004; local adaptation and diversification (Feldman & Feldman & Spicer, 2006; Chatzimanolis & Caterino, Spicer, 2006). 2007b), the mtDNA gene tree of N. carinata In the past decade, the Transverse Ranges have revealed a prominent east–west break in the middle received much of the attention of phylogeographers in of the Transverse Ranges (as also observed in Cali- the CFP (Calsbeek et al., 2003; Lapointe & Rissler, fornia mountain king snakes: Rodríguez-Robles et al. 2005; Chatzimanolis & Caterino, 2007a). Their 1999; pond turtles: Spinks & Shaffer, 2005; and unusual orientation, extending from the California Sepedophilus rove beetles: Chatzimanolis & coastline inland across southern California, and the Caterino, 2007a). Second, the deepest split in the N. diversity of habitats that they encompass have been carinata mtDNA gene tree separated the southern suggested as primary contributors to phylogeographic Sierra Nevada and Tehachapi populations from the complexity in this region. Although this intriguing central and southern California populations, corre- area deserves further attention, empirical studies sponding to the Sierran break discussed above. In yielding interesting results from populations in the addition, one of two sampling localities in the Teh- southern Sierra Nevada and Tehachapi Mountains achapi Mountains had two highly divergent mtDNA have also been accumulating (Macey et al., 2001; haplotypes (0.11 uncorrected, 0.25 corrected) with Segraves & Pellmyr, 2001; Jockusch & Wake, 2002; one haplotype sharing a more recent common ances- Matocq, 2002; Feldman & Spicer, 2006; Kuchta & tor with populations to the south, and the other Tan, 2006; Chatzimanolis & Caterino, 2007a, b; Davis being more closely related to populations in the et al., 2008; Caterino & Chatzimanolis, 2009). Many north. These results suggest that high diversity in of these studies provide evidence for deep divergence this region may result from mtDNA contact zone between southern Sierra clades and coastal and more between formerly isolated populations. A detailed southerly populations, leading some studies to recog- multilocus analysis investigating these patterns will nize the southern Sierras as a glacial refugium (Rich, add to a growing body of research aiming to under- Thompson & Fernandez, 2008), although evidence for stand and document the biodiversity of these impor- this is equivocal (Kuchta & Tan, 2006). Divergence tant ecological regions. times estimated for a Sierra Nevada break range The present study uses two independently evolving from approximately 2 Mya (Feldman & Spicer, 2006) sequence markers to evaluate phylogeographic to 9.9 Mya (Segraves & Pellmyr, 2001), further hypotheses using N. carinata. It has become apparent emphasizing that concordant spatial genetic patterns that the validity of phylogeographic hypotheses is may not imply concurrent chronology (Feldman & increased when biparentally inherited markers from Spicer, 2006), especially when long persisting geo- the nuclear genome are used in conjunction with graphical features are subject to climactic fluctua- uniparentally inherited mitochondrial markers. Theo- tions over an extended period of time. This long and retical and empirical studies have documented both dynamic history coupled with increased habitat com- stochastic and deterministic factors that limit the plexity resulting from the extremely varied topogra- ability of mtDNA to accurately represent the history phy of the Sierra Nevada highlight the importance of of a species (Irwin, 2002; Zhang & Hewitt, 2003; this area as a centre of genetic diversity and lineage Godinho, Crespo & Ferrand, 2008). Most significantly, diversification in the CFP. these include asymmetrical migration rates of males Comparative phylogeography demands a broad and females (Jockusch & Wake, 2002; Miller, Haig & perspective, based on many taxa and genetic Wagner, 2005; Berghoff et al., 2008; Kerth et al., 2008; markers. To date, our understanding of regional Ngamprasertwong et al., 2008), genetic drift, and/or patterns in the CFP is based heavily on both ver- lineage sorting (Maddison, 1997), with all of them tebrates and mitochondrial DNA. In the present potentially resulting in discordance among markers. study, we expand both perspectives in a multilocus Multilocus analysis of N. carinata phylogeography analysis of the flightless darkling beetle Nyctoporis provides a robust basis to test several hypotheses and carinata LeConte (Tenebrionidae) throughout its to examine the generality of patterns documented range in central and southern California. Previous to date. Specifically, we investigate whether both comparative analysis of N. carinata using mitochon- marker types support: (1) the east–west geographic drial DNA (mtDNA) (Caterino & Chatzimanolis, break through the middle of the Transverse Ranges 2009) revealed several interesting patterns within (sensu Chatzimanolis & Caterino, 2007a); (2) the iso- recognized evolutionary hotspots of the CFP (Davis lated southern Sierra Nevada clade with putative et al., 2008). First, althlough many flightless organ- contact to the Transverse Ranges through the

Tehachapi Mountains; (3) designated evolutionary as a potential outgroup. The differences between N. hotspots within the CFP (Davis et al., 2008); and (4) vandykei and N. carinata, on the other hand, are very lineages of cryptic species within N. carinata minor, and preliminary mitochondrial analyses did (Caterino & Chatzimanolis, 2009). not clearly separate these two species; thus, we considered them as probable synonyms in the present study. All specimens were collected directly into 100% MATERIAL AND METHODS ethanol and stored at -70 °C prior to extraction. FIELD COLLECTION Collection localities are grouped into regional ‘popu- Nyctoporis carinata is abundant and widely distrib- lations’ in accordance with previous studies (Fig. 1) uted throughout much of central and southern Cali- (Chatzimanolis & Caterino, 2007a; Caterino & Chat- fornia, and can be found in detritus material and zimanolis, 2009). These areas correspond to geologi- under rocks and loose debris. Both sexes of N. cari- cally discrete subunits of the mountainous areas nata lack wings and are thus ﬂightless. Specimens surrounding the Transverse Ranges, although bound- were collected throughout the Transverse Ranges, aries between a few, particularly in the western central and southern Sierra Nevada, and the Santa Transverse Ranges, are less pronounced. Most of Lucia mountains of the outer coast ranges. Most these, excluding the Santa Lucia Mountains to the specimens clearly represent the nominal species N. north-west, and the Tehachapi, Breckenridge and carinata LeConte. We sampled a small number of Sierra Nevada Mountains to the north-east, collec- specimens from the central and southern Sierra tively constitute the Transverse Ranges (Fig. 1). Nevada which correspond to Nyctoporis vandykei Blaisdell, and one specimen from Lake County (north of the San Francisco Bay) representing Nyctoporis FOCAL GENES aequicollis Eschscholtz. Based on morphology, the Genes from both the mitochondrial and nuclear latter appears quite distinct, and was initially treated genome were targeted in this study to compare infor-

3.1 3.2

1.1 3

4.1 2.2 4 2.1 5 2 5.1

7.2 8.1 7.1 7.4 7.3 7.5 6.1 9.1 6.2 6.3 6.4 10.1 6.5 10.2 11.1 8 10.3 6 7 9 11.2 11.3 13.1 10 11 12.1 13 12

Figure 1. Map of southern California with population designations and speciﬁc collecting localities as coded in the Appendix (i.e. 1.1 = locality one from population one; 1.2 = locality two from population one, etc.) (1, northern Santa Lucia Mountains; 2, southern Santa Lucia Mountains; 3, south-western Sierra Nevada; 4, Breckenridge Mountains; 5, Tehachapi Mountains; 6, Santa Ynez Mountains; 7, north-west Transverse Ranges; 8, central Transverse Ranges; 9, Sierra Pelona; 10, San Gabriel Mountains; 11, San Bernardino Mountains; 12, San Jacinto Mountains; 13, Santa Cruz Island). A few numbers cover multiple proximate populations, indistinguishable at this scale. Populations 6–12 comprise the entire Transverse Range region.

