Insect. Soc. DOI 10.1007/s00040-009-0041-1 Insectes Sociaux

RESEARCH ARTICLE

Preliminary analysis of a hybrid zone between two subspecies of

B. T. Aldrich Æ S. Kambhampati

Received: 6 May 2009 / Revised: 27 August 2009 / Accepted: 1 September 2009 Ó Birkha¨user Verlag, Basel/Switzerland 2009

Abstract For species largely allopatric in distribution, Keywords Zootermopsis Á Subspecies Á Hydrocarbons Á zones of contact provide an opportunity for hybridization, Hybridization Á Microsatellites Á Mitochondrial COI testing grounds for species boundaries, and may result in the formation of a new species. Thus, hybrid zones have Introduction the potential to provide important insights into speciation. In this study, we performed a preliminary analysis of a Dampwood in the genus Zootermopsis are the only hybrid zone between two subspecies of the dampwood endemic termites in the Nearctic temperate forests of wes- , Zootermopsis nevadensis (Z. n. nuttingi and Z. n. tern North America (Eggleton, 2000). Three species are nevadensis) near Bartle, CA, using 12 microsatellite loci known to occur in this area: Z. nevadensis (Hagen), Z. an- and a mitochondrial gene. Fifty-seven colonies collected in gusticollis (Hagen), and Z. laticeps (Banks). Of particular 36 locations were analyzed. The analysis of genetic interest to this study are two Z. nevadensis subspecies, Z. markers revealed a large hybrid zone approximately nevadensis nevadensis and Z. nevadensis nuttingi (Haverty 104 km in width. Although stepped clines best explained and Thorne, 1989) that diverged during the Pleistocene the data, we are unable to rule out the existence of a mosaic (approximately 2 million years ago; Broughton and Kistner, hybrid zone. We inferred weak selection (s* \ 0.05%) 1991). Although climatic oscillations and microhabitat against hybrids, but the data also suggested the existence of differentiation during this period led to extensive speciation a barrier to gene flow from Z. n. nevadensis to Z. n. nutt- (Hewitt, 1996, 2000; Avise et al., 1998), Pleistocene ingi, but not in the other direction. Given the large zone of vicariance has not resulted in complete speciation or genetic contact, extensive sampling is needed to obtain a more isolation between evolutionary lineages of many organisms complete characterization of the hybrid zone. However, the (Hewitt, 2000; Wake, 1997; Sequeira et al., 2005). results of this study suggest that despite the accumulation The two Z. nevadensis subspecies display two distinct of phenotypic and genetic differences and time since hydrocarbon phenotypes (phenotypes I: Z. n. nevadensis and divergence (*2 million years), Z. n. nuttingi and Z. n. III: Z. n. nuttingi; Haverty et al., 1988), interspecific nevadensis are capable of extensive hybridization. aggression between non-reproductive castes comparable to that of distinct Zootermopsis species (Haverty and Thorne, 1989), and subspecies-specific sequence differences in & B. T. Aldrich Á S. Kambhampati ( ) the mitochondrial cytochrome oxidase I gene (Aldrich Department of Entomology, Kansas State University, Manhattan, KS 66506, USA and Kambhampati, 2007). Laboratory matings between the e-mail: [email protected] two Z. nevadensis hydrocarbon phenotypes result in viable B. T. Aldrich immatures with an intermediate hydrocarbon phenotype; e-mail: [email protected] however, collection of these hybrid individuals in the field has yet to be reported (Thorne et al., 1993). Therefore, a Present Address: survey of natural populations of the Z. nevadensis subspecies B. T. Aldrich Department of Anesthesia, University of Iowa, using genetic markers is needed to examine potential Iowa City, IA 52242, USA hybridization between the subspecies in the field. B. T. Aldrich, S. Kambhampati

The degree of genetic isolation, or lack thereof, between of Bartle are hydrocarbon phenotype I (Z. n. nevadensis), evolutionary lineages can be examined in zones of primary whereas colonies to the northwest are hydrocarbon pheno- or secondary contact (hybrid zones), where two distinct type III (Z. n. nuttingi) (M. Haverty, pers. comm.; Aldrich gene pools meet, introgress, and produce hybrids (Barton and Kambhampati, 2007). Surveys by M. Haverty (pers. and Hewitt, 1985, 1989; Harrison, 1990; Arnold, 1997). comm.) indicated that a zone of contact between the two Hybrid zones may result in clines or gradual changes in subspecies occurs approximately 4–10-km east of Bartle. allele frequency across the zone of contact, which are Therefore, the area west and east of Bartle was extensively maintained by a balance between dispersal and selection sampled in an effort to identify the center of the zone of against hybrids (Barton and Gale, 1993; Barton and Hewitt, contact between the two subspecies (Fig. 1). Sampling was 1985). Depending on the level of selection against hybrids, performed extending some distance from each side of the a zone of contact may form a steep, narrow cline (strong presumed contact zone to ensure that sampling encom- selection) or a shallow, wide cline (weak selection) relative passed as much of the contact area as possible. Termite to dispersal. Thus, estimating the width of the cline will colonies were sampled from tree stumps and fallen logs. We indicate if there is a strong selection against hybrids pro- attempted to collect at least 40 individuals irrespective of moting reproductive isolation or weak selection against caste from each colony, which were preserved in vials of hybrids allowing admixture and gene flow between two 95% ethanol for genetic analysis. Live specimens were evolutionary lineages. However, not all hybrid zones are collected and sent to Michael Haverty (USDA Forest clinal; a number of studies (e.g., Britch et al., 2001; Rand Service, Berkeley, CA) for phenotyping using hydrocarbon and Harrison, 1989, Ross and Harrison, 2002; Sites et al. analysis as described by Haverty et al. (1988). 1995; Vines et al., 2003) have shown that mosaic hybrid Special considerations must be taken when performing a zones may arise when ecologically differentiated taxa hybrid zone analysis of termites. Typical hybrid zone hybridize across a network of habitat patches. Mosaic analyses require sampling of multiple individuals to esti- patterns can also arise through random genetic drift that mate the frequency of hybrid genotypes in the population. occurs due to low migration rate or long-range dispersal However, eusocial termites live in colonies, are largely into the space between two expanding populations (Vines monogamous, and display a reproductive division of labor. et al., 2003). Thus, mosaic hybrid zones may be generated Thus, although termite populations consist of thousands of by either low dispersal with drift or habitat associations. individuals, from a genetic standpoint, the number of For subspecies, hybrid zone characteristics can provide independent genotypes is quite small (i.e., 2–4 per colony; insight into the geographic and habitat structure influencing Aldrich and Kambhampati, 2007). Therefore, in termites hybridization (Bun˜o et al., 1994), the influence of hybrid- the ‘‘individual’’ is the colony and an appropriate strategy ization on phenotypic characters of subspecies (Largiade`r is to sample multiple colonies from a single site. Unfor- et al., 1994), niche differentiation between subspecies tunately, we encountered very low colony densities within (Wang et al., 1999), and the evolution of reproductive the Z. nevadensis hybrid zone. In fact, our density estimate isolation (Fel-Clair et al., 1998; Smadja et al., 2004). (2 colonies ha-1) is at the extreme low end of published The objectives of this study were as follows: (1) define termite density estimates ranging from 2 to 309 colo- the extent of the contact zone between the two Z. nevad- nies ha-1 (Lepage and Darlington, 2000). Therefore, in this ensis subspecies in the field and determine if intermediate hybrid zone, sampling multiple colonies is analogous to hydrocarbon phenotypes, resulting from hybridization sampling organisms with very low population numbers between the two subspecies, are present, (2) utilize 12 even though over 2,000 termites were analyzed in this microsatellite loci and one mitochondrial locus to deter- study. Despite these limitations, the area of presumed mine if genetic differences are breaking down in the zone contact was extensively sampled and when possible mul- of contact, and (3) provide a preliminary characterization tiple colonies were collected from a single site. Thus, of the zone of contact between the two subspecies. although the sampling in this study may appear coarse, given the low density of colonies our sampling is a good representation of the total number of colonies in this area Materials and methods and should provide a good approximation of the frequency of hybrid colonies (individuals) in this population. Sampling Mitochondrial DNA Sampling efforts were focused on a potential hybrid zone between Z. n. nevadensis and Z. n. nuttingi near the town of In a previous study, we (Aldrich and Kambhampati, 2007) Bartle, CA during June, 2003 (Bartle is located between identified two fixed base substitutions between the two sites 15 and 16; Fig. 1; Table 1). Colonies to the southeast Z. nevadensis subspecies in the mitochondrial COI gene. Hybrid zone between Zootermopsis subspecies

