mitochondrial dna in the bark weevils: phylogeny and evolution in

Mitochondrial DNA in the Bark Weevils: Phylogeny and Evolution in the Pissodes strobi Species Group (Coleoptera: Curculionidae)

Thomas M. Boyce, * Michael E. Zwick,t_ and Charles F. Aquadrot_ *Division of Entomology, C.S.I.R.O., and =/Section of Genetics and Development, Cornell University

The genus Pissodes (Coleoptera: Curculionidae ) contains four species informally grouped into the P. strobi species group: P. strobi, P. nemorensis, P. terminalis, and P. schwarzi. These species have been the focus of extensive investigations into their behavior and ecology in relation to reproductive isolation and evolution. We examined restriction-site polymorphism and divergence of mitochondrial DNA in these species and in an outgroup, P. afinis. Both diversity and divergence are high relative to that seen in other insects studied. Nucleotide diversity is 0.1%~ 1.3% within species, and net divergence among species is 2.0% 16.0%. Phylogenetic relationships of mitochondrial DNA haplotypes directly contradict several hypotheses of species relationship from previous studies of morphology, allozyme, and cytogenetic data indicating that mitochondrial and nuclear genes have evolved in distinctly different manners. Hybridization among species and high rates of sequence change may have contributed to these contra- dictions.

Introduction

Members of the Pissodes strobi species group of bark weevils (Coleoptera: Curculionidae) have provided a fertile arena for studies of fundamental issues in evolutionary biology. Many aspects of evolution of phyto- phagous insects have been examined in the group, including the role of pheromones and behavior in maintaining reproductive isolation (Phillips and Lanier 1983, 1986; Phillips et al. 1987). Extensive studies of the cytogenetics of the group have provided unique insights into the possible roles that chromosomal rear- rangements may have in speciation (Smith 1956, 1962; Smith and Takenouchi 1962). Moreover, some members of the group are significant pests of important North American trees.

Bark weevils utilize the cambial layers of either living or dead species of pine, spruce, and fir for feeding, oviposition, and larval development (Baker 1972, pp. 328-331; Furniss and Carolin 1977, pp. 330-334). In North America, Pissodes species are distributed throughout the pine forests of the northern portion of the continent and the extensive pine lands of the south- eastern states (figs. 1 and 2). The ecology and life cycles of various North American Pissodes species have been

Key words: Pissodes, bark weevil, mitochondrial DNA, phylogeny, hybridization.

Address for correspondence and reprints: Thomas M. Boyce, Di- vision of Entomology, C.S.I.R.O., G.P.O. Box 1700, Canberra, ACT 260 1, Australia.

Mol. Bid. Evol. 11(2):183-194. 1994. 0 1994 by The University of Chicago. All rights reserved. 0737-4038/94/l 102-2-003$02.00

described in detail by several authors (see Wallace and Sullivan 1985; also see references in Silver 1968; Baker 1972; Furniss and Carolin 1977 ).

Four species are currently recognized in the P. strobi species group: P. strobi (Peck), P. nemorensis Germar, P. terminalis Hopping, and P. schwarzi Hopkins (Smith and Takenouchi 1969). All share relatively similar, if not identical, adult morphologies, and they may be crossed in the laboratory to yield viable hybrids (Smith 1962; Godwin and ODell 1967; Phillips 1984). The most economically important pests in the genus are P. strobi and P. terminalis, which attack the growing terminals (the very topmost, dominant shoot) of young, living trees, killing 1 year’s, and sometimes 2 years’, growth. Trees attacked by these pests are rendered unsuitable for timber sale, since the trunk of the tree becomes de- formed after a lateral shoot takes over primary growth. The other members of the genus oviposit in dead or moribund branches, trunks, or roots but may be pests by acting as vectors for plant pathogens such as western gall rust ( Bella 1985 ) and root black stain ( Witcosky et al. 1986).

Despite the efforts mentioned above, the taxanomic status and phylogeny of the P. strobi species group of bark weevils have been uncertain for many years. This confusion has arisen in part because Hopkins ( 19 11) described many species solely on the basis of host-tree species and geographic locality. While many of these designations have been synonymized on the basis of chromosome number, morphology, allozyme identity, and pheromone use (table 1)) the species status of several

183 Downloaded from https://academic.oup.com/mbe/article-abstract/11/2/183/1113049by gueston 12 February 2018

184 Boyce et al.

_) P. strobi

1 P. nemorensis

FIG. 1 .-Distribution and collection localities of Pissodes strobi and P. nemorensis. Locality labels correspond to those in table 2.