© 2010 The Linnean Society of London, Biological Journal of the Linnean Society, 2010, 99, 424–444 PHYLOGEOGRAPHY OF N. CARINATA 427 mation from two independently evolving loci. The QUIN, version 3.1 (Excoffier, Laval & Schneider, well-studied mitochondrial protein coding cytochrome 2005). Significance was evaluated by comparing the oxidase I (COI) gene has been used in a large number observed test statistic to a distribution generated of phylogeographic and phylogenetic analyses, espe- from 1000 permutations of the original data such that cially among insect and arthropod groups (Caterino, P-values represent the proportion of the distribution Cho & Sperling, 2000). The nuclear marker used was that is less than or equal to the observed value. from an intron in the Guftagu (GFT) gene, identified Recombination in the nuclear gene was assessed during a screening of Tribolium introns (M. S. using the PHI (pairwise homoplasy index) statistic Caterino, M. Polihronakis, I. Ouzounov, unpubl. data) (Bruen, Phillipe & Bryant, 2006) implemented using provided by V. Krauss (Krauss et al., 2008). In Droso- the software SPLITSTREE, version 4.10 (Huson, phila, the GFT gene is important during the devel- 1998; Huson & Bryant, 2006). opment of adult morphological structures and is primarily associated with the external sensory organs. The targeted intron is located after the first ANALYSES OF POPULATION STRUCTURE AND region of coding sequence relative to both Drosophila SPATIAL AUTOCORRELATION and Tribolium annotated genomes. To test for significant genetic breaks across certain geographic barriers, we performed analysis of molecular variance (AMOVA) and spatial autocorre- GENERATING DNA SEQUENCE DATA lation analyses. The degree of population divergence

Genomic DNA was extracted from forebodies (head and haplotype exclusivity was determined using FST and prothorax) of field collected specimens using values (which differ from traditional FST values DNeasy Tissue extraction kits (Qiagen). Apart from because they incorporate both overall similarity and dissection, extractions were minimally destructive, haplotype frequencies; Excoffier, Smouse & Quattro, and chitinous parts were mounted as vouchers after 1992), as calculated in ARLEQUIN. Spatial autocor- extraction. Full collection data, including voucher relation analyses were performed to examine the numbers and corresponding DNA extractions, can be relationship between genetic and geographic dis- accessed through the California Beetle Project data- tance at the individual level, and to compare base (http://www.sbnature.org/calbeetles). Amplifica- markers for possible differences related to sex-biased tion of the latter half of COI was performed using dispersal. These analyses provide a finer-scale evalu- previously described polymerase chain reaction (PCR) ation of isolation-by-distance (IBD) than matrix cor- protocols and reagents (Caterino & Chatzimanolis, relation tests because it is possible to assess the 2009) with the primers C-J-2183 and C1-N-3014 behavior of the slope of autocorrelation values as (Simon et al., 1994). Amplification of the GFT intron the size of distance classes increase. These were required a nested PCR reaction with an initial ampli- performed in GENEALEX, version 6.1 (Peakall & fication using degenerate primers GFT37F (5′-AAA Smouse, 2006), with each marker treated individu- ATG MGN ATH CGN GCN TTT CC-3′) and GFT86R ally as a single population using the even sample (5′-TCN ACA TAC TTY TCR TCC AT-3′). The nested size option with 20 distance classes. Significance of amplification was done using an internal species the autocorrelation coefficient was assessed using specific forward primer located approximately 35 bp 999 permutations of the data in addition to 1000 downstream of the degenerate forward primer bootstrap replicates. (NycaGFT) with the initial reverse primer. PCR prod- Of the seven evolutionary hotspots designated by ucts were purified using Qiaquick PCR Purification Davis et al. (2008), our samples cover three of them kits (Qiagen) or ExoSAP-IT® (USB Corp.). Forward (Santa Lucia Mountains; north-eastern Transverse and reverse sequencing reactions and sequence visu- Ranges, Tehachapi and Piute Mountains; and San alization were performed by Macrogen, Inc. (Korea). Bernardino and San Jacinto Mountains). To deter- All DNA sequences were edited in GENEIOUS Pro, mine whether genetic diversity within N. carinata version 4.5.4 (Biomatters Ltd) and aligned in SE-AL, was consistently higher or more divergent in these version 2.0a11 (http://tree.bio.ed.ac.uk/software/seal/). areas, we generated a genetic landscape shape for Both genes were easy to align manually; the COI gene each of the markers using the software ALLELES IN did not have any indels, and the GFT intron had SPACE (AIS) (Miller, 2005). The genetic landscape several small indels that were accommodated by interpolation method implemented in AIS uses a inserting gaps. Sequence identity was checked by Delaunay triangulation-based connectivity network performing BLAST searches in GenBank (Altschul from X and Y geographic coordinates and assigns real et al., 1997). and interpolated genetic distances to the Z-axis to To test for patterns of selection at either locus, produce a three dimensional landscape (Miller et al., Tajima’s D (Tajima, 1989) was estimated in ARLE- 2006), which provides a visual representation of

© 2010 The Linnean Society of London, Biological Journal of the Linnean Society, 2010, 99, 424–444 428 M. POLIHRONAKIS and M. S. CATERINO genetic diversity patterns across a sample area. in GFT, where both haplotypes from this individual Unlike AMOVA, this method requires no a priori were minimally different from those in many popula- group designations. Genetic landscapes were interpo- tions of N. carinata. Therefore, we consider the lated using raw genetic distances and a 50 ¥ 50 grid species status and relationship of this entity uncer- with a distance weight value of one. Sensitivity of the tain and do not treat it as an outgroup. For this analysis to changes in grid size, weighting parameter reason, our trees should be interpreted as unrooted, values, and genetic distance measure were evaluated, although ‘midpoint rooting’ (as implemented in but resulted in minimal changes to the resulting PAUP*) is used for presentation. landscape plot. We also calculated genetic diversity within populations using uncorrected and corrected distances, as calculated in PAUP* (Swofford, 2002). RESULTS SEQUENCE DATA PHYLOGENETIC ANALYSIS AND A total of 130 and 109 individuals were sequenced for NETWORK CONSTRUCTION the COI (826 bp) (Genbank accessions EU37099- Relationships among individuals or haplotypes were EU37190 and GU049332-GU049339) and GFT inferred using phylogenetic and network methods to (497 bp) (Genbank accessions GU049270-GU049331) obtain both bifurcating and nonbifurcating perspec- genes, respectively. Of the 109 samples with GFT tives of the gene lineages. Models of molecular evolu- sequence, 34 had polymorphic sites (Appendix S1). tion for each individual marker were evaluated Individuals with one polymorphic site were phased in MRMODELTEST, version 2.3 (Nylander, 2004). manually and both haplotypes were included in the According to the Akaike information criterion, the COI analysis. Specimens with two or more polymorphic data best fit the HKY + I + G model, and the GFT data sites were phased using the software PHASE, version best fit the GTR + G model. Gene trees were inferred 2.1 (Stephens, Smith & Donnely, 2001; Stephens & in MRBAYES, imposing the model specified by AIC, Scheet, 2005). There were three cases that resulted in using default priors for all parameters. Branch length ambiguous phase calls and so autapomorphic poly- priors for each marker were determined through com- morphic sites for these individuals were coded as parison of branch lengths of maximum likelihood trees missing data. For all three cases, this resulted in just generated in GARLI, version 0.96 (Zwickl, 2006) to one parsimony informative polymorphic site that those obtained in the Bayesian analyses. Haplotype was phased manually. Thus, further discussion of networks were built using COMBINE TREES sequences from both genes will refer to haplotypes. (Cassens, Mardulyn & Milinkovitch, 2005), which The mean Tajima’s D-values for COI and GFT implements the union of maximum parsimonious estimated across all populations were not significant trees approach (UMP), using parsimony trees as (P = 0.55 and 0.52, respectively). At the population input. For the nuclear data set, parsimony trees were level, the northern Santa Lucia population had a generated with gaps coded as a fifth state, using a significant Tajima’s D-value (P = 0.01) for the COI heuristic search in PAUP* (Swofford, 2002) with tree gene, and the north-west Transverse Range popula- bisection–reconnection branch swapping and 500 rep- tion had a significant Tajima’s D-value (P = 0.05) for licates of random sequence addition, saving 200 trees the GFT gene, potentially indicating past selection or per replicate. Heuristic search options for the mito- population bottlenecks for these two populations. chondrial data set were modified to accommodate the There was no evidence for recombination in the GFT large number of trees generated (500 replicates of intron according the PHI test (P = 0.80). random sequence addition saving a maximum of 20 The latter half of the COI gene targeted for ampli- trees per replicate). Because of the computational fication had 224 variable sites, with 207 of those demands of the COMBINE TREES software, the first being parsimony informative. The COI data set had a 1000 trees from the parsimony search were analysed total of 100 unique haplotypes with no one haplotype in two sets of 500 trees each. Ten networks were dominating more than 10% of the entire sample. Of constructed for each data set which were then used to the 497 bp sequenced of GFT, 53 were variable, with generate a ‘best network’ with the fewest number of 27 of those being parsimony informative. The GFT cycles (Cassens et al., 2005). The best networks for data set had a total of 62 unique haplotypes, although each set of 500 trees were compared to ensure concor- the three most common haplotypes dominated 49% of dance when only subsets of trees from the parsimony the data set. Interestingly, approximately 20% of the search were used. haplotype diversity in the GFT data came from within Although mitochondrial DNA showed N. aequicollis the south-western Sierra population, whereas this to be rather divergent from other samples, tentatively population only accounted for 9% of the haplotype supporting its outgroup status, this was not the case diversity of the COI gene.