Fig. 1 Map showing the distribution of Z. nevadensis colonies sampled in this study. Bartle, CA located between sites 15 and 16

In this study, we sequenced the COI gene from one indi- zone, we used the Bayesian clustering method imple- vidual per colony in the presumed zone of contact to map mented in STRUCTURE 2.1 (Pritchard et al., 2000). the distribution of the two mitochondrial haplotypes. Owing to the high levels of relatedness among colony- Sequences are deposited in Genbank under accession mates, the data contained substantial colony-level numbers DQ133197–DQ133255. structuring. We attempted to remove the colony-level structuring effect by randomly selecting a single individual Microsatellite analysis from each colony for use in these tests. This enabled us to quantify sub-structuring above the level of the colony (i.e., DNA was extracted from the head and thorax of 26–54 grouping of colonies of the same subspecies). The resam- individuals from each colony as described (Aldrich and pling procedure was replicated 20 times for all colonies. Kambhampati, 2007). Twelve polymorphic microsatellite In the cluster analysis, individuals were treated as a loci were amplified for each individual using primers and single population without any prior assignment to subspe- methods reported by Aldrich and Kambhampati (2004). cies. We tested each of the 20 replicates for values of K Aldrich and Kambhampati (2007) provided the details of (optimal number of clusters or populations) from 1 to 20 allele visualization and scoring. with 20,000 iterations and a burn in of 10,000 iterations. Each test of K was replicated 10 times for a total of 4,000 Microsatellite differentiation across the hybrid zone tests [20 resampled datasets 9 20 (K) 9 10 replicates of (K)]. The 200 values of log likelihood [Ln P(D)] for each K To determine whether microsatellite allele frequency data were then averaged across datasets and replicates. The could be used to characterize the change from Z. n. nuttingi optimal K was chosen by comparing the average Ln P(D) colonies to Z. n. nevadensis colonies across the hybrid estimates for each K. Following the selection of K,we B. T. Aldrich, S. Kambhampati

Table 1 Sampling data for Zootermopsis nevadensis Site# Colony# Location Latitude Longitude Altitude Hydrocarbon Cluster 1 COI probability haplotype

1 1 Klamath National Forest 41.51.29 122.42.14 1,878 III 0.89 I 2 2 Hwy 96 Klamath National Forest 41.51.26 122.42.08 1,976 III 0.93 I 3 3 Klamath National Forest 41.50.50 122.41.01 2,070 0.94 I 4 4 Edgewood ? Calhan 41.25.85 122.38.33 4,770 III 0.86 I 5 5 Edgewood ? Calhan 41.24.73 122.37.94 4,509 III 0.66 I 6 6 Hwy 3 South of Scott Mts. Trail 41.15.56 122.40.20 4,875 III 0.93 I 7 7 Hwy 3 South of Scott Mts. Trail 41.14.79 122.39.89 4,874 III 0.93 I 8 8 Dunsmuir 41.09.75 122.17.28 1,944 0.90 I 9 9 Dunsmuir 41.10.32 122.15.78 1,944 0.94 I 10 10 McCloud 41.16.19 122.11.04 3,900 0.76 I 11 0.82 I 11 12 McCloud 41.15.13 122.06.37 3,370 0.58 I 12 13 Pilgrim creek road 0.2 mi north 89 41.19.12 122.00.97 3,803 0.91 I 13 14 Pilgrim creek road 0.5 mi north 89 41.19.32 122.00.78 3,850 III 0.62 I 15 III 0.79 I 16 0.81 I 17 0.91 I 18 0.60 I 14 19 4 mi West Bartle 41.16.09 121.54.75 3,314 III 0.89 II 15 20 0.5 mi North Bartle Hwy 89 41.15.64 121.50.58 3,949 III 0.76 II 21 3,949 III 0.78 I 16 22 1.2 mi South Bartle Hwy 89 41.14.89 121.48.79 3,948 III 0.87 I 17 23 5.6 mi South of Bartle Hwy 89 41.12.10 121.46.04 4,525 III 0.71 II 24 III 0.70 I 18 25 7 mi South Bartle 41.11.95 121.44.96 4,408 0.22 II 26 0.44 I 19 27 7.5 mi South Bartle 41.11.45 121.44.14 4,308 0.62 I 20 28 8 mi South of Bartle Hwy 89 41.11.23 121.43.68 4,321 H 0.55 II 29 H 0.63 I 30 I 0.65 I 21 31 9 mi South of Bartle 41.10.85 121.42.74 4,210 0.82 I 22 32 9.5 mi South of Bartle 41.10.65 121.42.30 4,155 0.25 I 33 0.35 II 23 34 10.1 mi South of Bartle Hwy 89 41.10.44 121.41.70 4,210 0.37 II 35 I 0.18 II 36 I 0.23 I 24 37 11.5 mi South of Bartle 41.09.74 121.39.90 4,073 H 0.13 II 25 38 12 mi South of Bartle Hwy 89 41.09.70 121.39.80 4,195 I 0.18 II 26 39 14.5 mi South of Bartle Hwy 89 41.08.02 121.38.35 4,075 I 0.07 II 40 I 0.16 II 41 0.10 II 27 42 16.1 mi South of Bartle Hwy 89 41.06.63 121.38.44 3,811 I 0.25 I 43 I 0.15 II 44 0.18 II 28 45 18.1 mi South of Bartle Hwy 89 41.04.94 121.38.54 3,800 I 0.12 I 46 0.08 I 47 H 0.11 I 48 I 0.05 II Hybrid zone between Zootermopsis subspecies