currently recognized taxa is still questionable, and their phylogeny is problematic. Analysis of sequence variation and phylogeny of mitochondrial DNA (mtDNA) has recently provided valuable insights into organismal evolution for a wide range of taxa (Wilson et al. 1985; Avise et al. 1987; Moritz et al. 1987; Avise 1989). In particular, restriction-enzyme-site polymorphisms of mtDNA may be used as genetic markers/of species status and phylogenetic relationship. We have applied this approach to the members of the P. strobi species group to better un- derstand their taxonomy and phylogenetic relationships. Our results show that, for these species, mtDNA phylogenies contradict phylogenetic expectations drawn from allozymes, cytogenetics, morphology, and ecology. Moreover, not only is the evolution of mitochondrial genome size and size polymorphism in the A+T-rich region exceptional (Boyce et al. 1989), but sequence polymorphism and divergence within the coding region is also unusuallv high.

Material and Methods

Collections of Pissodes species were made from localities described in table 2 and indicated in figures 1 and 2. Live adult weevils were shaken from the growing terminals of trees into 4-liter plastic jars and then were handpicked into vials. Adults aggregating at freshly cut pine logs were similarly collected by hand. These adults were most often found on the underside of logs, in relatively moist areas. Larvae and pupae were excavated from their respective host material directly from logs in the field or after the infested growing terminals had been cut from the tree.

Species identification followed the conventions dictated by current knowledge of the genus; that is, weevils aggregating in the spring to terminals of white pine were considered to be P. strobi, and those aggregating to logs in the same location and season were considered to be P. nemorensis (Phillips 1984; Phillips and Lanier 1986). Weevils emerging from terminals of lodgepole pine were considered to be P. terminalis (Drouin et al. 1963; Furniss and Carolin 1977, p. 334). Weevils emerging from tree species other than lodgepole pine were considered to be P. strobi (Smith and Sugden 1969). Weevils emerging from root collars of lodgepole pine from British Columbia were provisionally designated as “P. schwarzi” (Smith and Sugden 1969). Sub- sequent examinations of chromosome number and morphology of an individual male from this sample confirmed this identification (G. N. Lanier, personal communication). An individual of P. afinis Randall was also analyzed to provide an outgroup.

Total genomic DNA was prepared from individual weevils as described by Boyce et al. ( 1989) and was digested individually with 10 restriction endonucleases by following manufacturers’ recommendations; these en-

P. terminalis

Y v

i FIG. 2.-Distribution and collection localities of Pissodes ter-

minalis and P. schwarzi. Locality labels correspond to those in table 2.

Downloaded from https://academic.oup.com/mbe/article-abstract/11/2/183/1113049by gueston 12 February 2018

mtDNA Phylogeny in the Bark Weevils 185

Table 1 Host, Habitat, and Phenologies of Species in the Pissodes strobi Species Group

Species and Synonyms Diploid Number Host Species Habitat

strobi (Peck): 34 = engelrnanni Hopkins 34 = sitchensis Hopkins 34

nemorensis Germar: 30 (30-34) = approximatus Hopkins 30-34 = canadensis Hopkins 30-34 = deodarae Hopkins . 30-34

terminalis Hopping 28f, 29mb

schwarzi Hopkins . 38

= Yosemite Hopkins 38

SOURCE.-Smith and Sugden (1969); Phillips (1984).

Pinus, picea Terminals Pinus, Picea Terminals Picea sitchensis Terminals Pinus, Picea, Cedrus Slash, root collars Pinus, Picea Slash, root collars Pin us, Picea Slash, root collars Cedrus deodarae Slash and terminals Pinus contorta Terminals Pinus banksiana Terminals Pinus, Picea Slash, root collars Larix occidentalis Slash, root collars Pinus contorta Slash, root collars

a Since P. approximattrs has been synonymized with P. nemorensis, it is now considered polymorphic for its diploid number. However, only populations of the former P. approximattrs and its synonym, P. canadmsis, are known to be polymorphic in this way.

b Females (f) of P. terminalis are homozygous for a fusion product of two acrocentrics for which males (m) are heterozygous (Smith and Takenouchi 1962).

zymes were Hindlll, Bglll, Xbal, Pstl, Mspl, Haelll, Pvull, EcoRV, Seal, and Hhal. Digested DNAs were electrophoresed, transferred to nylon filters, and probed as by Boyce et al. ( 1989).

The mtDNA of Pissodes often reveals very complex restriction-fragment patterns because of its large size and the presence of repeated elements (Boyce et al. 1989). Variable numbers of repeated sequences of different size classes are observed in digestions with enzymes that are known to cleave within the repeated sequences (Hirtdlll, Xbal, Hhal, and Haelll). Restriction-site changes lo- calized to the A+T-rich region that appeared to be linked to changes in molecule size were not used in this analysis, since this type of rapid mutation between essentially in- distinguishable states has generally proved unreliable as a character for phylogenetic analysis (e.g., see Densmore et al. 1985; Harrison et al. 1987).