POPULATION STRUCTURE AND (Fig. 2). A decreasing slope of the spatial autocorre- AUTOCORRELATION ANALYSIS lation graphs is indicative of an IBD trend (i.e. autocorrelation values decrease as distance class High differentiation among populations was sup- size increases), whereas significant r-values indicate ported by both markers. All FST values based on the higher or lower genetic relatedness than predicted if COI data were significant, except for some of the individuals were randomly distributed (Smouse & comparisons with the Breckenridge Mountains popu- Peakall, 1999; Peakall, Ruibal & Lindenmayer, lation where only two individuals were sampled 2003). COI haplotypes showed significant IBD with

(Table 1, lower diagonal). Pairwise FST values based decreasing r-values from 0–165 km. The increase on the GFT locus were also almost all significant in r-values at approximately 200 km likely reflects (Table 1, upper diagonal), with exceptions between the close relationship of some individuals from the Sierra Pelona and central Transverse Range the central Transverse range population with indi- populations, as well as between the southern Santa viduals from the San Bernardino and San Jacinto Lucia and north-west Transverse Range populations. populations. Beyond 218 km, autocorrelation values Both of these insignificant values are between were relatively unchanging and significantly nega- geographically adjacent areas, suggesting gene flow tive. The GFT locus exhibited a very similar between them, although insufficient sampling cannot pattern, showing a general IBD trend up to appro- be ruled out as an explanation. ximately 150 km, at which point r-values re- Spatial autocorrelation analyses showed a strong mained negative with some fluctuation. The similar decline in relatedness with increasing distance relationships between genetic and geographic A) 0.8 0.7

0.6

0.5

0.4 r 0.3 0.2

0.1

0.0

-0.1

-0.2

-0.3 29 49 67 86 98 117 140 150 165 178 205 218 235 255 270 295 317 368 477 Distance (km) B) 0.7 0.6

0.5

0.4

0.3

r 0.2

0.1

0.0

-0.1

-0.2

-0.3 29 51 69 90 107 121 145 159 178 204 217 235 252 270 293 313 346 453 Distance (km)

Figure 2. Spatial autocorrelation graphs of COI haplotypes (A) and GFT haplotypes (B). Dotted lines represent the 95% conﬁdence interval with respect to the null hypothesis that individuals are randomly distributed.

Table 1. Pairwise population FST comparisons for COI (lower diagonal) and GFT (upper diagonal)

00TeLnenSceyo London, of Society Linnean The 2010 © Northern Southern Southwestern Santa Northwest Central San San San Northern Santa Santa Sierra Breckenridge Tehachapi Ynez Transverse Transverse Sierra Gabriel Bernardino Jacinto Channel Lucias Lucias Nevada Mountains Mountains Mountains Ranges Ranges Pelona Mountains Mountains Mountains Islands

Northern Santa – 0.29 0.81 – 0.58 0.32 0.30 0.73 0.70 0.52 0.66 0.70 0.43 Lucias Southern Santa 0.80 – 0.81 – 0.53 0.22 0.05 0.80 0.75 0.46 0.66 0.70 0.32 Lucias Southwestern 0.75 0.65 – – 0.79 0.77 0.85 0.82 0.81 0.76 0.86 0.87 0.82 Sierra Nevada Breckenridge 0.95 0.83 0.36 – – –––––––– Mountains Tehachapi 0.79 0.71 0.39 0.41 – 0.46 0.57 0.12 0.12 0.26 0.70 0.74 0.59 Mountains Santa Ynez 0.56 0.27 0.60 0.66 0.62 – 0.27 0.49 0.48 0.43 0.16 0.18 0.31 Mountains

ilgclJunlo h ina Society Linnean the of Journal Biological Northwest 0.73 0.31 0.70 0.81 0.73 0.32 – 0.68 0.66 0.53 0.64 0.66 0.30 Transverse Ranges Central 0.90 0.75 0.72 0.96 0.77 0.44 0.71 – 0.04 0.25 0.85 0.89 0.83 Transverse Ranges Sierra Pelona 0.75 0.56 0.64 0.80 0.69 0.19 0.59 0.62 – 0.25 0.83 0.87 0.78 San Gabriel 0.72 0.61 0.71 0.82 0.73 0.25 0.63 0.60 0.23 – 0.63 0.66 0.51 Mountains San Bernardino 0.75 0.63 0.69 0.83 0.73 0.39 0.63 0.65 0.47 0.48 – 0.11 0.70 Mountains San Jacinto 0.63 0.62 0.68 0.80 0.69 0.43 0.64 0.72 0.56 0.57 0.57 – 0.75 Mountains Northern 0.90 0.58 0.68 0.94 0.75 0.37 0.48 0.88 0.70 0.68 0.70 0.71 – Channel Islands 2010, , Bold values are signiﬁcantly different from zero (P < 0.01). 99 424–444 , PHYLOGEOGRAPHY OF N. CARINATA 431

A) B)

edg

Y (Easte

X (Southern edg e)

astern e

Y (

X (Southe rn edge)

Figure 3. Landscape shapes from ALLELES IN SPACE analyses. A, connectivity network based on Delaunay triangulation overlayed onto map of southern California (the vertex of each triangle represents a collecting locality). B, landscape interpolation of COI sequence data. C, landscape interpolation of GFT sequence data. The Z-axis in (B, C) represents the degree of genetic discontinuity between two localities. distance between the two genes suggest that dis- sent areas of high genetic diversity. On the other persal is not sex-biased. hand, when localities are further apart (long lines Comparison of the genetic landscape shapes for the between vertices), high peaks are more appropriately two genes provided a visual representation of the interpreted as genetic barriers. Both markers demon- genetic diversity patterns observed over the sampling strated high diversity among the south-western area (Fig. 3). Peaks on each landscape shape repre- Sierra populations as a result of high genetic diver- sent the degree of genetic discontinuity between two gence among several geographically proximate locali- localities as determined by the Delaunay triangula- ties, as well as a barrier across the Central Valley. tion method (Miller et al., 2006). Thus, when collect- The landscape based on the COI data (Fig. 3B) also ing localities are geographically close in proximity exhibited genetic discontinuity in the Transverse (short distances between vertices), high peaks repre- ranges, beginning at the southern tip of the Central

