Mitochondrial DNA Sequencing Reveals Extreme Genetic Differentiation in a Cryptic Species Complex of Neotropical Pseudoscorpions

MOLECULAR PHYLOGENETICS AND EVOLUTION

Vol. 7, No. 2, APRIL, pp. 208–216, 1997 ARTICLE NO. FY960388

Mitochondrial DNA Sequencing Reveals Extreme Genetic Differentiation in a Cryptic Species Complex of Neotropical Pseudoscorpions1 Thomas P. Wilcox, Lori Hugg, Jeanne A. Zeh, and David W. Zeh Department of Biology, University of Houston, Houston, Texas 77204-5513 Received June 13, 1996; revised October 7, 1996

The neotropical pseudoscorpion Cordylochernes scorpioides (Chernetidae: Lamprochernetinae) is currently described as a single species ranging from Central America to northern Argentina. However, interpopulation crosses have recently demonstrated that C. scorpioides actually represents a complex of cryptic species. Here we present mitochondrial COI gene sequence data from C. scorpioides individuals from Panama, Trinidad, and French Guiana which demonstrate little or no intrapopulation variability but divergence ranging from 2.6 to 13.8% between geographic populations. Phylogenetic analysis provides evidence of a major split between C. scorpioides lineages from Central and South America. Levels of interpopulation mtDNA divergence correspond well with previously established patterns of postzygotic reproductive incompatibility between geographically distinct units within the C. scorpioides complex. By contrast, multivariate morphometric analysis demonstrates that extensive sequence divergence has occurred in the absence of appreciable morphological differentiation between the populations. To provide a framework for assessing the scale of geographic divergence in C. scorpioides, Cordylochernes sequences were compared with homologous sequence from its presumed sister taxon, Lustrochernes, and from Parachernes and Semeiochernes, representatives of the second chernetid subfamily, the Chernetinae. Our preliminary, generic-level analysis suggests that COI sequence data may prove useful in resolving relationships within this problematic family. r 1997 Academic Press

INTRODUCTION The harlequin beetle-riding pseudoscorpion Cordylochernes scorpioides provides a model system for comprehensive investigation of diversification in the Neotropics. Ranging throughout rain forests from southern 1 Sequence data from this article have been deposited with the GenBank Data Libraries under Accession Nos. U70216–U70228.

1055-7903/97 $25.00 Copyright r 1997 by Academic Press All rights of reproduction in any form reserved.

Mexico to northern Argentina, this terrestrial arthropod has proved to be highly amenable not only to investigation at the molecular (Zeh et al., 1992a, 1994), behavioral (Zeh and Zeh, 1992a), and morphological (Zeh and Zeh, 1994) levels but also to assessment of interpopulation reproductive compatibility (Zeh and Zeh, 1994). Males of this species have been described as the most phenotypically variable pseudoscorpions known (Beier, 1948), and it was this extremely high level of variability in sexually dimorphic male traits that resulted in the original classification of C. scorpioides as four separate species (Beier, 1932a). Subsequent reexamination of hundreds of specimens revealed continuous variation in male morphology both within and between populations (Beier, 1948). This finding led Beier (1948) to classify C. scorpioides as a single, pan-neotropical species in the genus Cordylochernes (Chernetidae). Recent studies involving multilocus, minisatellite DNA fingerprinting (Zeh et al., 1992a), protein electrophoresis, morphometric discriminant function analysis, and reproductive compatibility experiments (Zeh and Zeh, 1994) have demonstrated, however, that populations of C. scorpioides from French Guiana and central Panama, although morphologically indistinguishable, are, in fact, genetically distinct and reproductively isolated at the postzygotic stage (Zeh and Zeh, 1994). Here, we report the results of a study designed to evaluate the utility of mitochondrial COI gene sequence data as indicators of evolutionary lineages within the C. scorpioides species complex. To provide a framework for assessing the scale of geographic divergence in C. scorpioides, mtDNA sequence data from four C. scorpioides populations were compared with homologous sequence from Lustrochernes, the probable sister taxon to Cordylochernes (Muchmore, 1984, personal communication) and from two additional chernetid genera, Parachernes and Semeiochernes. We also sequenced the same gene region in Paratemnoides elongatus, a member of the Atemnidae, to serve as an outgroup. The Atemnidae is hypothesized to be a close relative to the