For the purposes of this study, only sites that could be unambiguously mapped to the apparent coding region of the Pissodes mtDNA molecule were scored ( fig. 3 ) . Mapping and scoring of these sites were facilitated by sequentially probing filters with clones pPSBX.2, pPSBX.28, and pPSH.6.1 (fig. 3), obtained from P. strobi as described by Boyce et al. ( 1989). These clones themselves were carefully mapped in order to accurately resolve the site locations within them. The presence or absence of restriction sites in the mtDNA of individuals successfully scored for all enzymes makes up the haplotype of each individual.

Phylogenetic analysis of restriction-site differences as discrete characters utilized PAUP 3.1.1 (Swofford 1993) and PHYLIP 3.5~ (Felsenstein 1993). Details of

each analysis are presented with their results. The restriction-site maps and their frequencies in each species were also used to estimate haplotype diversity (h; Nei and Tajima 198 1 ), nucleotide diversity (n;; Nei and Li 1979)) and average percent sequence divergence among species and several populations (see below). Variances of divergence estimates and net divergences corrected for within-population or within-species polymorphism were estimated using the method of Nei and Jin ( 1989) with the neighbor-joining method of dendrogram con- struction ( Saitou and Nei 1987 ) . Divergence estimates were also used to construct dendrograms of mtDNA among species by using the neighbor-joining method ( Saitou and Nei 1987 ) .

Results Size Variation and Restriction-Site Maps

All of the 249 individuals examined were hetero- plasmic for several different size classes of mtDNA dif- fering by 2-7 kb in total length. The size variation re- sulted from varying numbers of tandem-repeated fragments within the A+T-rich region as described by Boyce et al. ( 1989). Although the presence of many size classes within individuals was apparent, it was not possible to reliably score individuals for the frequency of particular size classes because of difficulties in resolving the different size classes in digestions with enzymes that do not cleave in the repeated sequences. The fragments produced by these digestions occur in an 18-30-kb size range and do not resolve well under the conditions of agarose gel electrophoresis used for effective resolution of smaller restriction-enzyme digestion products.


186 Boyce et al.

Table 2 Collection Data for Populations of Pissodes Sampled for mtDNA Variation

P&odes Species and Population Locality; Date, Collector” Host Species and Site

Stage Collected

Voucher Specimens’

P. strobi: BC . . .._ Terrace, B.C.; 7/87, LM WOR . Waldport, Lincoln Co., (coastal) Oreg.; 8/87, TMB AOR Alsea, Benton Co., (inland), Oreg.; 8/87, TMB MT Flathead National Forest, Mont.; 7/87, JED co . . . . . . Bellvue, Larimer Co., Colo.; 9/82, TWP WI Black River Falls, Jackson Co., Wise.; 8/8 1, TWP NH . Durham, Strafford Co., N.H.; 8/81, TWP Ds-NY Danby, Tomkins Co., N.Y.; 5/87, TMB Df-NY . Danby, Tomkins Co., N.Y.; 8/87, TMB PA Fort Indiantown Gap, Pa.; 8/8 1, TWP VA . Northeast of Charlottesville, Louisa Co., Va.;

8181 TWP BNC

P. nemorensis: TX GA SC BNC _. _. DNC Df-NY TNY

P. terminalis: CA . . . . . . MT . . . . . . BC . . . . . .

P. sch warzi: BC

Brevard, Transylvania Co., N.C.; 8/87, SGB

Ratcliff, Houston Co., Tex.; 12/88, WWU Sharon, Warren Co., Ga.; 10/88, TMB Jamestown, Berkeley Co., S.C.; 10/88, TMB Brevard, Transylvania Co., N.C.; 5/87, SGB Durham, Durham Co., N.C.; 10/88, TMB Danby, Tompkins Co., N.Y.; 8/87, TMB Tully, N.Y.; 5/87, TMB

Hobart Mills, Nevada Co., Calif.; 7/87, TMB Flathead National Forest, Mont.; 7/86, JED North Ellis Creek, near Penticon, B.C.; 5/88, LM

North Ellis Creek, near Penticon, B.C.; 5/88 and 6189, LM

P. afinis . . . Tully, N.Y.; 5/87, TMB

Picea sitchensis terminals Picea sitchensis terminals Picea sitchensis terminals Picea engelmannii terminals Picea pungens terminals Pinus banksiana terminals Pinus strobus terminals Pinus strobus terminals Pinus strobus terminals Pinus strobus terminals Pinus strobus terminals