Valley separating the Tehachapis, Breckenridge clade, although two San Jacinto individuals (haplo- Mountains, and south-western Sierra populations type N21) were more closely related to a clade of from the eastern and western Transverse Range primarily San Bernardino (the neighbouring region populations. The plot of the GFT data (Fig. 3C) had a to the north) haplotypes. The large clade containing barrier that was not exhibited by the COI data the remaining eastern Transverse Range samples between the southern Sierra Nevada and northern contained San Bernardino, San Gabriel, and Sierra Transverse Ranges. Thus, the areas of high genetic Pelona haplotypes intermingled with minimal site diversity in N. carinata corresponded to the Trans- ﬁdelity. Although the San Bernardino haplotypes verse Ranges and southern Sierra hotspots desig- were the most cohesive of these, two of these indi- nated by Davis et al. (2008). However, in contrast to viduals were more closely related to haplotypes from Davis et al. (2008) our samples from the Santa Lucia, elsewhere in this broad region. The eastern Trans- San Bernardino, and San Jacinto Mountains did not verse Range clade included haplotypes from the exhibit increased levels of diversity. central Transverse Ranges (the Tecuya Trail locality in the San Emigdio Mountains), as well as one haplotype from the Tehachapis. Another interesting PHYLOGENETIC ANALYSIS AND element of this larger eastern Transverse Range clade NETWORK RECONSTRUCTION was a subclade of haplotypes from further west in mtDNA Ojai and the eastern-most Santa Ynez Mountains (the Analyses of COI revealed very high levels of divergence Murrietta Trail locality). Although the area covered among haplotypes, especially among regions. Uncor- by the eastern Transverse Range clade is more or less rected distances were as high as 0.11 uncorrected (0.27 contiguous, it is the most wide-ranging, with several corrected), or 0.07 uncorrected (0.11 corrected) if the extensions beyond its core area. The mainly western nominally distinct populations of N. aequicollis and N. Transverse Range clade included all individuals from vandykei are excluded. The UMP method of network the north-western Transverse Range and Santa Cruz building resulted in one network but with a large Island populations, as well as all individuals from the number of unsampled nodes. Thus, because of the high southern Santa Lucia Mountains. Most individuals haplotype diversity and non-network-like behavior of from the Santa Ynez Mountains were also included, the COI gene, these data are presented as a phyloge- with the exception of the Murrietta Trail samples netic tree only. The COI gene tree showed ﬁve main mentioned above. clades, principally: the western Transverse Ranges (most) + Channel Islands + southern Santa Lucias; GFT eastern, central, and small inclusions of western Divergence among GFT alleles was much lower than Transverse Ranges and Tehachapis; San Jacinto that seen in mtDNA. Most haplotypes differed by only Mountains (most); northern Santa Lucias; and Sierra one or a few changes. For this reason, the phylogenetic Nevada (including the Breckenridge Mountains) + tree based on these sequences was largely unresolved. Tehachapis (most) (Fig. 4). The most strongly sup- With such low divergence, the likelihood of direct ported relationship among the major clades linked ancestor–descendant relationships among sequences the eastern and western Transverse Range clades was high, and we therefore based our inferences on (PP = 1.0). This supported a single lineage covering the the UMP network (Fig. 5). Although genetic variation bulk of the Transverse Ranges with two exceptions: (1) was generally low, three of the four main groups in it did not include the San Jacinto Mountains and (2) it this network were recognized by a high frequency of did include the southern part of the Santa Lucia identical haplotypes over broad (mostly) contiguous Mountains, which is not considered part of the Trans- geographic regions, along with linked derivative verse Ranges. The break between these two major haplotypes. The most widespread of these was the Transverse Range groups corresponds loosely to the western group (orange), including all haplotypes from Ventura and Cuyama River drainages in the lower the northern and southern Santa Lucias, north- elevations. However, at higher elevations, no such western Transverse Ranges, Santa Ynez Mountains, obvious geographic barrier is observed. Santa Cruz Island, the central Tranverse Ranges, and Most populations were predominantly single lin- a single individual from the San Gabriel Mountains; eages. However, almost all had individuals appearing the central group (green) included most haplotypes elsewhere in the tree. The only two monophyletic from the Sierra Pelona, San Gabriel Mountains, Teh- populations were those from the northern Santa achapi Mountains, and the northern California repre- Lucia Mountains (the north-western part of our sam- sentative of N. aequicollis (G58 & G59); the eastern pling region) and from Santa Cruz Island. Most indi- group (blue) included all haplotypes from the San viduals from the San Jacinto Mountains population Jacinto Mountains, all but one from the San Bernar- (at the south-eastern extreme) fell out in an isolated dino Mountains, one haplotype from the San Gabriel

Figure 4. A, COI gene tree inferred in MRBAYES (*posterior probability Ն 0.80). B, corresponding clade designations on sample map. Arrow from the yellow clade refers to the placement of the Nyctoporis aequicollis specimen from the northernmost Lake County sampling area (not visible on map).

Mountains, and a disjunct group of haplotypes from Mountains, whereas GFT haplotypes in these areas the Ojai Valley and extreme eastern end of the Santa were more closely allied with those further east, in Ynez Mountains. The fourth main group (yellow) rep- the San Bernardino Mountains population. Although resented a diverse clade of related, mostly unique, the region where these discontinuities are observed haplotypes from the southern Sierra Nevada region. more-or-less coincides with the conﬂuence of several Both the COI and GFT gene trees supported an major drainages (principally the Ventura and Santa east–west break in the Transverse Ranges in the Clara Rivers), it is only approximate, and neither eastern Santa Ynez Mountains/western Ventura corresponds precisely to the Santa Clara break found County area. However, the genetic break between the by Chatzimanolis & Caterino (2007a) in Sepedophilus two clades supported by the COI data was positioned beetles, and among other species. slightly west of the break supported by the GFT data The results from an analysis of the COI and GFT set. In addition, the geographic relationships of popu- markers differed in many other respects. Although lations within each clade were somewhat different both gene trees revealed a few large clades with broad between genes in that the COI haplotypes from the geographic continuity, the exact regions deﬁned eastern group were almost all closely related to hap- by these clades showed little congruence. The GFT lotypes found in the Sierra Pelona and San Gabriel network did not reveal a strong relationship between

Figure 5. A, GFT union of maximum parsimonious trees approach network; sizes of boxes within coloured clades scaled to haplotype frequency, numbers in parentheses denote the frequency of the haplotype if > 1 (haplotypes in some haplotype boxes have different numbers as a result of missing data but are otherwise identical). B, corresponding clade designations on sample map. Arrow from the green clade refers to the placement of the Nyctoporis aequicollis specimen from the northernmost Lake County sampling area (not visible on map); ‘X’ in population four denotes no nuclear data are available for this region.

Sierran and Tehachapi populations as seen with COI, The COI sequences of N. carinata were character- instead grouping the Tehachapis strongly with the ized by high divergence, high haplotype diversity, low central portion of the Transverse Ranges, the San population connectedness, broad geographic cohesive- Gabriel Mountains, and the Sierra Pelona (haplotype ness, and a combination of mono- and polyphyletic G01 is the most common in all these areas). Sierran populations. This pattern signifies an old, broadly- populations were more closely related to Tehachapi distributed species with historically-isolated popula- individuals than to any other group, but were highly tions, some of which have come into secondary contact divergent relative to the minimal divergence observed with subsequent gene flow (Jockusch & Wake, 2002). throughout the remainder of the network. Most indi- This was supported by the overall non-monophyly of viduals from the San Jacinto and San Bernardino populations with otherwise high FST values. Interest- Mountains shared a GFT haplotype, whereas most ingly, geographically peripheral populations from the San Jacinto COI haplotypes were highly divergent. southern Sierra Nevada, the northern Santa Lucia Almost all northern Santa Lucia individuals had a Mountains, and the San Jacinto Mountains were the GFT sequence identical to those from the southern most isolated and exclusive (with one individual Santa Lucias, as well as to most individuals from the exception in the latter case), whereas more centrally north-western Transverse Ranges, and Santa Ynez located populations showed higher levels of admix- Mountains, whereas the COI data supported the ture. This suggests that, if populations had remained northern Santa Lucia population as being exclusive completely isolated, they would have had enough time and highly divergent. Finally, although the Lake to achieve monophyly. Monophyly of the Santa Cruz County specimen (corresponding to N. aequicollis) Island population can also be cited in this respect differed by only a single mutation in GFT from the because, although founder effects would be likely to eastern Sierra Pelona group, its COI sequence was explain monophyly for this disjunct population, there highly divergent from all other groups. was no evidence for a bottleneck based on estimates of Tajima’s D. The GFT data supported a somewhat different sce- DISCUSSION nario. The results obtained based on this marker Detailed comparative phylogeographic analyses showed low levels of divergence and low haplotype provide a means to study how geography influences diversity within and across most populations (with a the evolutionary history and biodiversity patterns of few common haplotypes shared among, and dominat- organisms. Within the CFP, there are many geo- ing most populations), but were similar to mitochon- graphic barriers that are hypothesized to have a drial results in low population connectedness, and pervasive effect on species across different taxonomic showed similar patterns of geographic relationships groups. Although evidence supporting this hypothesis among populations. This overall pattern is consistent is increasing, much of the data are focused on verte- with a series of relatively rapid, stepwise coloniza- brates, and, until recently, heavily dependent on tions from one or more ancestral pools of diversity single locus analyses of mitochondrial DNA. The into more peripheral areas, with historical continuity present study aimed to evaluate the phylogeographic between populations that now appear to be isolated. history of a dynamic part of the CFP through analysis of N. carinata, a flightless darkling beetle with East–west Transverse Range break a widespread distribution in central and southern The COI gene tree and GFT gene network both sup- California, using mitochondrial and nuclear genes. ported an east–west genetic break through the Trans- verse Ranges, although they did not completely overlap. The location of this break based on the COI THE PHYLOGEOGRAPHIC HISTORY OF N. CARINATA gene followed the eastern boundary of the north-west Phylogeographic expectations of N. carinata are Transverse Ranges but turned west to include the guided by two main ecological factors. The first is southernmost sampling locality from the Santa Ynez their flightlessness; we expect populations to be rela- Mountains (Fig. 4B). The location of the genetic break tively isolated as a result of their inability to traverse based on the GFT network was further east between unsuitable habitat, and for lineage breaks to corre- the central Transverse Ranges and the Sierra Pelona. spond well with major geographic or ecological fea- Although there are no significant geographic barriers tures (although these may have varied considerably that would explain a genetic break in either of these over time). Second, we expect similar dispersal abili- regions, there are two other examples of an east–west ties of males and females and therefore do not expect break further east in the Transverse Ranges, in Sepe- population structure to be significantly influenced dophilus rove beetles (Chatzimanolis & Caterino, by sex-biased dispersal (Jockusch & Wake, 2002; 2007a) and California mountain king snakes (Lam- Matyukhin & Gongalsky, 2007). propeltis zonata) (Rodríguez-Robles et al. 1999), that