Table 1 continued Site# Colony# Location Latitude Longitude Altitude Hydrocarbon Cluster 1 COI probability haplotype

29 49 2 mi South of Hat Creek 40.47.19 121.30.29 4,038 I 0.16 II 30 50 10 mi South of Hat Creek 40.44.54 121.28.33 3,590 I 0.09 II 31 51 3 mi South of Old Station 40.41.73 121.23.28 5,170 0.21 II 32 52 6 mi South of Old Station 40.40.00 121.21.51 5,620 0.06 II 33 53 10 mi South of Old Station 40.38.76 121.18.15 5,455 0.06 II 34 54 Milford 40.07.77 120.20.80 5,593 0.10 II 55 0.07 II 35 56 Milford 40.07.62 120.20.74 5,631 0.09 II 36 57 Milford 40.07.19 120.20.86 6,041 0.19 II Hydrocarbon: III Z. n. nuttingi, I Z. n. nevadensis, H hybrid colony Cluster probability indicates probability that individuals belong to cluster 1 (Z. n. nuttingi) COI haplotype sequence: I Z. n. nuttingi, II Z. n. nevadensis

analyzed the assignment of individuals into clusters by of the hybrid zone: contact with little or no barrier to gene averaging the probabilities of individuals within the same flow, represented by symmetrical sigmoid clines (‘‘Sig,’’ colony belonging to a cluster. two parameters: w, c); and contact with a barrier to gene flow represented by stepped clines (‘‘Step,’’ four parame- Cline fitting and hybrid zone analysis ters: w, c, b, h). The parameters are defined as follows; w is the width of the cline, c is the position of the center of the As suggested by the hydrocarbon phenotypes, cluster cline, b is the strength of the central selective barrier to analysis, and sequence of the COI gene fragment, two hybridization, and h is the exponential decay of cline genetically differentiated lineages are in contact near characters at the tail ends of the cline (Szymura and Barton, Bartle, CA (see results and Aldrich and Kambhampati, 1986). The cline (sigmoid vs. stepped) that best fits the data 2007). To analyze the extent and nature of the hybrid zone is the one with the highest log likelihood. If the log like- between the two lineages, we first identified loci containing lihood values between clines are not significantly different, private alleles between the subspecies with non-negligible the cline with fewest parameters is preferred (Barton, (C0.05) frequencies. All suitable loci were then com- 2000). Comparisons among clines were performed for each pressed into a two-allele system by allocating all alleles model and locus by exploring the likelihood surface step- into ‘‘western’’ or ‘‘eastern’’ alleles (Z. n. nuttingi and Z. n. wise along axes for both cline center c and width w with all nevadensis alleles, respectively) depending on their fre- parameters free to vary at each point as described by quency in the respective subspecies. For the purposes of Phillips et al. (2004). this analysis, all colonies within a site were treated as one subpopulation. The subpopulations were then collapsed Isolation by distance and dispersal distance into an one-dimensional transect beginning with site number 1 in the northwest moving southeast to site number Because of the uncertainties involving mating and identi- 36 (Fig. 1). Maximum likelihood clines were fitted inde- fication of subspecies in the hybrid zone, isolation by pendently across the transect for each locus using distance (IBD) was estimated among colonies 1–9 and 49– subpopulation allele frequencies by means of the com- 57, which according to cuticular hydrocarbon phenotypes pound tanh and exponential model of Szymura and Barton (Haverty et al., 1988), distributional data (Thorne et al., (1986) as implemented in the program ANALYSE (Barton 1993), and mitochondrial haplotypes (Aldrich and Kamb- and Baird, 1995). Clines were fit using four variables, pmax, hampati, 2007) correspond to Z. n. nuttingi and Z. n. pmin (maximum and minimum gene frequency values at nevadensis, respectively. Two methods were used to esti- the tail ends of a cline, respectively), cline center (distance mate IBD by means of a Mantel test with 2,000 from site 1 across transect), and cline width. Subpo- permutations in F-STAT ver. 2.9.3.2 (Goudet, 1995). First, pulations were weighted by effective sample size as IBD was assessed by testing the correlation between implemented in ANALYSE and described by Phillips et al. genetic distance and geographic distance using pair-