Pinus strobus terminals

Pinus taeda logs Pinus taeda logs Pinus taeda logs Pinus strobus logs Pinus taeda logs Pinus strobus logs Pinus resinosa logs

Pinus contorta terminals Pinus contorta terminals Pinus contorta terminals

Pinus contorta root collars

Pinus strobus

Larval Larval Larval Larval Larval Reared adult Reared adult Adult Larval Reared adult Reared adult

Larval

Adult Adult Adult Adult Adult Larval Adult

Larval Larval Reared adult

Reared adult

Adult

1192 A 1192 B 1192 C 1192 D 1192 E 1192 F 1192 G 1192 H 1192 I 1192 J 1192 K

1192 L

1192 M 1192 N 1192 0 1192 P 1192 Q 1192 R 1192 S

1192T 1192 U 1192 V

1192 W

. . .

’ LM = Loraine McLauchlin; TMB = William W. Upton.

b Deposited in the Cornell University

= Thomas M. Boyce; JED = Jerald E. Dewey; TWP = Thomas W. Phillips; SGB = Stephen G. Boyce; and WWL

Insect Collection, Ithaca, N.Y., under lot number 1192.

For the restriction sites that cleave within the apparent coding region of the Pissodes mitochondrial genome, the 10 restriction endonucleases revealed a total of 80 mapped cleavage sites (fig. 3) and a total of 7 1 different haplotypes found across all individuals and species surveyed. The haplotypes for the different species were aligned on the basis of restriction sites whose ho- mology was deduced from detailed maps of the regions probed with the cloned regions from P. strobi. No instances of heteroplasmy for restriction-site polymorphisms were detected. Haplotypes listed by individual and restriction sites are available on request from the authors and have been deposited with the voucher specimens (table 2 ) .

Phylogeny of mtDNA Haplotypes

As a first step in analysis, we attempted to construct a minimal mutation network of all haplotypes

(fig. 4). The size and complexity of the data set pre- vented searching via techniques that would guarantee a minimal distance tree for all haplotypes at once. An initial search using the heuristic search algorithm of PAUP, branch swapping using tree bisection-recon- nection, and retaining as many trees as possible produced 2,800 equally parsimonious trees of 148 mutational steps. Figure 4 displays the strict concensus tree for these 2,800 trees. .The resulting consensus relationships support the division of haplotypes into several distinct groups, separated by a substantial number of mutational steps (fig. 4).

The resulting groupings of haplotypes do not always reflect species boundaries ( see below). For convenience, groups will therefore be referred to as haplotype “assemblages.” Since they are usually congruent with species designations, they will be designated in most cases by the nominal species from which they were collected. For



+T - nch r.%jKm Drosophila

pFSBX.2 m

pPSBX.28 w mpPSH.6.1

cloned .sf+ynents of Pisoda mtDNA

strobi

\;4 HI3

nemorensis A4 A5 M5 s2 Pz AI0 MB M9 Ml0 E6 E7 X8 88X9

M6 HI0

Texas B2 H10 E7

terminalis

schwani 83 F-2 AI0 M9 El

P. afinis

-12 -11 -10 3 -8 -7 d -5 -4 -3 -2 -1 0 t1 +2 +3 t4 +5 4.6 +7 +8

Map Position .

FIG. 3.-Region of the P&odes mitochondrial genome surveyed for restriction-site polymorphisms in Pissodes species. The genetic map of Drosophila melanogaster has been aligned to the restriction map of Pissodes as described by Boyce et al. ( 1989). Note that the scale has been centered at the BarrzHI site in P. strobi. Haplotypes are represented from each major assemblage and P. a&is (see fig. 4 and text). Sites listed above each line are polymorphic, while those below the line are fixed within the species or assemblage. Sites are labeled as in the Appendix.

example, the “terminalis” assemblage comprises haplotype terA and its relatives, and the “strobi” assemblage comprises strA and its relatives (fig. 4).

To further evaluate the strength of these groupings and the relationships among the assemblages, two further approaches were used. First, the DOLLOP parsimony program of PHYLIP was applied to all haplotypes, using the polymorphism option and

bootstrap resampling (Felsenstein 1985) for a total of 50 replicates. The Do110 polymorphism algorithm of parsimony analysis can be particularly appropriate to restriction-site data (Felsenstein 1993 ). While the algorithm allows only a single origin of a site, it also allows the retention of polymorphism at a site while minimizing the total number of polymorphic characters. Results of this analysis (not shown) were con-


188 Boyce et al.

FIG. 4.-Strict-consensus phylogenetic network of P&odes haplotypes. Haplotypes are named as in the Appendix. Major assemblages are grouped together and indicated by the assemblage name. Numbers of mutational changes are indicated along each branch.

gruent with those of the simple parsimony analysis using PAUP.