© 2010 The Linnean Society of London, Biological Journal of the Linnean Society, 2010, 99, 424–444 436 M. POLIHRONAKIS and M. S. CATERINO approximately correspond with N. carinata. Increased that this area represents an evolutionary hotspot. sampling of N. carinata in these regions, in addition Can these data shed light on any of the underlying to further comparative work including other taxa, processes? The maintenance of high diversity in would help determine whether these genetic breaks these populations supports the hypothesis that this are a result of stochastic processes (Irwin, 2002), or region once served as a glacial refugium that poten- the result of geography. tially sourced expansion into coastal and southern populations via the Transverse Ranges. However, this Isolation of southern Sierra Nevada population population was paraphyletic in the mtDNA tree with On the basis of the high diversity of the nuclear gene respect to haplotypes from the Tehachapi and Breck- in the southern Sierras, we propose that this popula- enridge Mountains, but monophyletic and quite tion was ancestral and potentially served as a glacial divergent in the GFT network. Non-monophyly of refugium (Hewitt, 2000; Rich et al., 2008). Further- mtDNA combined with monophyly of the nDNA is more, the genetic data from both the mitochondrial considered characteristic of differential introgression and nuclear markers supported a south and westward patterns between the two genomes (Chan & Levin, dispersal out of this area with an ancient stepping 2005; Weisrock et al., 2006; Linnen & Farrell, 2007; stone colonization through the Transverse Ranges. Peters et al., 2007; Spinks & Shaffer, 2007; Zhang & Potential geographic barriers reducing southward Sota, 2007; Gompert et al., 2008), and is suggestive of dispersal for flightless N. carinata beetles are the some interesting evolutionary dynamics in the Teh- Kaweah and Kern River drainages (Segraves & achapi Mountains. Pellmyr, 2001; Feldman & Spicer, 2006). These drain- Conflicting relationships inferred by the nuclear ages may have served as successive filters to dis- and mitochondrial markers in the Tehachapi Moun- persal, resulting in reduced population sizes during tains are consistent with differential introgression colonization as demonstrated by a single derived combined with several range fluctuations, which nuclear haplotype in the Tehachapis (G01), which resulted in unique contact zones for each of the two linked the larger western, central, and eastern groups markers. The haplotype relationships of the two throughout the rest of the range. Interestingly, all of markers from the southern Sierra Nevada suggest the main groups in the GFT network, with the excep- that migration out of this region comprised multiple tion of the southern Sierra population, were domi- events that occurred over an extended period of nated by one main haplotype with several derivative time. Under this scenario, an initial population haplotypes that suggest there were subsequent filters expansion southward was followed by subsequent to dispersal east and west out of the Tehachapis isolation (i.e. population contraction or relatively Mountains and Sierra Pelona region. simple isolation-by-distance) between populations in the south-western Sierra, and those in the Breck- enridge and Tehachapi Mountains. The divergence EVOLUTIONARY HOTSPOTS patterns of haplotypes in this region suggest that Our finding of both high genetic divergence and diver- this isolation lasted long enough for several sity of N. carinata populations from the Sierra and population-specific mutations to accumulate in both Tehachapi Mountains populations is interesting in the nuclear and mitochondrial genes. To explain the light of similar results from an increasing number of discrepancy of relationships of the Tehachapi Moun- other studies focusing on a wide variety of taxa dis- tains population between the two markers, we tributed throughout the CFP (Macey et al., 2001; Seg- hypothesize secondary contact between these iso- raves & Pellmyr, 2001; Jockusch & Wake, 2002; lated populations with differential introgression of Matocq, 2002; Feldman & Spicer, 2006; Kuchta & mitochondrial genes from the south-western Sierra Tan, 2006; Chatzimanolis & Caterino, 2007a, b; Davis populations south into the Breckenridge and Teh- et al., 2008). In addition to evidence of divergent achapi Mountains populations. This scenario would southern Sierran populations, two of these studies explain why individuals from the Tehachapi with sufficient taxon sampling also support higher Mountains had a derived nuclear haplotype and an genetic diversity in this region (Kuchta & Tan, 2006; ancestral mitochondrial haplotype. The data are Rich et al., 2008). This contrasts with previous work suggestive of asymmetrical mitochondrial introgres- revealing higher genetic diversity within southern or sion during secondary contact as a result of a lack coastal populations (Law & Crespi, 2002; Matocq, of derived nuclear or mitochondrial haplotypes in 2002), suggesting that refugia and/or genetic diversity the south-western Sierra population. We also did hotspots have existed in multiple areas, although not sample any individuals with an ancestral possibly over different time periods. nuclear haplotype and a derived mitochondrial hap- The elevated diversity in the southern Sierran lotype. Finer-scale sampling would help resolve his- region for both genes clearly supports the contention torical connections in this area.

High genetic diversity of N. carinata from the which revealed that many of the individuals from both southern Sierras and Transverse Ranges corre- of these populations shared identical haplotypes with sponded to evolutionary hotspots designated by Davis neighbouring populations. Mitochondrial studies of et al. (2008). However, only the southern Sierras other arthropod taxa in the CFP have found levels received support from both marker types, whereas much lower than this to correspond with well differ- diversity in the Transverse Ranges was supported by entiated subspecies (e.g. 0.014 in the yucca moth the COI data only. Evidence for other evolutionary Tegiticula maculata; Segraves & Pellmyr, 2001), and hotspots (sensu Davis et al., 2008) within the south- for comparable divergences to correspond to fully dis- ern CFP is equivocal; other recognized areas for tinct species (0.08 in a species complex of the Califor- which our data might be pertinent (i.e. the central nia turret spider, Antrodiaetus riversi; Starrett & coast ranges, the San Bernardino Mountains, and Hedin, 2007). In the latter case, nuclear data were San Jacinto Valley) showed little above baseline levels fully concordant with the mitochondrial, offering a of diversity in either locus. stark contrast to the situation in N. carinata.

EVIDENCE FOR CRYPTIC SPECIES? GENETIC VARIATION: MITOCHONDRIAL VERSUS Although some of the suggested morphological char- NUCLEAR DATA acters separating the three species N. carinata, N. It is often assumed that large-scale population-level vandykei, and N. aequicollis are relatively discern- processes will result in relatively parallel patterns able, high variation in N. carinata precludes their between marker types, although this is not always consistent application. Two of the main characters the case (Ballard, Chernoff & James, 2002; Bensch originally cited to distinguish N. vandykei from N. et al., 2006; Weisrock et al., 2006). Thus, the most carinata are the former’s comparatively slender, less difficult questions for multilocus phylogeography are robust form, and less pronounced frontal carina what differences are expected, what differences (ridge) on the head (Blaisdell, 1931). However, both exceed expectations, and what are the underlying these characters vary considerably within N. cari- causes in either case (Hudson & Turelli, 2003; Zink & nata, such that some specimens are easily diagnosed, Barrowclough, 2008). Most of the expected differences whereas others are not. The uncertainty surrounding between mitochondrial and nuclear markers in the the morphological distinctions was not necessarily present study likely arise from two factors: differen- clarified by the genetic data. In the COI gene tree, tial mutation rates and uneven effective population ‘N. vandykei’ specimens were paraphyletic with sizes (Forister et al., 2004). In general, the possibility respect to N. carinata specimens from the Tehachapi of recombination in nuclear loci and selection in both and Breckenridge Mountains, whereas, in the GFT marker types should also be considered, although we network, the ‘N. vandykei’ individuals are monophyl- found no evidence for either of these in our dataset. etic and deeply diverged from the rest of the individu- There are little data on expected mutation rates in als sampled (discussed above). The uncertainty nuclear introns relative to mitochondrial protein- surrounding N. aequicollis is even greater as a result coding sequences (Zhang & Hewitt, 2003) but the of its very close ties with populations in the central background rate should be significantly lower, portion of the Transverse Ranges according to the perhaps one-tenth the mtDNA rate (Hare & Palumbi, GFT gene, at the same time as being quite distinct 2003). The interaction of mutation generation and from most other lineages in the COI tree (but grouped maintenance via larger Ne is generally expected to with the south-west Sierra and Tehachapi clade with yield lower variation in nuclear haplotypes in a popu- midpoint rooting). lation sample (Zink & Barrowclough, 2008), although Aside from the questionable status of the named just how much less will be determined by many species, high divergence among mtDNA haplotypes additional factors (Hudson & Turelli, 2003). was also initially suggestive of additional cryptic The clearest finding supported by both genes was species within N. carinata. Where populations were the generally low population connectedness. Both FST monophyletic, average uncorrected distances to neigh- and autocorrelation analyses indicated rapid attenu- bouring populations exceeded 0.05 in all cases ation of similarity with distance. This indicates that (Caterino & Chatzimanolis, 2009). The high diver- the shared haplotypes or related haplotypes we gence of the two most northerly and southerly popu- observed across populations did not result from recent lations (northern Santa Lucias and San Jacinto gene flow. This is further supported by COI relation- Mountains, respectively), in particular, suggested that ships, where such geographic discordance was most these lineages represented genealogically distinct obvious (San Bernardino haplotypes in the San Jacin- units as a result of allopatric divergence. However, this tos; Central Transverse Range haplotypes in the Teh- hypothesis was not supported by the nuclear data, achapis). In no cases were ‘rogue’ haplotypes identical