(2004). Two hypotheses were tested concerning the nature wise FST/(1 - FST) estimates and logarithm of pairwise B. T. Aldrich, S. Kambhampati geographic distances, as described by Rousset (1997). colonies were located northwest of McCloud, CA (site # 11 Second, IBD was assessed by testing the correlation in Fig. 1; Table 1), and only Z. n. nevadensis colonies were between pairwise colony coefficient of relatedness esti- located southeast of Hat Creek, CA (site # 29 in Fig. 1; mates and logarithm of pairwise geographic distances. Table 1). Colonies with the intermediate hydrocarbon Correlations using pairwise relatedness estimates were phenotypes were located proximal to where Z. n. nevad- chosen because relatedness values take into account the ensis and Z. n. nuttingi distributions abut (Fig. 1; Table 1). allele frequencies in the entire population, whereas pair- Thus, it is likely that colonies with the intermediate wise FST values use only the alleles present in the two hydrocarbon profile are the result of hybridization between colonies (Vargo, 2003). Pairwise FST and relatedness Z. n. nevadensis and Z. n. nuttingi and will be referred to as values among colonies were obtained using the programs hybrid phenotypes. Out of the 30 colonies that were phe- F-STAT and RELATEDNESS ver. 5.0.8 (Queller and notyped, hydrocarbon data identified 14 colonies as Z. n. Goodnight, 1989), respectively. nuttingi,12asZ. n. nevadensis, and 4 as hybrids (Table 1). Dispersal distance per generation was estimated using the relationship between pairwise coefficient of relatedness Genetic differentiation across the hybrid zone estimates and the natural logarithm of distance between sites. The slope (b) of a regression through data of this sort Aldrich and Kambhampati (2007) identified fixed base should yield an estimate of the product of density (d) and substitutions in the mitochondrial COI gene of the two dispersal (r) such that 1/b = 4dr2 (Rousset, 1997). Least subspecies. Examination of all colonies revealed a clear squares regression was used to determine the slopes and pattern in the distribution of the two haplotypes (Table 1), standard errors for Z. n. nuttingi and Z. n. nevadensis data. with haplotype I (Z. n. nuttingi) in colonies located Estimated slopes and the estimated density of Zootermopsis northwest of the contact zone and haplotype II (Z. n. ne- colonies in our sampling area (colonies ha-1) were used to vadensis) in colonies southeast of the contact zone. calculate r2. However, within the contact zone, there is a large overlap Because not all hybrid zones are clinal and may be of about 31 km between the two haplotypes. The log of restricted to specific habitat patches (i.e., a mosaic hybrid probability of the microsatellite data [Ln P(D)] averaged zone), IBD was also examined among colonies 10–48 across the 200 replicates for each number of inferred across the zone of contact. If there is no breakdown of IBD, clusters indicated the highest break in slope and the highest it would suggest that this area of contact was one large log likelihood [Ln P(D) =-1,746.1 ± 3.9 (SE)] occurred clinal hybrid zone. In contrast, a significant breakdown in at K = 2. In other words, the number of clusters (K) that IBD would be suggestive of a mosaic hybrid zone made up best explains the data is two. The probabilities of cluster of a mixture of colonies of the two subspecies and hybrids. for each colony are listed in Table 1 and shown in Fig. 2 Within the hybrid zone, IBD was assessed as described from colony 1 in the northwest to colony 57 in the south- above using a Mantel test and the correlation between east. The assignment of each colony into clusters resulted pairwise colony coefficient of relatedness estimates and in a strong sigmoidal cline across the zone of contact logarithm of pairwise geographic distances. (Fig. 2). That is, colonies in the northwest and southeast of

Results

Hydrocarbon identification

Specimens of the two Z. nevadensis subspecies were sampled from 36 different sites in northern California (Fig. 1). Multiple colonies were sampled at several sites yielding a total of 57 colonies (Table 1). Thirty of the 57 colonies were identified by means of hydrocarbon analysis. The remaining colonies were not hydrocarbon phenotyped because too few individuals were collected to permit both hydrocarbon and microsatellite analysis. We identified three different hydrocarbon profiles corresponding to Z. n. nevadensis, Z. n. nuttingi (Haverty et al., 1988), and one Fig. 2 Probability of assignment for each colony into cluster 1. intermediate phenotype. In agreement with the distribu- Probability of 1 indicates Z. n. nuttingi and 0 indicates Z. n. tional data (Thorne et al., 1993), only Z. n. nuttingi nevadensis Hybrid zone between Zootermopsis subspecies our sampling area were assigned with high probabilities to (Table 3). Because of the significant improvement for clusters 1 and 2, respectively, with intermediate individuals locus ZOOT11 and slight increase in likelihood values for in the center of the sampling area (Table 1). This change in locus ZOOT29 in stepped clines over sigmoidal clines, assignment probability is consistent with the hybridization stepped clines better represent the data and were used for between the two subspecies and supported further investi- cline-fitting analysis (Fig. 3; Table 4). Although the sim- gation into the patterns of hybridization. pler model is preferred over the more complex model when the differences in likelihood are not significant, the Hybrid zone analysis and cline fitting parameters c and w were nearly identical for the optimal stepped and sigmoid clines for locus ZOOT29 (Step: Examination of microsatellite allele frequency data c = 89.47, w = 109.6; Sig: c = 89.62, w = 108.23). revealed that two loci, ZOOT11 and ZOOT29, showed Therefore, our choice of model (stepped vs. sigmoidal) for allelic differences between colonies west and east of locus ZOOT29 did not significantly influence estimates of Bartle, CA. Locus ZOOT11 exhibited two alleles specific the cline width or center and the conclusions drawn from to western colonies, one allele specific to eastern colonies, this analysis. Allowing all parameters to vary resulted in and one allele common to both sets of colonies. Locus similar cline width maximum likelihood estimates (MLE) ZOOT29 exhibited two alleles specific to eastern colonies, for the two clines (MLEs: locus ZOOT11, 100 km; locus one allele private to western colonies, and one allele found ZOOT29, 108 km). Cline center estimates were more in both sets of colonies. Although there were significant variable with ZOOT11 indicating a cline center 137 km differences in allele frequency and private alleles between and locus ZOOT29 indicating a cline center 90 km from the subspecies at other loci, the 10 remaining loci showed site 1 across the transect. Constraining the clines to a significant overlap in allele frequencies and were not suitable for independent cline fitting. Allocation of private Table 3 Maximum likelihood of sigmoidal and stepped models for alleles into the two-allele system for each locus and site is ZOOT11 and ZOOT29 given in Table 2. Depending on the position of the cline center, stepped Locus Sig ML Step ML GdfP clines showed equal or greater likelihood values compared ZOOT11 -345.42 -328.01 34.82 2 \0.01 with sigmoid clines for locus ZOOT11 and locus ZOOT29. ZOOT29 -427.29 -427.26 0.06 2 0.97 The difference in likelihood between the optimal stepped G statistics (twice the difference in log likelihood) are used to assess and sigmoid model was significant for locus ZOOT11, but the significance of the improvement in stepped clines versus sig- was not significantly different for locus ZOOT29 moidal clines

Table 2 Frequency of diagnostic alleles used for cline-fitting analysis at each site in this study Locus Allele Site 1 Site 2 Site 3 Site 4 Site 5 Site 6 Site 7 Site 8 Site 9 Site 10 Site 11 Site 12