208

Cordylochernes mtDNA PHYLOGENY

Chernetidae based on cladistic analyses of morphological and behavioral traits (Harvey, 1992; Proctor, 1993). MATERIALS AND METHODS Study Populations Individuals of the Cordylochernes scorpioides species complex were obtained from two sites in Panama: Parque Nacional Soberanı´a in the Canal Zone (CZ) and western Chiriqui province near the border with Costa Rica (CHIR), approximately 10 km north of Santa Clara (see Fig. 1). Individuals from South American populations of Cordylochernes were collected from the Blanchisseuse Road in north central Trinidad (TRIN) and from the Piste du Kaw in French Guiana (FG) (see Zeh et al., 1992b). Semeiochernes armiger individuals were obtained from Parque Nacional Soberanı´a and Trinidad, Lustrochernes consocius from Barro Colorado Island (CZ), and Parachernes setosus from Parque Nacional Soberanı´a. The subsocial atemnid species, Paratemnoides elongatus (Brach, 1978; Zeh and Zeh, 1990), was collected from four colonies near Gamboa, Panama (CZ). Colonies were sampled on both the western (Gamboa and Cerro Pelado) and eastern (C25B and Camino de Cruces) sides of the Rı´o Chagres, a large river which is the primary freshwater source of the Panama Canal. Pseudoscorpions used in this study were either field-collected individuals or laboratoryreared offspring of field-inseminated or laboratorymated females. Voucher specimens have been deposited with W. B. Muchmore (University of Rochester), V. Mahnert (Museum d’Histoire Naturelle, Switzerland), and D. Quintero (Universidad de Panama´). DNA Extraction, PCR Amplification, and DNA Sequencing We extracted pseudoscorpion DNA from single individuals, using the 23 CTAB protocol described by Zeh et al. (1992a). Standard PCR techniques and degenerate versions of the primers Ron (58 GGAKCACCTGATATAGCATTYCC 38) and Nancy (58 CCCGGTAARATTAAAATATAAACTTC 38) (Simon et al., 1994) were used to amplify an approximately 440-bp fragment of the mitochondrial cytochrome oxidase subunit 1 (COI) gene, corresponding to the region between 1751 and 2191 bp in the Drosophila yakuba mtDNA sequence (Simon et al., 1994), from the following specimens: 6 C. scorpioides individuals from CZ; 2 C. scorpioides each from CHIR, FG, and TRIN; 2 L. consocius from CZ; 2 S. armiger from CZ and 1 from TRIN; 1 P. setosus from CZ; and 6 P. elongatus from CZ. PCR products were dideoxy sequenced (Sanger et al., 1977) in both the 58 and 38 directions with the Direct PCR Sequencing kit (Amersham), 35S (Amersham Redivue), and either the 58 PCR primer (Ron) or a modified, nondegenerate 38 primer (58 ATTAAAATATAAACTTC 38). DNA sequences were not determined for primer

209

regions nor those immediately adjacent to them. Sequence autoradiographs were read and compiled using MacDNASIS (Hitachi v3.2, 1994). Phylogenetic Analysis Since there were no insertions or deletions, manual alignment with the MacDNASIS multiple sequence editor (Hitachi v3.2, 1994) was straightforward. The invertebrate mitochondrial code was used to translate aligned sequences, and phylogenies were inferred from both nucleotide and amino acid sequences. Based on cladistic analysis of morphological characters, the Atemnidae is hypothesized to be closely related to the Chernetidae (Harvey, 1992). We thus placed the four P. elongatus sequences in the outgroup. Additionally, characters were assumed to be unpolarized for all analyses. Two optimality criteria were applied during tree searches: maximum parsimony and maximum likelihood (Felsenstein, 1981). Branch and bound parsimony searches were carried out using PAUP v3.1.1 (Swofford, 1994), and all minimal trees were saved (MULPARS). If multiple, most parsimonious trees were recovered, a single tree was selected by using Farris’ (1969) successive weighting approach with PAUP’s REWEIGHT command. Maximum likelihood tree estimations (Felsenstein, 1981) for nucleotide and amino acid sequences were performed with PHYLIP’s DNAML and PROTML, respectively (Felsenstein, 1995). All PHYLIP analyses employed global branch swapping with 10 replicate random taxa additions. Finally, support for each clade was examined by analyzing 200–500 bootstrap replicate datasets for each tree estimation method employed (Felsenstein, 1985). We also calculated Bremer’s support indices for all parsimony analyses (Bremer, 1994). In the parsimony analyses, we examined the effects of several weighting schemes on tree topology: equal character weighting, character weighting by codon position, transversion weighting, and combined codon position and transversion weighting. Codon position weights of 4-8-1 were applied to the first, second, and third codon positions, respectively, while transversions were weighted three times more heavily than transitions (see below). In maximum likelihood analyses, equivalent alternative character weighting schemes were obtained by using the categories option of DNAML and assuming no correlations among sites (Felsenstein and Churchill, 1996). To estimate transition bias, we used the Changes and Stasis procedure in MacClade v3.0.4 (Maddison and Maddison, 1992) to enumerate transitions and transversions over a series of successively pruned most parsimonious (MP) trees. Beginning with an equally weighted MP tree, we successively pruned the most distant taxa in the following order: Paratemnoides, Parachernes, Semeiochernes, Lustrochernes, and Cordylochernes FG 1 TRIN. A transition:transversion weighting scheme of 1:3 was employed because transitions were found to

210

WILCOX ET AL.

occur approximately three times as often as transversions in cases where transitions were not yet saturated.

reciprocal substitutions (Table 1), with the degree of bias depending upon taxa included in the analysis. A & T accounted for approximately 70% of all transversions.