© 2010 The Linnean Society of London, Biological Journal of the Linnean Society, 2010, 99, 424–444 438 M. POLIHRONAKIS and M. S. CATERINO to any in their source areas, indicating that we new generation of protein database search programs. sampled descendants of the original contact, and that Nucleic Acids Research 25: 3389–3402. the observed relationships resulted from either Avise JC. 2000. Phylogeography. The history and formation of ancient migration or the retention of ancestral poly- species. Cambridge, MA: Harvard University Press. morphism from a more broadly-distributed historical Ballard WJ, Chernoff B, James AC. 2002. Divergence of population. In the GFT network, these particular sets mitochondrial DNA is not corroborated by nuclear DNA, of areas shared common haplotypes but did not share morphology, or behavior in Drosophila simulans. Evolution any of the more recently derived variants within their 56: 527–545. Bensch S, Irwin DE, Irwin JH, Kvist L, Åkesson S. 2006. respective groups. This supports the latter scenario Conflicting patterns of mitochondrial and nuclear DNA suggesting these populations were broadly connected diversity in Phylloscopus warblers. Molecular Ecology 15: in the past. 161–171. Further work resolving details of the evolutionary Berghoff SM, Kronauer JC, Edwards KJ, Franks NR. history in this group would continue to solidify 2008. Dispersal and population structure of a New World support for the importance of this region with regard predator, the army ant Eciton burchellii. Journal of Evolu- to the unique evolutionary dynamics and species tionary Biology 21: 1125–1132. diversity patterns found in the present study. Specifi- Blaisdell FE. 1931. Studies in the Tenebrionidae, number 3. cally, we need a more robust outgroup for this species, Pan-Pacific Entomologist 8: 41–46. which may be difficult as a result of the fact that it is Bruen TC, Phillipe H, Bryant D. 2006. A simple and robust the only genus in its tribe worldwide. However, statistical test for detecting the presence of recombination. because of the relatively low variation of the nuclear Genetics 172: 2665–2681. gene in the present study, it may be possible to use Calsbeek R, Thompson JN, Richardson JE. 2003. Pat- species in related tribes of Pimeliinae to root the N. terns of molecular evolution and diversification in a biodi- carinata nuclear gene tree. Increased taxon sampling versity hotspot: the California Floristic Province. Molecular in areas of high diversity, such as the southern Sierra Ecology 12: 1021–1029. Nevada, and throughout the more peripheral parts of Cassens I, Mardulyn P, Milinkovitch MC. 2005. Evaluat- the range would also provide further insight into ing intraspecific ‘network’ construction methods using simu- important evolutionary processes in the species and lated sequence data: do existing algorithms outperform the the region. global maximum parsimony approach? Systematic Biology 54: 363–372. Caterino MS, Chatzimanolis S. 2009. Conservation genet- ACKNOWLEDGEMENTS ics of three flightless beetle species in southern California. Conservation Genetics 10: 203–216. Our field collections were assisted and facilitated by Caterino MS, Cho S, Sperling FAH. 2000. The current numerous individuals and agencies. We thank R. state of insect molecular systematics: a thriving Tower of Aalbu, S. Chatzimanolis, K. Hopp, P. Jump, S. Babel. Annual Review of Entomology 45: 1–54. Mulqueen, S. Russell, P. Russell, A. Short, K. Will, the Chan KMA, Levin SA. 2005. Leaky prezygotic isolation and California Department of Fish and Game, Los Padres porous genomes: rapid introgression of maternally inherited National Forest, Angeles National Forest, Sequoia DNA. Evolution 59: 720–729. National Forest, San Bernardino National Forest, Chatzimanolis S, Caterino MS. 2007a. Toward a better San Dimas Experimental Forest, Arroyo Hondo Pre- understanding of the ‘Transverse Range break’: lineage serve (Land Trust for Santa Barbara County), Camp diversification in southern California. Evolution 61: 2127– Cedar Falls, the UC McLaughlin Reserve, UC Sedg- 2141. wick Reserve, UC Whitaker Forest, the UC James Chatzimanolis S, Caterino MS. 2007b. Limited phylogeo- Reserve, and the UC Santa Cruz Island Reserve. We graphic structure in the flightless ground beetle, Calathus appreciate the assistance of I. Ouzounov and G. ruficollis, in southern California. Diversity and Distribu- Betzholtz in the laboratory, J. Q. Richmond for help tions 13: 498–509. with the data interpretation, and V. Krauss for Conservation International. 2009. Biodiversity Hotspots. Downloaded on 13 July 2009. Available at: http://www. sharing his Tribolium intron data. This work was biodiversityhotspots.org/xp/hotspots/california_floristic/ supported by the Schlinger Foundation, a bequest Pages/default.aspx from G. Ostertag, and National Science Foundation Davis EB, Koo MS, Conroy C, Patton JL, Moritz C. 2008. award DEB0447694 to M. Caterino. The California Hotspots Project: identifying regions of rapid diversification of mammals. Molecular Ecology 17: 120–138. Excoffier L, Smouse P, Quattro J. 1992. Analysis of REFERENCES molecular variance inferred from metric distances among Altschul SF, Madden TL, Schäffer AA, Zhang J, Miller DNA haplotypes: application to human mitochondrial DNA W, Lipman DJ. 1997. Gapped BLAST and PSI-BLAST: a restriction data. Genetics 131: 479–491.