ZOOT11 Western 1.00 0.71 1.00 1.00 1.00 1.00 1.00 1.00 1.00 0.55 1.00 1.00 Eastern 0.00 0.29 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.45 0.00 0.00 ZOOT29 Western 1.00 1.00 1.00 1.00 0.00 1.00 1.00 1.00 1.00 0.00 0.76 1.00 Eastern 0.00 0.00 0.00 0.00 1.00 0.00 0.00 0.00 0.00 1.00 0.24 0.00 Locus Allele Site 13 Site 14 Site 15 Site 16 Site 17 Site 18 Site 19 Site 20 Site 21 Site 22 Site 23 Site 24

ZOOT11 Western 0.89 1.00 1.00 0.96 1.00 0.88 1.00 1.00 0.00 0.50 1.00 1.00 Eastern 0.11 0.00 0.00 0.04 0.00 0.12 0.00 0.00 1.00 0.50 0.00 0.00 ZOOT29 Western 0.36 0.93 0.00 0.67 0.00 1.00 0.52 0.68 0.00 0.40 0.49 0.00 Eastern 0.64 0.07 1.00 0.33 1.00 0.00 0.48 0.32 1.00 0.60 0.51 1.00

Locus Allele Site 25 Site 26 Site 27 Site 28 Site 29 Site 30 Site 31 Site 32 Site 33 Site 34 Site 35 Site 36

ZOOT11 Western 1.00 0.58 0.28 0.68 0.00 0.00 0.00 0.36 0.00 0.24 0.00 0.00 Eastern 0.00 0.42 0.72 0.32 1.00 1.00 1.00 0.64 1.00 0.76 1.00 1.00 ZOOT29 Western 0.00 0.03 0.19 0.44 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 Eastern 1.00 0.97 0.81 0.56 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00

Alleles for two loci, ZOOT11 and ZOOT29, are allocated into Western (Z. n. nuttingi) and eastern (Z. n. nevadensis) alleles depending on their frequency in the respective groups B. T. Aldrich, S. Kambhampati

Subspecies-specific markers, hybrid colonies, and cline center

Within the estimated hybrid zone (sites 6–30; using the average width and center for the 2 loci) we observed 33 colonies (73%) with some combination of subspecies- specific cuticular hydrocarbons, microsatellite alleles, or mitochondrial haplotypes from both subspecies (see Table 1). First generation hybrids from matings between the two ‘‘pure’’ subspecies are expected to possess private microsatellite alleles from both subspecies at both loci. Considering all colonies sampled, we observed 11 (19%) with Z. n. nuttingi and Z. n. nevadensis private alleles at Fig. 3 Maximum likelihood fitted sigmoidal cline curves for diag- nostic loci ZOOT11 (solid black line) and ZOOT29 (dashed gray both loci. Furthermore, if hybrid colonies were fertile, we line). Distance across X axis is in km, starting from the northwest- might expect to find colonies with private alleles from ernmost site ending in the southeastern most site. The allele frequency both subspecies at one locus, but only the private alleles of Z. n. nevadensis alleles (Y axis) for each site across the axis is from one subspecies at the other locus. To this end, we plotted for OOT11 (black squares) and ZOOT29 (gray circles) observed 24 colonies (42%), with private alleles from both subspecies at one locus and private alleles from one Table 4 Summary of best-fit cline parameters using stepped clines of the subspecies at the other locus. These data suggest

Locus w c bSE– hSE– bNW– hNW– ML extensive hybridization between the subspecies within our (km) (km) NW NW SE SE sampling area and at least some capacity for hybrids to reproduce. However, the estimates of barrier strength ZOOT11 100 137 179.97 1.00 1.42 0.55 -328.01 generated in the cline-fitting analysis suggested an uneven ZOOT29 108 90 24.81 1.00 15.09 1.00 -427.26 gene flow between the subspecies. Assuming a cline w cline width, c cline center from site 1 across the transect, b selective center located between sites 23 and 24 (114 km from site barrier strength from southeast to northwest (SE–NW) and northwest 1; estimated using the average center of the 2 independent to southeast (NW–SE), h rate of exponential decay from southeast to northwest (SE–NW) and northwest to southeast (NW–SE), ML loci) we could find no obvious pattern in the distribution maximum likelihood of colonies with microsatellite alleles from only one subspecies, hydrocarbon phenotypes, or mitochondrial common cline width (104 km; average width of the two haplotypes around the cline center that would provide clines) did not result in a significant decrease in likelihood insight into potential uneven geneflow between the

(GW same-W diff = 1.88, 2 df, P = 0.39). However, con- subspecies. straining the clines to a common center (114 km; average center for the two clines) resulted in a significant decrease IBD, dispersal distance, and selection strength in likelihood (GC same-C diff = 315.30, 2 df, P \ 0.01). Although the width of the cline can be reasonably There was a significant correlation between genetic and approximated as 104 km, the position of the cline across geographic distance for Z. n. nuttingi and Z. n. nevadensis the transect cannot be clearly defined and may range from colonies 1–9 and 49–57, respectively. Correlation tests 90 to 137 km from site 1 across the transect. using pairwise coefficient of relatedness estimates resulted For stepped clines, additional cline characteristics can in larger correlations and more significant probabilities be obtained from the estimates of barrier strength (b) and compared with tests using FST/(1 - FST) estimates (data rate of exponential decay (h). For both clines, the selective not shown). The correlation coefficient (r =-0.18; barrier from southeast (Z. n. nevadensis) to northwest (Z. n. P \ 0.01) between pairwise relatedness estimates and the nuttingi) was significantly greater than the barrier from logarithm of geographic distance for Z. n. nuttingi and Z. n. northwest to southeast (Table 4; v2 = 94, 2 df, P \ 0.01). nevadensis was identical. Thus, relatedness between colo- The average selective barrier from the southeast to nies is decreasing as geographic distance is increasing. northwest for the two clines was 102 km and was nearly as Restricting the test to colony pairs separated by \30 km large as the maximum cline width estimate (ZOOT29 = improved the correlations somewhat (Fig. 4). In the 108 km). Furthermore, the average decay parameter esti- restricted tests, however, correlation between pairwise mates were larger northwest of the barrier region when relatedness estimates and the logarithm of geographic compared with estimates southeast of the barrier region distance for Z. n. nuttingi (r =-0.23; P \ 0.05) varied (Table 4). from that for Z. n. nevadensis (r =-0.29; P \ 0.01). Hybrid zone between Zootermopsis subspecies

P \ 0.05). Restricting the test to colony pairs separated by \30 km lowered the correlation between pairwise relat- edness estimates and the logarithm of geographic distance (r =-0.25; P \ 0.05). However, the correlations between genetic and geographic distance for colonies separated by \30 km were comparable for the Z. n. nuttingi, Z. n. nevadensis, and hybrid zone colonies (r =-0.23, -0.29, and -0.25, respectively).