Second, a two-tiered bootstrap resampling scheme was implemented using PAUP. The first level of bootstrap resampling occurred at the level of the assemblages. One haplotype was randomly selected to represent each assemblage, creating a reduced data set of six haplotypes. The P. afinis haplotype was included as an outgroup, for a total of seven haplotypes. These then were analyzed

using the bootstrap facility of PAUP with a branch-and- bound search for the shortest tree for each replicate. One hundred resampling replicates were performed at each level, for a total of 10,000 replicates. The overall majority-rule consensus tree was then calculated (fig. 5), and it again supported the initial phylogeny (fig. 4).

The majority of ambiguity in phylogeny recon- struction lies in the inability to resolve relationships within the major assemblages. With the exception of the placement of P. schwarzi and the Texan population of P. nemorensis, relationships among the major assemblages are robustly resolved. The haplotypes of the P. nemorensis and P. terminalis assemblages share more recent common ancestry than they share with any other group, and the P. strobi assemblage forms the sister group to the other members of the P. strobi species group

(fig. 5). In two instances, the haplotype assemblages defined

by this analysis are not concordant with the species identifications of the individuals in which they are found. First, the individuals identified as P. strobi and collected from Brevard, N.C. (BNC) and Charlottesville, Va. (VA) (haplotype nemS ) , and four individuals (nemP, nemQ, nemR, and nemS) collected from the spring population of Danby, N.Y. ( Ds-NY ), contain mitochondrial haplotypes that are much more closely related to those usually found in P. nemorensis than they are to any other haplotype (figs. 4 and 5 ) . These haplotypes form a group that unambiguously lies within the P. nemorensis lineage (figs. 4 and 5). Forty-three mutational steps separate this assemblage and the nearest other haplotype found within P. strobi. Yet, the group is only two mutation steps away from its nearest relatives in P. nemorensis.

A second instance of unexpected phylogenetic af- finity is seentin a group of haplotypes from individuals of P. nemorensis collected in Texas (TX). These haplotypes are separated from all other groups: 15 steps to the nearest other P. nemorensis haplotype and 13 steps to the nearest P. schwarzi haplotype (fig. 4). Yet, the relationship of these haplotypes is as intermediates between the other major assemblages, and they are not resolvable as a separate, monophyletic assemblage

(fig. 5). While recombination among mitochondrial ge-

nomes is not considered a likely event (Moritz et al. 1987), such ambiguity in the placement of the Texan P. nemorensis haplotypes could be the result of recombination among more distantly related haplotypes. To examine this possibility, phylogenetic analysis of subsets of restriction-site changes among the representative haplotypes was performed. These subsets were constructed by a “sliding window” technique of choosing 20 adjacent sites every 10 sites along the restriction-site



map. Analysis of these subsets by PAUP revealed phylogenetic relationships of haplotypes congruent with the complete analysis, although relationships were often less well resolved (data not shown). If recombination had been apparent, phylogenies of some regions would have consistently been in conflict with those for other regions, reflecting their possibly disparate ancestries (e.g., see Satta and Takahata 1990). Additionally, visual inspec- tion of the restriction-site maps of these haplotypes did not reveal any obvious indications of recombination events. Although not a monophyletic assemblage, the Texas mtDNA assemblage does appear to represent a distinct grouping and is treated as such in the remaining analysis.

Sequence Divergence among Species and Assemblages

The estimated mtDNA sequence divergences among the populations and species defined by the phylogenetic assemblages of haplotypes are presented in table 3. Values of overall divergence are presented, as well as values corrected for within-species polymorphism, because the species (or lineages as defined in fig. 4) differ substantially in the degrees of within-species diversity. Correction for polymorphisms will decrease divergence estimates for all comparisons of the most polymorphic species (e.g., the P. strobi lineage) with all others, but not among those other lineages. Estimates of per-nucleotide sequence divergence among assemblages, corrected for within-lineage polymorphism, are 2%- 16% and are uniformly larger than diversity estimates within these assemblages (table 3). A dendrogram of genetic distances among the groups, based on adjusted divergence estimates, was constructed using the neighbor- joining method (Saitou and Nei 1987). The relationships indicated by the dendrogram are congruent with those of the phylogenetic analysis, given the uncertainty of the relationships of P. schwarzi and the P. nemorensis from Texas (fig. 5A and B).