Morphological Analysis Multivariate morphological divergence in the C. scorpioides species complex was quantified from measurements of 10 male and 10 female pseudoscorpions from each of the four populations. Measured individuals were reared under standard conditions in the laboratory (see Zeh and Zeh, 1992b) in order to minimize environmental effects on trait expression. High magnification (approximately 303) video images of live pseudoscorpions, restrained under a glass slide with the right pedipalp fully extended, were loaded into NIH Image (version 1.58) in order to compute nine linear measurements of the pedipalps and cephalothorax (for trait descriptions, see Zeh and Zeh, 1992a). We used discriminant analysis (Dillon and Goldstein, 1986) to estimate canonical population means, with variances around the means estimated by projecting 1000 bootstrap replicate sample means from each population into canonical space. Classification error rate estimates were obtained with the crossvalidate option in the SAS procedure DISCRIM, a method which measures the mean error rate in classifying a single individual to its source population when the discriminant function is computed from the jackknifed data set (SAS Institute, 1989b). All other morphological analyses were carried out using SAS IML (SAS Institute, 1989a). RESULTS Sequence Characteristics Analysis of 313 bp of COI sequence, corresponding to positions 1804 through 2117 of the D. yakuba sequence (Clary and Wolstenholme, 1985), from the 24 pseudoscorpion individuals, revealed 13 unique sequences, with no variation in length observed. We detected no missense or stop codons when sequences were translated, using the invertebrate mitochondrial codon table. Additionally, all translated sequences reveal high amino acid sequence homology with arthropod COI sequences deposited in GenBank, strongly suggesting that all sequences are from functional COI genes. Of the 313 bp sequenced, 175 bp are invariant (56%), 24 are autapomorphic (8%), and 114 are potentially phylogenetically informative (36%). Of the potentially phylogenetically informative sites, 69% occur at third codon positions, 21% at first positions, and 10% at second positions. In addition, we observed A or T at 66% of all nucleotide sites, with A1T bias most apparent at third codon positions (84%). Transition:transversion ratios, estimated from changes over successively pruned unweighted MP trees, ranged from 1:1 to 3.75:1, and increased as the maximum pairwise divergence among included taxa decreased. A = G and T = C substitutions occurred 2.6 to 3 times more frequently than their

mtDNA Sequence Variation Within-population COI sequence variation appears to be low in Cordylochernes, ranging from 0% in the FG, TRIN, and CZ populations to 0.64% between the two sequences identified in the CHIR population. For each of the CZ and FG populations, low intrapopulation variability was verified by analysis of partial sequences (200 bp) from an additional four individuals. Again, no within-population variation was detected. At the between-population level, pairwise sequence divergence in C. scorpioides varies extensively, ranging from 2.6 to 13.8% (Table 2). FG and TRIN populations exhibit the least divergence, while sequence divergence is greatest between the CHIR and TRIN populations. In general, pairwise divergence corresponds to the magnitude of geographic distance separating populations (Fig. 1). However, sequences from the CHIR and CZ populations, which are separated by only 385 km, differ at 8.2% of their nucleotides. By contrast, sequences from FG and TRIN are only approximately 33% as divergent, despite a threefold greater geographic distance between the populations (Fig. 1, Table 2). The mtDNA divergence between populations of C. scorpioides from FG and TRIN is completely silent, while two amino acid differences (AA positions 1 and 28) exist between these South American populations and the CZ population from Panama. An additional amino acid difference (position 23) distinguishes the CHIR population from the remaining C. scorpioides populations (Fig. 1). In the case of S. armiger, the approximately 10% sequence divergence between individuals collected from Trinidad and Panama corresponds closely to that observed between C. scorpioides individuals from the same two localities (Fig. 1, Table 2). Semeiochernes armiger populations from Central and South America, however, differ by only one amino acid (AA position 28). In the outgroup taxon P. elongatus, nucleotide variation between individuals from colonies separated by less than 11 km ranged from 2.9 to 7.7% (Table 2). However, all differences are silent. This extreme mtDNA divergence

TABLE 1 The Average Proportion of Nucleotide Changes in Each Substitution Class across MP Phylogenies To From

A

C

G

T

A C G T

— 0.009 0.054 0.170

0.033 — 0.017 0.256

0.139 0.027 — 0.020

0.159 0.088 0.029 —

211

Cordylochernes mtDNA PHYLOGENY

TABLE 2 Pairwise Nucleotide Sequence Divergence Estimated Using Kimura’s Two-Parameter Model (Upper Half) and the Absolute Number of Nucleotide Differences between Sequences (Lower Half) (1) (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13)

P. e. C25B(1) P. e. Gamb(1) P. e. Cam(2) P. e. Cerro(2) P. s.(1) S. a. CZ(2) S. a. TRIN(1) L. c.(2) C. s. CZ(6) C. s. CHIR(1) C. s. CHIR(1) C. s. FG(2) C. s. TRIN(2)

20 23 17 69 74 77 70 74 75 75 73 77

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

(11)

(12)

(13)

0.0672

0.0774 0.0294

0.0565 0.0431 0.0463

0.2857 0.2997 0.2729 0.2928

0.3294 0.3318 0.3139 0.3268 0.3028

0.3150 0.3018 0.2747 0.3123 0.2927 0.1006

0.2957 0.3058 0.3007 0.3186 0.3042 0.2471 0.2609

0.3139 0.3190 0.3190 0.3320 0.3072 0.2698 0.2561 0.2287

0.3168 0.2966 0.2966 0.3141 0.3417 0.2997 0.2756 0.2500 0.0817

0.3168 0.3066 0.3066 0.3244 0.3472 0.3099 0.2856 0.2500 0.0817 0.0064

0.3077 0.3230 0.3077 0.3256 0.2916 0.2876 0.2967 0.2309 0.1111 0.1299 0.1299

0.3294 0.3452 0.3399 0.3587 0.2825 0.2937 0.2977 0.2502 0.1077 0.1379 0.1302 0.0260