Discussion

Characterization of the hybrid zone

Using microsatellite loci and one mitochondrial locus, we identified genetic differences between the subspecies. As expected with hybridization between two populations with unique nuclear markers, we observed a gradual change in allele frequencies moving northwest to southeast across our sampling area from colonies displaying Z. n. nuttingi alleles to those displaying Z. n. nevadensis alleles with intermediate individuals in the center of the contact zone. Fig. 4 Isolation by distance analysis for Z. n. nevadensis and Z. n. Consistent with Thorne et al. (1993), several colonies nuttingi colonies. The relationship between pairwise coefficients of sampled in the area of distributional overlap were inter- relatedness for colonies (Y axis) and log of geographic distance (X mediate with respect to the originally described hydro- axis) is shown. Zootermopsis n. nuttingi: r =-0.23, P \ 0.05 and Zootermopsis n. nevadensis: r =-0.29, P \ 0.01 carbon phenotypes I and III. Thus, our analyses of genetic markers confirm the existence of a hybrid zone between the two Z. nevadensis subspecies centered within an area 6 km Regression of the pairwise coefficient of relatedness to the northwest and 32 km to the southeast of Bartle, CA. estimates and log distance estimates generated average Although colonies with intermediate hydrocarbon phe- slopes of -0.02 ± 0.008 (SE) and -0.03 ± 0.0006 for Z. notypes were identified in this study, hydrocarbon n. nuttingi and Z. n. nevadensis, respectively. Although not phenotypes were not always reflective of colonies des- used in these analyses, the regression of all pairwise cended from hybridization between the two subspecies. combinations (data not shown) displayed nearly identical That is, some colonies containing combinations of sub- slopes to regression of the restricted dataset (pairs species-specific microsatellite alleles and mitochondrial \30 km). Slope estimates and standard errors were then haplotypes from both subspecies were not identified as used as bounds for estimating dispersal ranges for hybrids through hydrocarbon analysis (e.g., colonies 19, Zootermopsis. Using the estimated density of Zootermopsis 20, 23, 30, 36, and 42). Very little is known about how colonies within our sampling area (d = 2 colonies per ha) hydrocarbon phenotypes are inherited in Zootermopsis. yielded dispersal values (r) ranging from 164 to Plausibly, hydrocarbon analysis may not be sensitive 375 m gen-1/2. enough to detect all hybrid colonies. For example, inter- Assuming that the cline in this study has reached equi- mediate hydrocarbon phenotypes may only be detectable librium between selection and migration, we next from first generation hybrids and any subsequent crosses calculated the effective level of selection, s*. We used the between hybrids or between hybrids and one of the two equation s* = 8(r/w)2 (Barton and Gale, 1993) which subspecies may result in an incongruence between hydro- calculates the selection against heterozygotes required to carbon phenotypes and genetic markers. maintain a cline of a specific width. Using our estimates of The two clines generated in this study showed signifi- cline widths (w = 100–108 km), estimates of dispersal cantly different estimates for cline centers, but not cline (r = 164 to 375 m gen-1/2), and the above equation we widths. Differences in cline center estimates between locus estimate a range of s* from 7.4 9 10-5 to 4.5 9 10-4. ZOOT11 and locus ZOOT29 can be best explained by Within the hybrid zone (colonies 10–48), there was a sig- differences in the number of private alleles for each locus nificant correlation between pairwise relatedness estimates and their frequencies. Cline centers were skewed to the and the logarithm of geographic distance (r =-0.30; right for locus ZOOT11 and skewed to the left for locus B. T. Aldrich, S. Kambhampati

ZOOT29; that is, each locus-specific cline is skewed in the et al., 1993) and some habitat specialization between the direction of the subspecies with the fewest private alleles or subspecies cannot be ruled out. A GIS-based analysis lowest frequency of private alleles for that locus. Poten- (e.g., Kambhampati and Peterson, 2007) may shed some tially, a reexamination of this hybrid zone using additional light on this issue. loci might help clarify the ‘‘true’’ cline center. Considering the distribution of cuticular hydrocarbon phenotypes it was Reproductive compatibility between the Z. nevadensis somewhat unexpected that we found such a large hybrid subspecies zone in this area. According to cuticular hydrocarbons, hybrid colonies are located within a relatively small area Given the changes in microsatellite allele frequencies and (*29 km). However, our genetic data suggest a much mtDNA haplotypes across a large contact zone and the larger hybrid zone (*104-km wide) exists in this area. weak selection against hybrids, there appears to be exten- Thus, additional sampling over a much larger area, par- sive hybridization between the two subspecies of Z. ticularly near the edges of the estimated hybrid zone, is nevadensis. The rather large cline width relative to dis- needed to more precisely define the boundaries of the persal resulted in a relatively weak effective selection hybrid zone. Potentially, this may reveal geographic or coefficient (s*)of\0.05%. This estimate is substantially habitat factors influencing the hybridization dynamics lower than those from studies involving distinct species, between the subspecies. all showing s* [ 20% (e.g., Szymura and Barton, 1991; Although stepped clines best explained the data, we Mallett et al., 1990; Porter et al., 1997). Our estimate of explored the possibility that this area may actually consist weak selection against hybrids was supported by the of a large mosaic hybrid zone (e.g., Britch et al., 2001; significant number of colonies with mixed genetic and Rand and Harrison, 1989, Ross and Harrison, 2002; Sites phenotypic markers from both subspecies and our inference et al., 1995; Vines et al., 2003), made up of distinct that hybrid colonies are likely reproducing. habitat and population patches, at a scale smaller than Although the width of the cline relative to dispersal was surveyed for this study. Potentially, at larger scales indicates relatively weak selection against hybrids for at mosaic hybrid zones can create the appearance of a large least one locus, stepped clines best explained the data clinal hybrid zone. However, we found no evidence for a indicating that some barrier to gene flow exists in the breakdown in IBD within the hybrid zone, as would be sampling area. Cline-fitting estimates of barrier strength expected in a mosaic hybrid zone. Within the hybrid (b) and rate of exponential decay (h) suggested that there zone, most colonies contain some combination of sub- is a stronger selective barrier from east to west (i.e., Z. n. species-specific cuticular hydrocarbons, microsatellite nevadensis to Z. n. nuttingi) relative to the barrier from alleles, or mitochondrial haplotypes from both subspecies. west to east; that is, there appears to be a greater intro- This suggests that ‘‘pure’’ subspecies colonies are infre- gression of Z. n. nuttingi alleles into Z. n. nevadensis than quent within the area sampled and inter-subspecies vice versa. We were unable to identify any obvious pat- matings occur fairly frequently. Unfortunately, the terns (e.g., unidirectional reproductive incompatibility) in extensive nature of this hybrid zone and low colony the nuclear or mitochondrial markers that would explain densities in this area prevent us from unambiguously the uneven gene flow between the subspecies. For identifying true nature of this hybrid zone (i.e., clinal vs. instance, colonies containing subspecies-specific micro- mosaic). Thus, we do not rule out the possibility that this satellite alleles from both subspecies were observed in hybrid zone is actually a mosaic occurring at a scale both mitochondrial haplotypes (data not shown). Thus, larger than that sampled in this study. Additional sam- males and females of each subspecies appear to be pling near the edges of our estimated hybrid zone or equally compatible and generate viable and fertile off- perpendicular to our sampling transect may reveal some spring. For now, we are unable to determine which model mosaic structure. Furthermore, the low colony densities (sigmoidal vs. stepped) best explains this zone of contact; may make it difficult to detect the presence of a fine-scale however, a reexamination of this area using additional mosaic structure. Therefore, although the hybrid zone molecular markers may help clarify the pattern of gene appears clinal in our analyses, additional information at flow in this area. Controlled mating experiments may also both larger and smaller scales is needed before any provide some clues and help determine if our analysis is definitive conclusions can be made. Finally, it is not detecting true inter-subspecies differences in hybridization known whether the two subspecies differ in their habitat or hybrid fitness not detectable by the molecular markers associations or exhibit ecological specialization. If they used here. If a barrier to geneflow does exist in this differ in their ecological associations, a mosaic hybrid hybrid zone, it may provide a unique opportunity to study zone becomes a more likely scenario. The distribution of the evolution of reproductive isolation mechanisms in the Z. nevadensis subspecies is largely allopatric (Thorne social . Hybrid zone between Zootermopsis subspecies