Discussion

mtDNA variation in Pissodes is unusual, not only because of the genome’s large size and’extensive heteroplasmy for size variants (Boyce et al. 1989)) but also because levels of sequence divergence in mtDNA among these species are unusually high for what are otherwise closely related taxa. Moreover, levels of mtDNA sequence variation within lineages of Pissodes are high (particularly in P. strobi) and are comparable to values typically found in vertebrates (Moritz et al. 1987 ) . This result is important in that it contrasts with the lower levels found, to date, in several other insect species, particularly in the Lepidoptera (Brower and Boyce 199 1; Pashley and Ke 1992 ) .

Mitochondrial and Species Phylogenies

It is surprising that the phylogenetic relationships constructed for the major mitochondrial haplotype assemblages conflict in several instances with the evolutionary relationships that may be inferred from morphological similarity and other types of genetic data. Overall morphological similarity would suggest that the sibling species P. strobi and P. nemorensis are most closely related within the group (Hopkins 19 11; Phillips 1984). In contrast, the phylogeny of mitochondrial haplotypes indicates that P. strobi is the sister lineage to all other members of the species group (figs. 4 and 5A and B). Morphologically, P. terminalis and P. schwarzi are distinguished from P. nemorensis and P. strobi by col- oration, distribution, and host use (Hopping 1920). However, both P. terminalis and P. schwarzi share a more recent common mitochondrial ancestry with P. nemorensis than with P. strobi. These conflicts arise re- gardless of whether the mitochondrial network is rooted (a) at the midpoint of the greatest patristic distance or (b) by using the haplotype of P. afinis as an outgroup.

The mitochondrial phylogeny also conflicts with hypotheses of relationship suggested by cytogenetic analyses. Smith and Takenouchi ( 1962) proposed that P. terminalis is a hybrid species formed by the crossing of female P. strobi with male P. schwarzi. Their hypothesis could account for the unique sex-specific di- morphism in the C autosome observed in P. terminalis. A pair of acrocentric C chromosomes were to have been derived from P. strobi with the metacentric C acquired from P. schwarzi, the only other species in which it is currently seen. Laboratory crosses of the two species appeared to confirm the hypothesis (Smith and Taken- ouchi 1962). This proposed hybridization also could be interpreted to explain in a vague sense the acquisition of terminal oviposition in P. terminalis, i.e., that it was acquired as a trait from P. strobi. However, Smith and Takenouchi’s hypothesis of the hybrid origin of P. terminalis is not supported by the mitochondrial haplotype phylogeny. If a hybridization event were responsible for the origin of P. terminalis, the mtDNA data suggest that it would most likely have involved a female parent most closely related to P. nemorensis, not to P. strobi.

The study of 11 allozyme loci by Phillips ( 1984) suggested that, like morphological similarity, allozyme frequencies were most similar for P. strobi and P. nemorensis ( fig. 5C). Indeed, no loci fixed for alternate alleles were detected among the two species. Overall, frequency differences between the two are significant, but the estimated genetic distance between them is small (Nei’s D = 0.02; Phillips 1984). Pissodes terminalis is more differentiated in allozyme frequencies (Nei’s D = 0.16 between P. terminalis and P. strobi). Pissodes


(A) MtDNA - Bootstrapped parsimony

P. affinis

P. strobi

P. sch warzi

P. nemorensis (Texas)

P. terminalis

P. strobi (VA, BNC, and Ds-NY in part )

P. nemorensis

(B) MtDNA - Neighbor joining

21.34 P. affinis

5.44 P. strobi

3.95 P, schwa&

4.28 P. nemorensis (Texas)

0.92 0.32 w P. terminalis

0.39 3.24

’ -0.20 - P. strobi

r 1.90 . (VA, BNC, and Ds-NY in part )

1 o 2. P. nemorensis .


mtDNA Phylogeny in the Bark Weevils 19 1

(C) Allozyme genetic distance

0.137

c p* affinis

0.069

0.012 P. strobi

0.055 P. nemorensis 0.0 12 (Eastern U.S.)

1 0.082 P. terminalis

FIG. 5.-Relationships of Pissodes species estimated from (A) bootstrap analysis of mitochondrial haplotypes, (B) mitochondrial genetic distance, and (C) allozyme genetic distance. The majority-rule consensus tree of panel A displays the frequency at which particular groupings occur in replications of the procedure in the text. In panel C the dendrogram of genetic distance based on allozyme variation is taken from Phillips ( 1984) and was constructed using UPGMA. Not all assemblages or species represented in the mitochondrial trees were examined for allozyme variation.

afinis, which is not considered a member of the P. strobi species group, is differentiated even more (Nei’s D = 0.30).

A similar conflict between mtDNA and nuclear gene phylogenies has been found in Hawaiian species of Drosophila ( DeSalle and Giddings 1986)) although the levels of divergence among the Drosophila mtDNA lineages are substantially less than that found in Pissodes. There is no immediate reason to doubt the allozyme data available for Pissodes species. Indeed, the data are congruent with morphological similarities. However, the

addition of a single fixed allozyme difference among any of the species would greatly affect the estimated genetic distances among them -and potentially would also affect the relationships inferred from those data. A more extensive survey of allozyme variation or nuclear gene DNA sequences is warranted.