9 13 71 71 77 72 75 71 73 76 80

14 66 66 74 71 75 71 73 73 79

70 73 76 74 77 74 76 76 82

68 71 70 71 77 78 69 67

29 63 62 66 68 70 70

61 65 71 73 68 69

58 63 63 58 62

24 24 32 31

2 37 39

37 37

37

Note. P. e., Paratemnoides elongatus; P. s., Parachernes setosus; S. a., Semeiochernes armiger; L. c., Lustrochernes consocius; C. s., Cordylochernes scorpioides. Location abbreviations are: Gamb, Gamboa; Cerro, Cerro Pelado; and Cam, Camino de Cruces; all other site abbreviations are as defined in the methods. The sample size for each sequence is in parentheses following the taxon label.

does not appear to correlate with geographic distance or location relative to the Rı´o Chagres. Phylogeny Reconstruction A branch and bound search in which all nucleotide sites were equally weighted and in which Paratemnoides was specified as the outgroup, recovered three MP trees (282 steps, CI 5 0.681, HI 5 0.319, RI 5 0.751), which differed only in the relative positions of P. elongatus colonies within the Paratemnoides clade (Fig. 2). Successive reweighting recovered a single tree identical to one of the three unweighted MP trees. Very similar topologies, which differed only in the relative positions of colonies within the Paratemnoides clade, were recovered when transversions were weighted 3 times transitions, when relative weights of 4-8-1 were

applied to 1st, 2nd, and 3rd codon positions, and when both codon position weights and transition bias were jointly incorporated (Fig. 2). Although neither character position nor substitution weighting altered tree topology, these analyses often resulted in higher bootstrap values and support and tree consistency indices, particularly for deeper branches (Fig. 2). Parsimony analysis of inferred amino acid sequences recovered a single tree (41 steps, CI 5 0.951, HI 5 0.049, RI 5 0.96), congruent with the MP mtDNA tree, but lacking resolution within the Cordylochernes clade. Topology of trees recovered using maximum likelihood both with and without character weighting were well supported and congruent with MP trees, although the relative positions of Paratemnoides remained weakly supported (data not shown).

FIG. 1. Pseudoscorpion collection localities in Central and South America. Numbers within arrows connecting sites denote percentage nucleotide sequence divergence and number of amino acid (AA) differences between geographic populations of Cordylochernes scorpioides. For site details and species collected, see text.

212

WILCOX ET AL.

FIG. 2. Maximum parsimony (MP) trees recovered from unweighted and weighted parsimony. The tree shown for unweighted parsimony is one of three MP trees recovered and is the same tree recovered when characters were reweighted using Farris’s (1969) successive reweighting approach. Weighted parsimony with 4-8-1 codon position weighting and 1:3 TS:TV substitution weighting only recovered the one MP tree depicted, though several rearrangements within the Paratemnoides are only one step longer. Numerals above the internal branches indicate the proportion of 500 bootstrap replicate samples recovering each clade. Numerals below the internal branches are Bremer’s support indices. For weighted parsimony, support values for 4-8-1 codon position weighting are to the left of each slash and for combined codon position and 1:3 TS:TV weighting are to the right of each slash. Bremer’s support indices for weighted parsimony are rescaled by the ratio of weighted to unweighted treelengths for comparison with unweighted values (Bremer, 1994). NS, supported by fewer than 50% of bootstrap replicates.

The same branching order was recovered among ingroup taxa in all phylogenetic reconstructions, regardless of the weighting scheme employed. Cordylochernes scorpioides populations form a strongly supported, monophyletic group, sister to L. consocius, in which the Panamanian and South American populations clearly represent sister taxa (Fig. 2). Semeiochernes armiger sequences from Panama and Trinidad form a monophyletic group, basal to the Cordylochernes 1 Lustrochernes clade, while P. setosus represents the basal lineage of the chernetid genera investigated in this study.

respectively. Although the overall discriminant functions were statistically significant for both sexes (Wilk’s Lambda, P , 0.003), only FG females were clearly separated from any other group (Fig. 3). Additionally, FG females are only distinguishable along the first canonical axis, which is strongly correlated with overall body size. Finally, cross classification errors were relatively large, ranging from 10% for FG females to a maximum of 70%. DISCUSSION

Morphological Analysis Because C. scorpioides is highly sexually dimorphic (Zeh and Zeh, 1994), discriminant analysis of morphometric data was carried out separately on each sex. Discriminant analysis recovered two canonical axes for both males and females (likelihood ratio, P , 0.05) which account for 86 and 95% of the overall variance,

Generic-Level Phylogenetic Inferences With 110 genera and 603 described species (Harvey, 1990), the Chernetidae is the most diverse of the 22 pseudoscorpion families. Phylogenetic relationships among genera within this family are poorly understood and controversial (Muchmore, 1972, 1974, personal

Cordylochernes mtDNA PHYLOGENY

213

Divergence within the C. scorpioides Clade

FIG. 3. Multivariate analysis of morphometric data. Top panels depict the mean canonical score (695% bootstrap CI) for morphological measurements of C. scorpioides males and females on the first two discriminant axes. (s), CHIR; (k), CZ; (n), FG; (q), TRIN.