Evolution and taxonomic status of Z. nevadensis Barton N.H. and Gale K.S. 1993. Genetic analysis of hybrid zones. In: subspecies Hybrid Zones and the Evolutionary Process (Harrison R.G., Ed), Oxford Univ. Press: Oxford, UK. pp 13–45 Barton N.H. and Hewitt G.M. 1985. Analysis of hybrid zones. Annu. The weak selection against hybrids suggests subspecies- Rev. Ecol. Syst. 16: 113–148 specific alleles should readily pass through the zone of Barton N.H. and Hewitt G.M. 1989. Adaptation, speciation, and contact, but possibly with greater gene flow from Z. n. hybrid zones. Nature 341: 497–503 Blum M.J. 2002. Rapid movement of a Heliconius hybrid zone: nuttingi to Z. n. nevadensis populations. Potentially, this evidence for phase III of Wright’s shifting balance theory? may cause a shift in the location of the hybrid zone, a phe- Evolution 56: 1992–1998 nomenon observed in other species and believed to provide Britch S.C., Cain M.L. and Howard D.J. 2001. Spatio-temporal evidence for Wright’s Shifting Balance theory (Dasmaha- dynamics of the Allonemobius fasciatus- A. socius mosaic hybrid zone: a 14-year perspective. Mol. Ecol. 10: 627–638 patra et al., 2002; Blum, 2002). Temporal sampling over a Broughton R.E. Jr. and Kistner D.H. 1991. A DNA hybridization larger geographic range may help determine the evolution- study of the termite genus Zootermopsis (Isoptera: Termopsi- ary fate of the two subspecies. For instance, the two dae). Sociobiology 19: 15–40 subspecies may be differentially adapted to their own unique Bun˜o I., Torroja E., Lo´pez-Ferna´ndez C., Butlin R.K., Hewitt G.M. and Gosa´lvez J. 1994. A hybrid zone between two subspecies of climate and habitat, resulting in a mosaic hybrid zone at a the grasshopper Chorthippus parallelus along the Pyrenees: the larger scale not detectable in this study. Alternatively, the west end. Heredity 73: 625–634 zone of overlap may be acceptable to both subspecies, but Dasmahapatra K.K., Blum M.J., Aiello A., Hackwell S., Davies N., selection against hybrids may increase as hybrid colonies Berminghan E.P. and Mallet J. 2002. Inferences from a rapidly moving hybrid zone. Evolution 56: 741–753 spread west and east into Z. n. nuttingi and Z. n. nevadensis Eggleton P. 2000. Global patterns of termite diversity. In: Termites: territories, respectively. In contrast, the two subspecies may Evolution, Sociality, Symbioses, Ecology (Abe T., Bignell D.E. be in the process of converging into a single evolutionary and Higashi M., Eds), Kluwer Academic Publishers: Dordrecht, unit. However, given that the two subspecies are actively The Netherlands. 25–51 pp Fel-Clair F., Catalan J., Lenormand T. and Britton-Davidian J. 1998. mating in nature, but display distinctive phenotypic and Centromeric incompatibilities in the hybrid zone between house genetic differences outside of the hybrid zone, their current mouse subspecies from Denmark: evidence from patterns of nor taxonomic status as subspecies seems appropriate. activity. Evolution 52: 592–603 Goudet J. 1995. F-STAT Version 1.2: a computer program to Acknowledgments We thank M. Haverty for alerting us to the calculate F-statistics. J. Hered. 86: 485–486 presence and location of the hybrid zone between the two subspecies. Harrison R.G. 1990. Hybrid zones: windows on the evolutionary We thank M. Haverty and L. Nelson for identifying and help in process. Oxford Surv. Evol. Biol. 7: 69–128 collecting Zootermopsis. Without their help, this study would not Haverty M.I. and Thorne B.L. 1989. Agonistic behavior correlated have been possible. We thank E. Krafsur and B. Thorne for helpful with hydrocarbon phenotypes in dampwood termites, Zooter- comments. We thank E. Vargo C. DeHeer, and J. Marshall for mopsis (Isoptera: ). J. Behav. 2: 523–543 reviewing earlier versions of this manuscript and providing con- Haverty M.I., Page M., Nelson L.J. and Blomquist G.J. 1988. structive criticism. The distribution map was prepared by the Cuticular hydrocarbons of dampwood termites, Zootermopsis: Geographic Information Systems Spatial Analysis Laboratory intra- and intercolony variation and potential as taxonomic (GISSAL), Department of Geography, Kansas State University. This characters. J. Chem. Ecol. 14: 1035–1058 study was funded by a National Science Foundation grant Hewitt G.M. 1996. Some genetic consequences of ice ages, and their (DEB9806710 to S. K.). This is journal article number 10-050-J of the role in divergence and speciation. Biol. J. Linn. Soc. 58: 247–276 Kansas Agricultural Experiment Station. Hewitt G.M. 2000. The genetic legacy of the quaternary Ice Ages. Nature 405: 907–913 Kambhampati S. and Peterson A.T. 2007. Ecological niche conser- vation and differentiation in the wood-feeding cockroaches, References Cryptocercus, in the United States. Biol. J. Linn. Soc. 90: 457– 466 Aldrich B.T. and Kambhampati S. 2004. Microsatellite markers for Largiade`r C.R., Klingenberg C.P. and Zimmermann M. 1994. two species of dampwood termites in the genus Zootermopsis Morphometric variation in a hybrid zone of two subspecies of (Isoptera: Termopsidae). Mol. Ecol. Notes 4: 719–721 Gerris costae (Heteroptera: Gerridae) in the Maritime Alps. J. Aldrich B.T. and Kambhampati S. 2007. Population structure and Evol. Biol. 7: 697–712 colony composition of two Zootermopsis nevadensis subspecies. Lepage M. and Darlington J.P.E.C. 2000. Population dynamics of Heredity 99: 443–451 termites. In: Termites: Evolution, Sociality, Symbioses, Ecology Arnold M.L. 1997. Natural Hybridization and Evolution. Oxford (Abe T., Bignell D.E. and Higashi M., Eds), Kluwer Academic Univ. Press: New York, NY Publishers: Dordrecht, The Netherlands. pp 333–361 Avise J.C., Walker D. and Johns G.C. 1998. Speciation durations and Mallett J., Barton N., Lamas G.M., Santisteban J.C., Musedas M.M. Pleistocene effects on vertebrate phylogeography. Proc. R. Soc. and Eeley H. 1990. Estimates of selection and gene flow from Lond. B Biol. 265: 1707–1712 measures of cline width and linkage disequilibrium in Heliconius Barton N.H. 2000. Estimating multilocus linkage disequilibria. hybrid zones. Genetics 124: 921–936 Heredity 84: 373–389 Phillips B.L., Baird S.J.E. and Moritz C. 2004. A narrow zone of Barton N.H. and Baird S.J.E. 1995. ANALYSE: an application for secondary contact between morphologically cryptic phylogeo- analyzing hybrid zones. Freeware, Edinburgh. Available at graphic lineages of the rainforest skink, Carlia rubrigularis. http://helios.bto.ed.ac.uk/evolgen/mac/analyse Evolution 58: 1536–1548 B. T. Aldrich, S. Kambhampati