Pissodes nemorensis in Texas

The unexpected discovery of a well-differentiated, (although not monophyletic) group of haplotypes in individuals of P. nemorensis collected in Texas indicates

Table 3 Estimated Sequence Variation within (R; along the Diagonal) and Differentiation between (d,,) Pissodes Species or Assemblages Identified by Major Assemblages of Mitochondrial Haplotypes

P. strobi (n = 114)

P. strobi (VA, BNC, Ds-NY,

in part) (n = 20)

P. nemorensis (n = 61)

P. nemorensis (Texas) (n = 9)

P. terminalis (n = 33)

P. sch warzi (n = 13)

P. strobi . . . . . . . . . . 0.0225 (0.0028) 0.1694 (0.0296) 0.1595 (0.0377) 0.1118 (0.0350) 0.1119 (0.03 10) 0.1472 (0.0349) P. strobi (VA, BNC, Ds-NY,

in part) . . . . . . . . . . 0.1770 (0.0297) 0.0031 (0.0019) 0.0004 (0.0002) 0.0546 (0.0154) 0.0224 (0.0088) 0.0877 (0.02 14) P. nemorensis . . . . . . . 0.1696 (0.0378) 0.0042 (0.00 17) 0.0045 (O.OOZ9) 0.0522 (0.0193) 0.0208 (0.0083) 0.088 1 (0.019 1) P. nemorensis (Texas) . . 0.1197 (0.0351) 0.0579 (0.0155) 0.0042 (0.0017) 0.0034 (0.0023) 0.0515 (0.0196) 0.0492 (0.0144) p. terminalis . . . . . 0.1167 (0.03 1 I) 0.0245 (0.0089) 0.0234 (0.0086) 0.0538 (0.0154) 0.0011 (0.0007) 0.0924 (0.0195) P. schwarzi . . . . . . . 0.1557 (0.0350) 0.09 15 (0.02 16) 0.0925 (0.0 193) 0.053 1 (0.0 153) 0.0952 (0.0228) 0.0045 (0.0021) P. afinis . . . . . . . . . 0.2678 (0.082 1) 0.3393 (0.0924) 0.3208 (0.0916) 0.2748 (0.0762) 0.2752 (0.0775) 0.285 1 (0.0765)

NOTE.-Divergence between populations corrected for within-species/assemblage variation is above the diagonal; and uncorrected divergence is below the diagonal. Estimates and their standard errors (in parentheses) are based on the method of Nei and Jin (1989), with a clustering of haplotypes by the neighbor-joining method.


192 Boyce et al.

that P. nemorensis as currently recognized may be com- posed of at least two different populations. It is interesting that populations of P. nemorensis have been described infesting the terminals of Pinus taeda as well as Cedrus species in eastern Texas (Ollieu 197 1). Moreover, although P. deodare was originally described by Hopkins ( 19 11)) from collections made in Georgia, it was distinguished from P. nemorensis in part because the specimens were reared from the living branches and terminals of C. deodara. Further investigation of the ecology and genetics of Texan populations of P. nemorensis and their possible relationship to P. deodarae is needed. In contrast, the synonymy of P. approximatus with P. nemorensis is supported by the nearly continuous distribution of one or a number of closely related mtDNA haplotypes throughout the rest of the sampled range of P. nemorensis. Our data do not address the synonymy of P. canadensis with P. nemorensis.

Hybridization and Mitochondrial Introgression

The presence of mitochondrial haplotypes most closely related to those of P. nemorensis in individuals collected and identified as P. strobi appears to be best explained by introgression of mtDNA across the species boundary. Alternative explanations, such as the retention of ancestral mtDNA polymorphisms or polymorphism of host use in P. nemorensis, are not supported by other data. Studies have indicated that eastern populations of P. nemorensis sympatric with P. strobi successfully oviposit and develop in pine terminals at a very low frequency, if at all, even when forced to do so (Phillips and Lanier 1983), indicating specificity of host resource in P. nemorensis. Likewise, the probability that the P. nemorensis mitochondrial haplotypes have simply been retained in P. strobi since the divergence of the two species is low, in light of the extensive amount ( 16%) of sequence divergence between the two mitochondrial lineages and the low amount of divergence between the haplotypes in question and those of P. nemorensis.