communication). Traditional classification, based on type of body setae, sculpturing of the cuticle, and number of setae in the cheliceral flagellum, divides the family into two subfamilies: the Lamprochernetinae and the Chernetinae (Beier, 1932a,b, 1933). The characters used to erect this classification have been criticized as erroneous by Legg (1971, 1987) who has argued for the systematic importance of male genitalic and female spermathecal characters. Determining whether or not such traits yield robust phylogenies consistent with Beier’s original classification must await a full-scale, morphological reexamination of the genera (Muchmore, 1972, 1974, personal communication). The results of our very preliminary application of molecular methods for generic-level phylogenetic reconstruction of the Chernetidae suggest that mtDNA COI sequence data may ‘‘. . . give some valuable clues to what the higher categories (subfamilies, tribes, etc.) will include’’ (W. B. Muchmore, personal communication). Both our weighted nucleotide and amino acid sequence analyses provide strong support for a monophyletic Lustrochernes1Cordylochernes clade, a result consistent with the traditional classification of these two genera in the subfamily Lamprochernetinae (Beier, 1932a,b, 1933; Muchmore, 1976, 1984). By contrast, the two genera in the Chernetinae, Parachernes and Semeiochernes, did not cluster together as the sister taxon to the Lamprochernetinae, suggesting that this subfamily may comprise a paraphyletic group. Phylogenetic analysis of the translated sequence also reveals a pattern in the types of amino acid substitutions occurring at different levels within the phylogenetic hierarchy. Whereas substitutions within the Lustrochernes1Cordylochernes clade involve only changes within side-chain families of amino acids (e.g., the replacement of one nonpolar side-chain amino acid for another, such as leucine for valine), substitution of one amino acid type for a different side-chain type occurred between more distantly related taxa.

Our mtDNA sequence analysis revealed extreme differentiation within the C. scorpioides species complex with a maximum divergence of 13.8% between individuals from western Panama and Trinidad. Assuming an arthropod pairwise mtDNA sequence divergence rate of 2.3% per million years (Brower, 1994b), this corresponds to a divergence time of approximately 6 million years. In fact, the level of divergence within Cordylochernes greatly exceeds that exhibited by many congeneric species of birds in North America (see Seutin et al., 1993) and provides evidence of a major split between C. scorpioides lineages from South and Central America. Further support for an ancient divergence between the two clades is provided by differences not only at synonymous sites but also in amino acid sequences. Two amino acid differences distinguish the Canal Zone and French Guiana/Trinidad populations and three differences exist between the Chiriqui and the South American populations. The South and Central American lineages differ markedly in the geographic scale over which nucleotide divergence has occurred. Sequences from individuals collected 1200 km apart in Trinidad and French Guiana differ by only 2.6%, compared to a difference of 8.2% between individuals collected in western and central Panama, a distance of less than 400 km. In addition, while there are no amino acid substitutions within the South American clade, a first position transversion between the Chiriqui and Canal Zone sequences results in an amino acid change (valine to leucine) between the two Panamanian populations. Previous research involving interpopulation crosses has shown C. scorpioides individuals from French Guiana and from the Canal Zone to be completely reproductively incompatible at the postzygotic stage (Zeh and Zeh, 1994). Reproductive divergence between these two cryptic species was found to be accompanied by little morphological but strong molecular differences in terms of both allozymes (Zeh and Zeh, 1994) and hypervariable nuclear DNA (Zeh et al., 1992a). The high level of mtDNA COI sequence divergence reported here between the C. scorpioides populations from French Guiana and the Canal Zone provides additional support for the conclusion that ‘‘high levels of variability within each population . . . have obscured a speciation event in which genetic divergence and postzygotic incompatibility have clearly outpaced the evolution of prezygotic isolation’’ (Zeh and Zeh, 1994). Our mtDNA sequence data have now revealed a similar relationship involving molecular divergence and morphological stasis in association with breakdown in reproductive compatibility between two further pairs of C. scorpioides populations. The outcome of matings between individuals from Trinidad and the Canal Zone is very similar to that observed in French Guiana/Canal Zone

214

WILCOX ET AL.

crosses: females produce external brood sacs containing embryos which invariably abort early in development (J. A. Zeh, unpublished data). This complete reproductive incompatibility is accompanied by two amino acid differences and 11% nucleotide sequence divergence. The second case involves what appears to be incipient speciation between the C. scorpioides populations from Chiriqui and the Canal Zone. Morphologically similar but genetically distinct in their mtDNA sequences, individuals from these two populations exhibit partial reproductive incompatibility at the postzygotic stage (J. A. Zeh, unpublished data). Here, partial incompatibility is associated with only one amino acid difference but 8.2% nucleotide sequence divergence. Other studies of terrestrial arthropods have noted as much as 25% divergence between cytochrome oxidase sequences from congeneric insect species, with generally less than 1% divergence between individuals of the same species (reviewed in Simon et al., 1994). Beckenbach et al. (1993) examined 688 bp of the COII gene, and found less than 0.3% intraspecific and a maximum of 11% interspecific divergence among 12 species in the Drosophila obscura species group. A complete absence of intraspecific divergence has been documented in 532 bp of the mtDNA COI gene (a region completely overlapping the one sequenced by us) in two genera of temperate bumblebees, with interspecific divergence ranging from 5 to 11% in Bombus and 6 to 9% in Psithyrus (Pedersen, 1996). In a study of 16 morphologically distinguishable lepidopteran species in the genus Greya, sequences spanning 765 bp of the COI and COII genes (again, a region which encompasses the one examined in this study) exhibited 1 to 5.7% divergence within species and a maximum of 16% divergence between species (Brown et al., 1994). Maximum intraspecific divergence was found within widespread species, while species with restricted ranges were less than 1% divergent. Similarly, Brower (1994a) examined 945 bp of COI and COII from Heliconius butterflies and found pairwise divergences of less than 1% between conspecific races and up to 13% between the ingroup and closely related outgroup genera. In a more detailed phylogenetic study of the same gene region, Brower (1994b, 1996) uncovered less than 5% sequence divergence between conspecific races within two species of Heliconius sampled across South America. Thus, the 13.6% COI sequence divergence reported here between morphologically very similar South and Central American C. scorpioides populations corresponds to divergence levels found between morphologically distinct, congeneric species in other terrestrial arthropods, supporting our conclusion of extreme genetic differentiation within C. scorpioides. Extensive ‘‘intraspecific’’ mtDNA divergence, suggestive of the existence of cryptic species, has also been detected between disjunct populations of the wood-