Excoffier L, Laval G, Schneider S. 2005. ARLEQUIN ver. phy of codistributed species in California. American Natu- 3.0: an intergrated software package for population genetics ralist 166: 290–299. data analysis. Evolutionary Bioinformatics Online 1: 47– Law JH, Crespi BJ. 2002. The evolution of geographic 50. parthenogenesis in Timema walking-sticks. Molecular Feldman CR, Spicer GS. 2006. Comparative phylogeogra- Ecology 11: 1471–1489. phy of woodland reptiles in California: repeated patterns of Linnen CR, Farrell BD. 2007. Mitonuclear discordance is cladogenesis and population expansion. Molecular Ecology caused by rampant mitochondrial introgression in Neodip- 15: 2201–2222. rion (Hymenoptera: Diprionidae) sawflies. Evolution 61: Forister ML, Fordyce JA, Shapiro AM. 2004. Geological 1417–1438. barriers and restricted gene flow in the holarctic skipper Macey JR, Strasburg JL, Brisson JA, Vredenburg VT, Hesperia comma (Hesperiidae). Molecular Ecology 13: 3489– Jennings M, Larson A. 2001. Molecular phylogenetics of 3499. western North American frogs of the Rana boylii species Godinho R, Crespo EG, Ferrand N. 2008. The limits of group. Molecular Phylogenetics and Evolution 19: 131–143. mtDNA phylogeography: complex patterns of population Maddison WP. 1997. Gene trees in species trees. Systematic history in a highly structured Iberian lizard are only Biology 46: 523–536. revealed by the use of nuclear markers. Molecular Ecology Maldonado JE, Vila C, Wayne RK. 2001. Tripartite genetic 17: 4670–4683. subdivisions in the ornate shrew (Sorex ornatus). Molecular Gompert Z, Forister ML, Fordyce JA, Nice CC. 2008. Ecology 10: 127–147. Widespread mito-nuclear discordance with evidence Matocq MD. 2002. Phylogeographical structure and regional for introgressive hybridization and selective sweeps in history of the dusky-footed woodrat, Neotoma fuscipes. Lycaeides. Molecular Ecology 17: 5231–5244. Molecular Ecology 11: 229–242. Hare MP, Palumbi SR. 2003. High intron sequence conser- Matyukhin AV, Gongalsky KB. 2007. The home range size vation across three mammalian orders suggests functional of two darkling beetle species (Coleoptera, Tenebrionidae) constraints. Molecular Biology and Evolution 20: 969–978. from southern Kazakhstan. Zoologichesky Zhurnal 86: Hewitt G. 2000. The genetic legacy of the Quaternary ice 1446–1451. ages. Nature 405: 907–913. Miller MP. 2005. Alleles in Space (AIS): computer software Hudson RR, Turelli M. 2003. Stochasticity overrules the for the joint analysis of interindividual spatial and genetic ‘three-times rule’: genetic drift, genetic draft, and coales- information. Journal of Heredity 96: 722–724. cence times for nuclear loci versus mitochondrial DNA. Miller MP, Haig SM, Wagner RS. 2005. Conflicting pat- Evolution 57: 182–190. terns of genetic structure produced by nuclear and mito- Huson DH. 1998. SplitsTree: a program for analyzing and chondrial markers in the Oregon slender salamander visualizing evolutionary data. Bioinformatics 14: 68–73. (Batrachoseps wrighti): implications for conservation efforts Huson DH, Bryant D. 2006. Application of phylogenetic and species management. Conservation Genetics 6: 275–287. networks in evolutionary studies. Molecular Biology Miller MP, Bellinger R, Forsman ED, Haig SM. 2006. and Evolution 23: 254–267. Software available at http:// Effects of historical climate change, habitat connectivity, www.splitstree.org and vicariance on genetic structure and diversity across Irwin DE. 2002. Phylogeographic breaks without geographic therange of the red tree vole (Phenacomys longicaudus)in barriers to gene flow. Evolution 56: 2383–2394. the Pacific northwestern United States. Molecular Ecology Jockusch EL, Wake DB. 2002. Falling apart and merging: 15: 145–159. diversification of slender salamanders (Plethodontidae: Ngamprasertwong T, Mackie IJ, Racey PA, Piertney SB. Batrachoseps) in the American West. Biological Journal of 2008. Spatial distribution of mitochondrial and microsatel- the Linnean Society 76: 361–391. lite DNA variation in Daubenton’s bat within Scotland. Kerth G, Petrov B, Conti A, Anastasov D, Weishaar M, Molecular Ecology 17: 3243–3258. Gazaryan S, Jaquiéry J, König B, Perrin N, Bruyn- Nylander JAA. 2004. MrModeltest, Version 2. Program dis- donckx N. 2008. Communally breeding Bechstein’s bats tributed by author. Evolutionary Biology Centre, Uppsala have a stable social system that is independent from the University. postglacial history and location of the populations. Molecu- Peakall R, Smouse PE. 2006. GENALEX 6: genetic analysis lar Ecology 17: 2368–2381. in Excel. Population genetic software for teaching and Krauss V, Thümmler C, Georgi F, Lehmann J, Stadler research. Molecular Ecology Notes 6: 288–295. PF, Eisenhardt C. 2008. Near intron positions are reliable Peakall R, Ruibal M, Lindenmayer DB. 2003. Spatial phylogenetic markers: an application to holometabolous autocorrelation analysis offers new insights into gene flow insects. Molecular Biology Evolution 25: 821–830. in the Australian bush rat, Rattus fuscipes. Evolution 57: Kuchta SR, Tan AM. 2006. Lineage diversification on an 1182–1195. evolving landscape: phylogeography of the California newt, Peters JL, Zhuravlev Y, Fefelov I, Logie A, Omland KE. Taricha torosa (Caudata: Salamandridae). Biological 2007. Nuclear loci and coalescent methods support ancient Journal of the Linnean Society 89: 213–239. hybridization as cause of mitochondrial paraphyly between Lapointe FJ, Rissler LJ. 2005. Notes and comments – gadwall and falcated duck (Anas spp.). Evolution 61: 1992– congruence, consensus, and the comparative phylogeogra- 2006.

Rich KA, Thompson JN, Fernandez CC. 2008. Diverse Starrett J, Hedin M. 2007. Multilocus genealogies reveal historical processesshape deep phylogeographical diver- cryptic species and biogeographical complexity in the Cali- gence in the pollinating seed parasite Greya politella. fornia turret spider Antrodiaetus riversi (Mygalomorphae, Molecular Ecology 17: 2430–2448. Antrodiaetidae). Molecular Ecology 16: 583–604. Rissler LJ, Hijmans RJ, Graham CH, Moritz C, Wake Stephens M, Scheet P. 2005. Accounting for decay of linkage DB. 2006. Phylogeographic lineages and species compari- disequilibrium in haplotype inference and missing data sons in conservation analyses: a case study of California imputation. American Journal of Human Genetics 76: 449– herpetofauna. American Naturalist 167: 655–666. 462. Rodríguez-Robles JA, Denardo DF, Staub RE. 1999. Stephens M, Smith NJ, Donnely P. 2001. A new statistical Phylogeography of the California mountain king snake, method for haplotype reconstruction from population data. Lampropeltis zonata (Colubridae). Molecular Ecology 8: American Journal of Human Genetics 68: 978–989. 1923–1934. Swofford DL. 2002. PAUP*. Sunderland, MA: Sinauer Segraves KA, Pellmyr O. 2001. Phylogeography of the yucca Associates. moth Tegeticula maculata: the role of historical biogeogra- Tajima F. 1989. Statistical method for testing the neutral phy in reconciling high genetic structure with limited spe- mutation hypothesis by DNA polymorphism. Genetics 123: ciation. Molecular Ecology 10: 1247–1253. 585–595. Sgariglia EA, Burns KJ. 2003. Phylogeography of the Cali- Weisrock DW, Shaffer HB, Storz BL, Storz SR, Voss SR. fornia thrasher (Toxostoma redivivum) based on nested- 2006. Multiple nuclear gene sequences identify phylogenetic clade analysis of mitochondrial-DNA variation. Auk 120: species boundaries in the rapidly radiating clade of Mexican 346–361. ambystomatid salamanders. Molecular Ecology 15: 2489– Simon C, Frati F, Beckenback A, Crespi BJ, Liu H, 2503. Flook P. 1994. Evolution, weighting, and phylogenetic Zhang DX, Hewitt GM. 2003. Nuclear DNA analyses in utility of mitochondrial gene sequences and a compilation of genetic studies of populations: practice, problems and pros- conserved polymerase chain reaction primers. Annals of the pects. Molecular Ecology 12: 563–584. Entomological Society of America 87: 651–701. Zhang AB, Sota T. 2007. Nuclear gene sequences Smouse PE, Peakall R. 1999. Spatial autocorrelation analy- resolve species phylogeny and mitochondrial introgression sis of individual multiallele and multilocus genetic struc- in Leptocarabus beetles showing trans-species polymor- ture. Heredity 82: 561–573. phisms. Molecular Phylogenetics and Evolution 45: 534– Spinks PQ, Shaffer HB. 2005. Range-wide molecular analy- 546. sis of the western pond turtle (Emys marmorata): cryptic Zink RM, Barrowclough GF. 2008. Mitochondrial DNA variation, isolation by distance, and their conservation under siege in avian phylogeography. Molecular Ecology 17: implications. Molecular Ecology 14: 2047–2064. 2107–2121. Spinks PQ, Shaffer HB. 2007. Conservation phylogenetics of Zwickl DJ. 2006. Genetic algorithm approaches for the phy- the Asian box turtles (Geomydidae, Cuora): mitochondrial logenetic analysis of large biological sequence data sets introgression, numts, and inferences from multiple nuclear under maximum likelihood criterion, PhD Dissertation, loci. Conservation Genetics 8: 641–657. University of Texas at Austin.