Porter A.H., Wenger R., Geiger H., Scholl A. and Shapiro A.M. 1997. Szymura J.M. and Barton N.H. 1986. Genetic analysis of a hybrid The Pontia daplidice-edusa hybrid zone in northwestern Italy. zone between the fire-bellied toads, Bombina bombina and B. Evolution 51: 1561–1573 variegata near Cracow in southern Poland. Evolution 40: 1141– Pritchard J.K., Stephens M. and Donnelly P.J. 2000. Inference of 1159 population structure using multilocus genotype data. Genetics Szymura J.M. and Barton N.H. 1991. The genetic structure of the 155: 945–959 hybrid zone between the fire-bellied toads Bombina bombina and Queller D.C. and Goodnight K.F. 1989. Estimating relatedness using B. variegata: Comparisons between transects and loci. Evolution genetic markers. Evolution 43: 258–275 45: 237–261 Rand D.M. and Harrison R.G. 1989. Ecological genetics of a mosaic Thorne B.L., Haverty M.I., Page M. and Nutting W.L. 1993. hybrid zone: mitochondrial, nuclear, and reproductive differen- Distribution and biogeography of the North American termite tiation of crickets by soil type. Evolution 43: 432–449 genus Zootermopsis (Isoptera: Termopsidae). Ann. Entomol. Soc. Ross C.L. and Harrison R.G. 2002. A fine-scale analysis of the mosaic Am. 86: 532–544 hybrid zone between Gryllus firmus and Gryllus pennsylvanicus. Vargo E.L. 2003. Hierarchical analysis of colony and population Evolution 56: 2296–2312 genetic structure in the eastern subterranean termite, Reticuli- Rousset F. 1997. Genetic differentiation and estimation of gene flow termes flavipes, using two classes of molecular markers. from F-statistics under isolation by distance. Genetics 145: Evolution 57: 2805–2818 1219–1228 Vines T.H., Ko¨hler S.C., Thiel M., Ghira I., Sands T.R., MacCallum Sequeira F., Alexandrino J., Rocha S., Arntzen J.W. and Ferrand N. C.J., Barton N.H. and Nu¨rnberger B. 2003. The maintenance of 2005. Genetic exchange across a hybrid zone within the Iberian reproductive isolation in a mosaic hybrid zone between the fire- endemic golden-striped salamander, Chioglossa lusitanica. Mol. bellied toads Bombina bombina and B. variegata. Evolution 57: Ecol. 14: 245–254 1876–1888 Sites J.W. Jr., Nicholas H.B. and Reed K.M. 1995. The genetic Wake D.B. 1997. Incipient species formation in salamanders of the structure of a hybrid zone between two chromosome races of the Ensatina complex. Proc. Natl. Acad. Sci. USA 94: 7761–7767 Sceloporus grammicus complex (Sauria, Phrynosomatidae) in Wang H., McArthur E.D. and Freeman D.C. 1999. Narrow hybrid central Mexico. Evolution 49: 9–36 zone between two subspecies of big sagebrush (Artemisia Smadja C., Catalan J. and Ganem G. 2004. Strong premating tridentata: Asteraceae). IX. Elemental uptake and niche separa- divergence in a unimodal hybrid zone between two subspecies of tion. Am. J. Bot. 86: 1099–1107 house mouse. J. Evol. Biol. 17: 165–176