feeding roach Cryptocercus punctulatus, from eastern and western United States. These populations exhibit approximately 11% divergence in their 12s rRNA sequences, and this high level of sequence divergence corresponds with complete reproductive isolation and variation in chromosome number between sampled populations (Kambhampati et al., 1996). Similarly, in a study of populations of the widespread darkling beetle, Heteger politus, collected from several of the Canary islands, Juan et al. (1996) found interpopulation COI divergences of 5 to 12%. Although information on the reproductive compatibility of the populations is not available, gene flow between island populations is likely to be restricted since H. politus is flightless. These comparative data thus indicate that the levels of mtDNA sequence divergence reported here are consistent with there being barriers to gene flow between geographic populations of C. scorpioides. With the exception of the previously discussed Heliconius butterflies, which presumably exhibit limited differentiation due to recency of divergence (Brower, 1996), little is known of the geographic scale over which mtDNA divergence occurs in terrestrial invertebrates in the Neotropics. Recent studies of vertebrates have revealed pairwise mtDNA divergences of 1 to 11% between populations of several putative neotropical species, including birds (Seutin et al., 1993, 1994), akodontine rodents (Smith and Patton, 1991), boid snakes (Henderson and Hedges, 1995), and frogs (L. Weigt, personal communication). Of particular interest are data from a study of the tungara frog, Physalemus pustulosus, sampled from Central and South America. Levels of COI sequence divergence between Panamanian populations in Chiriqui and the Canal Zone range from 4 to 9%, representing the largest phylogeographic break observed between adjacent populations of this frog (L. Weigt, personal communication). These results correspond well with the 8.2% divergence found in Cordylochernes from the same areas, suggesting that a barrier to gene flow may occur between western and central Panama. Interestingly, however, divergence between the Canal Zone and Trinidad is lower for P. pustulosus (L. Weigt, personal communication) than for Cordylochernes sampled from the same areas. In contrast to the extensive nucleotide divergence between geographic populations of C. scorpioides, within-population sequence diversity appears to be low. Sequence data from several individuals from both the Canal Zone and French Guiana detected only a single sequence within each population. This lack of diversity could be accounted for by a recent historical event such as a severe population bottleneck. However, the extremely high level of heterozygosity detected at two minisatellite loci in these populations ($95% at both loci in the two populations; see Zeh et al., 1994) would seem to be inconsistent with a recent bottleneck. The

215

Cordylochernes mtDNA PHYLOGENY

pattern for Cordylochernes contrasts sharply with that found in P. elongatus, which displayed a remarkably high level of intrapopulation mtDNA sequence divergence (up to 7.7%) between individuals collected from different colonies only a few kilometers apart within the Canal Zone. The causes of this marked difference remain to be established, but social behavior seems likely to be an important contributory factor. Paratemnoides elongatus exhibits the highest level of sociality known in pseudoscorpions, involving cooperative predation and the construction of communal molting nests (Brach, 1978; Zeh and Zeh, 1990). Brach (1978) hypothesized that the evolution of cooperative behavior in Paratemnoides was linked to interrelatedness (inbreeding) among colony members and field data on colony composition and sex ratio are consistent with this hypothesis (Zeh and Zeh, 1990). A pattern in which females generally remain in their natal colonies while males occasionally disperse could result in the high level of local sequence diversity detected by our mtDNA analysis. The mtDNA sequence data for S. armiger appear to provide an interesting parallel to the C. scorpioides pattern of divergence. This pseudoscorpion was originally described as three species (S. armiger, S. extraordinarius, and S. militaris) on the basis of sexually dimorphic traits of the male pedipalpal chela (Beier, 1933, 1954). Recent rearing experiments, however, have demonstrated that intrapopulation variability encompasses the full range of Semeiochernes ‘‘interspecific’’ chelal morphology and that separate species status cannot be established on the basis of male morphology (Zeh and Zeh, 1992b). At the molecular level, however, mitochondrial COI sequence divergence of almost 10% between Semeiochernes individuals from Panama and Trinidad may well indicate that the pattern exhibited by C. scorpioides is a general one and that many so-called species of terrestrial arthropods in the Neotropics may actually represent complexes of cryptic species. ACKNOWLEDGMENTS We thank V. Mahnert and W. B. Muchmore for identifying the pseudoscorpions and for comments on the manuscript, and L. Meffert for assistance with morphometric analysis. We also thank R. DeSalle and two anonymous reviewers for helpful comments which improved the manuscript. Particular thanks go to R. Hartman (Chiriqui), to C. Starr and R. Hernandez (Trinidad), and to G. Tavakilian, J.-M., and O. Baloup (French Guiana) for logistical support. We are grateful to El Instituto Nacional de Recursos Naturales Renovables (INRENARE) and the Forestry Division of the Republic of Trinidad and Tobago for permission to collect in Panama and Trinidad, respectively. This research was supported by grants from the Environmental Institute of Houston and the University of Houston’s L.G.I.A. program to D.W.Z. and by grants to D.W.Z. and J.A.Z. from the National Geographic Society (Grant 5333-94) and the Smithsonian Tropical Research Institute.