APPENDIX SAMPLING LOCALITIES BY REGION, WITH HAPLOTYPES REPRESENTED IN EACH (LOCALITY NUMBERS REFER TO POINTS ON FIG.1)

Population N Locality Coordinates COI haplotypes GFT haplotypes

1. Northern 10 1.1 – CA: Monterey 36.2341°N, N88, N89, N90, N91 G04, G55, G56, G57 Santa Lucias Co., Arroyo Seco 121.4886°W County Park Total 10 2. Southern 5 2.1 – CA: San Luis 35.3731°N, N13, N14, N15 G04, G09, G10 Santa Lucias Obispo Co., LPNF, 120.6859°W Cuesta Ridge 1 2.2 – CA: San Luis 35.5861°N, N93 G60 Obispo Co., Santa 121.0057°W Rosa Creek Total 6 3. Southwestern 2 3.1 – CA: Fresno Co., 36.7718°N, N78, N79 G39, G40, G41, G42 Sierra Nevada Sequoia NF, SE of 118.8854°W Hume Lake 8 3.2 – CA: Tulare Co., 36.7025°N, N72, N73, N74, N75, G32, G33, G34, G35, Sequoia NF, UC 118.9322°W N76, N77 G36, G37, G38, G49 Whitaker Forest 1 3.2 – CA: Tulare Co., 36.6720°N, N80 G43 Sequoia NF, F.S. 118.9798°W 14S75 Total 11 4. Breckenridge 2 4.1 – CA: Kern Co., 35.4753°N, N02 – Mountains Sequoia NF, Kern 118.7284°W Cyn. Total 2 5. Tehachapi 2 5.1 – CA: Kern Co., 35.0487°N, N61, N62 G01 Mountains Oak Creek Cyn. 118.3605°W 5 5.1 – CA: Kern Co., 35.0506°N, N61, N63 G01 Oak Creek Cyn. 118.3567°W 3 5.1 – CA: Kern Co., 35.0673°N, N81, N82 G01, G44, G45, G46, Tehachapi Mountain 118.4820°W G47 Park 1 5.1 – CA: Kern Co., 35.1807°N, N81 G45, G48 Woodford-Tehachapi 118.5150°W Rd Total 11 6. Santa Ynez 3 6.1 – CA: Santa 34.4855°N, N64, N65, N66 – Mountains Barbara Co., Arroyo 120.1423°W Hondo Preserve 2 6.1 – CA: Santa 34.4855°N, N7, N8 G07 Barbara Co., Arroyo 120.1417°W Hondo Preserve 1 6.2 – CA: Santa 34.47°N, N44 G14 Barbara Co., Santa 119.73°W Barbara 5 6.3 – CA: Ventura Co., 34.5009°N, N50, N51, N52, N53, G26, G05, G04, G27, LPNF, Murrietta Tr. 119.3899°W N54 G17, G28 1 6.4 – CA: Ventura Co., 34.4504°N, N42 G05, G22 Upper Ojai Valley 119.1207°W 3 6.5 – CA: Ventura Co., 34.293°N, N28, N29, N30 G05, G17, G01, G18 Ventura 119.204°W

APPENDIX Continued

Population N Locality Coordinates COI haplotypes GFT haplotypes

Total 15 7. Northwest 1 7.1 – CA: Santa 34.6796°N, N9 G04 Transverse Barbara Co., UC 120.0387°W Ranges Sedgwick Reserve 1 7.1 – CA: Santa 34.6825°N, N9 Barbara Co., UC 120.0445°W Sedgwick Reserve 2 7.1 – CA: Santa 34.6842°N, N11, N12 G04 Barbara Co., UC 120.0459°W Sedgwick Reserve 1 7.1 – CA: Santa 34.6889°N, N49 G24, G25 Barbara Co., UC 120.0428°W Sedgwick Reserve 1 7.1 – CA: Santa 34.7127°N, N35 G04, G20 Barbara Co., UC 120.0396°W Sedgwick Reserve 1 7.1 – CA: Santa 34.7132°N, N41 G04 Barbara Co., UC 120.0395°W Sedgwick Reserve 1 7.1 – CA: Santa 34.7164°N, N23 G04, G13 Barbara Co., UC 120.0396°W Sedgwick Reserve 2 7.1 – CA: Santa 34.7197°N, N31, N39 G04 Barbara Co., UC 120.0366°W Sedgwick Reserve 1 7.1 – CA: Santa 34.7261°N, N43 G20, G23 Barbara Co., UC 120.0483°W Sedgwick Reserve 1 7.1 – CA: Santa 34.7269°N, N9 G04, G08 Barbara Co., UC 120.0474°W Sedgwick Reserve 1 7.1 – CA: Santa 34.7197°N, – G04 Barbara Co., UC 120.0415°W Sedgwick Reserve 1 7.2 – CA: Santa 34.7538°N, N3 G04 Barbara Co., LPNF, 119.9429°W Sunset Valley 1 7.3 – CA: Santa 34.5410°N, N24 G14, G15 Barbara Co., LPNF, 119.9110°W Rancho Alegre 2 7.4 – CA: Santa 34.7021°N, N4 G04 Barbara Co., LPNF, 119.6547°W Big Pine Mountain 2 7.4 – CA: Santa 34.7032°N, N36, N38 G04 Barbara Co., LPNF, 119.6526°W Big Pine Mountain 1 7.5 – CA: Ventura Co., 34.6585°N, N1 – LPNF, north slope 119.3800°W Pine Mountain Total 20 8. Central 2 8.1 – CA: Kern Co., 34.8484°N, N94, N96 G01 Transverse LPNF, Tecuya Rd. 119.0697°W Ranges 4 8.1 – CA: Kern Co., 34.8300°N, N94, N95 G01 LPNF, Tecuya Trail 119.0078°W

APPENDIX Continued

Population N Locality Coordinates COI haplotypes GFT haplotypes

Total 6 9. Sierra 7 9.1 – CA: Los Angeles 34.6408°N, N10, N18, N26, N37, G01, G02, G03 Pelona Co., Angeles NF, 118.4148°W N46, N47, N48 Grass Mountain Total 7 10. San Gabriel 7 10.1 – CA: Los 34.3764°N, N55, N56, N57, N58, G01 Mountains Angeles Co., 118.4403°W N59, N60 Placerita Cyn. Co. Pk. 2 10.2 – CA: Los 34.2717°N, N25, N27 G16 Angeles Co., 118.0608°W Angeles NF, Angeles Crest Hwy. 3 10.2 – CA: Los 34.2964°N, N25, N67, N68 G16, G29, G30 Angeles Co., 118.1625°W Angeles NF, Hwy N3 2 10.2 – CA: Los 34.2910°N, N71, N85 G01, G50 Angeles Co., 118.1695°W Angeles NF, Hwy N3 2 10.3 – CA: Los 34.2012°N, N86, N87 G51, G52, G53, G54 Angeles Co., 117.7736°W Angeles NF, San Dimas Exp. Forest Total 16 11. San 1 11.1 – CA: San 34.2955°N, N84 G05, G31 Bernardino Bernardino Co., 117.2134°W Mountains SBNF, Hwy 173 1 11.2 – CA: San 34.1864°N, N45 G05, G21 Bernardino Co., 117.1836°W SBNF, City Ck. 1 11.3 – CA: San 34.1741°N, N40 G06, G21 Bernardino Co., 116.9844°W SBNF, Deer Creek 2 11.3 – CA: San 34.1678°N, N05, N06 G05, G06 Bernardino Co., 116.9143°W SBNF, W. of Barton Flats 4 11.3 – CA: San 34.1819°N, N06, N70 G05, G06 Bernardino Co., 116.8884°W SBNF, Santa Ana R. 2 11.3 – CA: San 34.1646°N, N69 G05 Bernardino Co., 116.9366°W SBNF, Camp Cedar Falls Total 11

APPENDIX Continued

Population N Locality Coordinates COI haplotypes GFT haplotypes

12. San Jacinto 8 12.1 – CA: Riverside 33.8092°N, N16, N17, N19, N20, G05, G11 Mountains Co., UC James 116.7650°W N21, N22 Reserve 1 12.1 – CA: Riverside 33.8081°N, N83 G11 Co., UC James 116.7784°W Reserve 1 12.1 – CA: Riverside 33.8448°N, N19 G05 Co., SBNF, F.S. 116.7322°W 4S01 1 12.1 – CA: Riverside 33.7949°N, N32 G05 Co., SBNF, Marion 116.7214°W Mountain Total 11 13. Santa Cruz 2 13.1 – CA: Santa 34.0191°N, N33, N34 G04, G04, G19 Islands Barbara Co., Santa 119.6878°W Cruz Islands 4 13.1 – CA: Santa 34.0222°N, N97, N98, N99, G04, G19, G61, G62 Barbara Co., 119.6906°W N100 Pelican Bay Tr. Total 6

SUPPORTING INFORMATION Additional Supporting Information may be found in the online version of this article: Appendix S1. Haplotypes and GenBank accession numbers. ‘N’ haplotypes refer to the Nyctoporis carinata COI sequence; ‘G’ haplotypes refer to the Nyctoporis carinata GFT sequence. Please note: Blackwell Publishing are not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.