REFERENCES Beckenbach, A. T., Wei, Y. W., and Liu, H. (1993). Relationships in the Drosophila obscura species group inferred from mitochondrial cytochrome oxidase II sequences. Mol. Biol. Evol. 10: 619–634. Beier, M. (1932a). Pseudoscorpionidea II. Subord. Cheliferinea. Das Tierreich. 58: 1–294. Beier, M. (1932b). Zur Kenntnis der Lamprochernetinae (Pseudoscorp.). Zool. Anz. 97: 258–267. Beier, M. (1933). Revision der Chernetidae (Pseudoscorp.). Zool. Jahrb. 64: 509–548. Beier, M. (1948). Phoresie und Phagophilie bei Pseudoscorpionen. ¨ esterr. Zool. Z. 1: 441–497. O Beier, M. (1954). Einer Pseudoscorpioniden: Ausbeute aus Venezuela. Memorie Mus. civ. Stor. nat. Verona 4: 131–142. Brach, V. (1978). Social behavior in the pseudoscorpion Paratemnus elongatus (Banks) (Pseudoscorpionida: Atemnidae). Insectes Sociaux 25: 3–11. Bremer, K. (1994). Branch support and tree stability. Cladistics 10: 295–304. Brower, A. V. Z. (1994a) Phylogeny of Heliconius butterflies inferred from mitochondrial DNA sequences (Lepidoptera: Nymphalidae). Mol. Phylogenet. Evol. 3: 159–174. Brower, A. V. Z. (1994b). Rapid morphological radiation and convergence among races of the butterfly Heliconius erato inferred from patterns of mitochondrial DNA evolution. Proc. Natl. Acad. Sci. USA 91: 6491–6495. Brower, A. V. Z. (1996). Parallel race formation and the evolution of mimicry in Heliconius butterflies: A phylogenetic hypothesis from mitochondrial DNA sequences. Evolution 50: 195–221. Brown, J. M., Pellmyr, O., Thompson, J. N., and Harrison, R. G. (1994). Phylogeny of Greya (Lepidoptera: Prodoxidae), based on nucleotide sequence variation in mitochondrial cytochrome oxidase I and II: Congruence with morphological data. Mol. Biol. Evol. 11: 128–141. Clary, D. O., and Wolstenholme, D. R. (1985). The mitochondrial DNA molecule of Drosophila yakuba: Nucleotide sequence, gene organization, and genetic code. J. Mol. Evol. 22: 252–271. Dillon, W. R., and Goldstein, M. (1986). ‘‘Multivariate Analysis Methods and Applications,’’ Wiley, New York. Farris, J. S. (1969). A successive approximations approach to character weighting. Syst. Zool. 34: 21–34. Felsenstein, J. (1981). Evolutionary trees from DNA sequences: A maximum likelihood approach. J. Mol. Evol. 17: 368–376. Felsenstein, J. (1985). Confidence limits on phylogenies: An approach using the bootstrap. Evolution 39: 783–791. Felsenstein, J. (1995). ‘‘PHYLIP (Phylogeny Inference Package), Version 3.5c,’’ Department of Genetics, Univ. Washington, Seattle. Felsenstein, J., and Churchill, G. A. (1996). A hidden Markov model approach to variation among sites in rate of evolution. Mol. Biol. Evol. 13: 93–104. Harvey, M. S. (1990). Catalogue of the Pseudoscorpionida. Manchester Univ. Press, Manchester. Harvey, M. S. (1992). The phylogeny and classification of the Pseudoscorpionida (Chelicerata: Arachnida). Invertebr. Taxon. 6: 1373– 1435. Henderson, R. W., and Hedges, S. B. (1995). Origin of West Indian populations of the geographically widespread boa Corallus enydris inferred from mitochondrial DNA sequences. Mol. Phylogenet. Evol. 4: 88–92. Hitachi (1994). ‘‘MacDNASIS (Macintosh DNA and Protein Analysis System Version 3.2),’’ San Bruno, CA.

216

WILCOX ET AL.

Juan, C., Oromi, P., and Hewitt, G. M. (1996). Phylogeny of the genus Heteger (Tenebrionidae, Coleoptera) and its colonization of the Canary Islands deduced from cytochrome oxidase I mitochondrial DNA sequences. Heredity 76: 392–403. Kambhampati, S., Luykx, P., and Nalepa, C. A. (1996). Evidence for sibling species in Cryptocercus punctulatus, the wood roach, from variation in mitochondrial DNA and karyotype. Heredity 76: 485–496. Legg, G. (1971). The Comparative and Functional Morphology of the Genitalia of the British Pseudoscorpiones. Ph.D. Thesis, Univ. Manchester. Legg, G. (1987). Proposed taxonomic changes to the British pseudoscorpion fauna (Arachnida). Bull. Br. Arachnol. Soc. 7: 179–182. Maddison, W. P., and Maddison, D. R. (1992). ‘‘MacClade: Analysis of Phylogeny and Character Evolution, Version 3,’’ Sinauer Associates, Inc., Sunderland, MA. Muchmore, W. B. (1972). Observations on the classification of some European chernetid pseudoscorpions. Bull. Brit. Arachnol. Soc. 2: 112– 115. Muchmore, W. B. (1974). Clarification of the genera Hesperochernes and Dinocheirus (Pseudoscorpioinida, Chernetidae). J. Arachnol. 2: 25–36. Muchmore, W. B. (1976). Pseudoscorpions from Florida and the Caribbean area. 5. Americhernes, a new genus based upon Chelifer oblongus Say (Chernetidae). Florida Entomol. 59: 151–163. Muchmore, W. B. (1984). The pseudoscorpions described by R. V. Chamberlin (Pseudoscorpionida, Olpiidae and Chernetidae). J. Arachnol. 11: 353–362. Pedersen, B. V. (1996). The phylogenetic analysis of cuckoo bumblebees (Psithyrus, Lepeletier) and bumblebees (Bombus, Latreille) inferred from sequences of the mitochondrial gene cytochrome oxidase I. Mol. Phylogenet. Evol. 5: 289–297. Proctor, H. (1993). Mating behavior resolves trichotomy for cheliferoid pseudoscorpions (Pseudoscorpioinida, Cheliferoidea). J. Arachnol. 21: 156–158. Sanger, F., Nicklen, S., and Coulson, A. R. (1977). DNA sequencing with chain-terminating inhibitors. Proc. Natl. Acad. Sci. USA 74: 5463–5467. SAS Institute Inc. (1989b). ‘‘SAS/IML Software: Usage and Reference, Version 6, First Edition,’’ SAS Institute Inc., Cary, NC. SAS Institute Inc. (1989a). ‘‘SAS/STAT User’s Guide, Version 6, Fourth Edition, Volume 1,’’ SAS Institute Inc., Cary, NC.

Seutin, G., Brawn, J., Ricklefs, R., and Bermingham, E. (1993). Genetic divergence among populations of a tropical passerine, the streaked saltator (Saltator albicollis). Auk 110: 117–126. Seutin, G., Klein, N., Ricklefs, R., and Bermingham, E. (1994). Historical biogeography of the bananaquit (Coereba flaveola) in the Caribbean region: A mitochondrial DNA assessment. Evolution 48: 1041–1061. Simon, C., Frati, F., Beckenbach, A., Crespi, B., Liu, H., and Flook, P. (1994). Evolution, weighting and phylogenetic utility of mitochondrial gene sequences and a compilation of conserved polymerase chain reaction primers. Ann. Entomol. Soc. Am. 87: 651–701. Smith, M. F., and Patton, J. L. (1991). Variation in mitochondrial cytochrome b sequence in natural populations of South American akodontine rodents (Muridae: Sigmodontinae). Mol. Biol. Evol. 8: 85–103. Swofford, D. (1994). ‘‘PAUP: Phylogenetic Analysis Using Parsimony, Version 3.1.1,’’ Formerly distributed by Illinois Natural History Survey, Champaign, IL. Zeh, D. W., and Zeh, J. A. (1992a). Dispersal-generated sexual selection in a beetle-riding pseudoscorpion. Behav. Ecol. Sociobiol. 30: 135–142. Zeh, D. W., and Zeh, J. A. (1994). When morphology misleads: interpopulation uniformity in sexual selection masks genetic divergence in harlequin beetle-riding pseudoscorpion populations. Evolution 48: 1168–1182. Zeh, D. W., Zeh, J. A., Coffroth, M.-A., and Bermingham, E. (1992a). Population-specific DNA fingerprints in a neotropical pseudoscorpion (Cordylochernes scorpioides). Heredity 69: 201–208. Zeh, D. W., Zeh, J. A., and Tavakilian, G. (1992b). Sexual selection and sexual dimorphism in the harlequin beetle Acrocinus longimanus. Biotropica 24: 86–96. Zeh, D. W., Zeh, J. A., and May, C. A. (1994). Charomid cloning vectors meet the pedipalpal chelae: single-locus minisatellite DNA probes for paternity assignment in the harlequin beetle-riding pseudoscorpion. Mol. Ecol. 3: 517–522. Zeh, J. A., and Zeh, D. W. (1990). Cooperative foraging for large prey by Paratemnus elongatus (Pseudoscorpionida, Atemnidae). J. Arachnol. 18: 307–311. Zeh, J. A., and Zeh, D. W. (1992b). Are sexually-selected traits reliable species characters? Implications of intra-brood variability in Semeiochernes armiger (Balzan) (Pseudoscorpionida: Chernetidae). Bull. Br. Arachnol. Soc. 9: 61–64.