The expected time to a common ancestor for mtDNA is N generations, where N is the long-term effective population size (Kingman 1982; Tajima 1983; Avise et al. 1988). If it is assumed that the population ancestral to P. strobi and P. nemorensis was a single, panmictic unit, that bark weevils typically complete one generation per year, and that the rate of mtDNA evolution in bark weevils is comparable to that estimated for some Drosophila species (0.1% divergence/ bp/Myr ; Solignac et al. 1986; Satta et al. 1987), then the time to common ancestry for the two most closely related haplotypes of P. nemorensis and P. strobi is - 80 million generations (or 80 Myr). As- suming a faster rate typical of vertebrate mtDNA (2% divergence /Myr ; Brown et al. 1982) reduces this es-

timate to -6 million generations, or a long- ,term fen male effective population size of 26 million. Curren female effective population sizes estimated from tht mean divergence of mtDNA haplotypes in single pop, ulations of P. strobi and P. nemorensis are estimatec to be < 1 million (Boyce 1990)) making the probabilit! of retaining such divergence in mtDNA haplotype: over long periods apparently rather small.

Hybridization between the two species is a man likely explanation for the patterns observed. Hybridiza, tion is attainable in laboratory conditions, and natura hybridization between the two species has been proposec by Smith ( 1962). Indeed, Smith argued from cytogenetic data that P. strobi had hybridized with a 2N = 30 species possibly P. nemorensis, to form a hybrid swarm corre, sponding to the then-recognized P. approximatus ant P. canadensis (Smith 1962). The putatively hybrid in dividuals in our study were collected in areas broadly sympatric with current P. nemorensis, within the distri bution of the former P. approximatus, and thus woulc be sympatric with hybrids as suggested by Smith. More, over, the two larval samples of P. strobi fixed for P. nem orensis mtDNA were from the southern end of the rangt of P. strobi, where P. nemorensis is the more abundan species.

Aspects of the two species’ behaviors may con tribute to the distribution of mtDNA haplotypes thal we have observed. Adults of both species actively feet on live cambial tissues of branches, and presumably on terminals, of various pine species during spring ac tivity (Godwin et al. 1982). Thus, mating among P strobi and P. nemorensis adults would be possible al feeding sites if behavioral isolation between the twc species is not strong. In tests by Phillips and Laniei ( 1983)) mixed pairs of female P. strobi and male P nemorensis from Tully, N.Y. ( = P. approximatus) pro. duced progeny, from white pine terminals, at levels in distinguishable from those of pure P. strobi pairs. The reciprocal pairings produced progeny at half these lev. els. If female P. nemorensis mated by P. strobi male: at terminals then oviposited in those terminals, then they would potentially contribute P. nemorensis haplotypes to the following generation of weevils emerging from terminals.

The putatively hybrid weevils were either (a) adults collected while aggregating to white pine terminals during spring oviposition by P. strobi ( Ds-NY), (b) excavated as larvae from an infested white pine terminal (BNC), or (c) reared to the adult stage, from infested white pine terminals (VA). While it is possible that the spring adults collected in New York were P. nemorensh simply feeding and not mating or ovipositing, currently accepted knowledge of the habitat and oviposition pref-



erences of P. strobi would indicate that the weevils collected as larvae from white pine leaders were P. strobi (Phillips 1984; Wallace and Sullivan 1985; Phillips and Lanier 1986). Morphological characters for the sepa- ration of P. nemorensis and P. strobi larvae as described by Boving ( 1929) were examined in the larvae collected from white pine terminals in Brevard, N.C. Only one character could be consistently observed: the relative tanning of the peritreme of the spiracles (darkly tanned in P. strobi). All of these larvae displayed the P. strobi type of peritreme tanning. Thus, P. nemorensis mtDNA was found in individuals displaying characters seemingly typical of P. strobi.

In a survey of allozyme variation among P. approximatus, P. nemorensis, and P. strobi, no fixed differences and few statistically significant frequency shifts were found; overall genetic identities among the three species averaged 0.96 (Phillips 1984). These data taken together suggest that hybridization has led to mitochondrial haplotypes from P. nemorensis introgressing into P. strobi and that nuclear gene flow may either occur currently between P. strobi and P. nemorensis or have occurred at some point in the recent past. Occasional hybridization allowing bidirectional exchange of nuclear genes and unidirectional exchange of mtDNA would lead to similar allozyme frequencies, shared chromosome polymorphisms, and possibly similar morphologies. If hybridization and introgression of nuclear genes account for the apparent similarities of P. strobi and P. nemorensis, then it has been continuing for perhaps at least 6 Myr or as long as 80 Myr.

Clearly, currently recognized taxonomic relationships based on morphological or chromosomal similarities alone need to be reevaluated. Moreover, the nature of mtDNA sequence change in insects as a whole may also need reevaluation in light of the patterns found in Pissodes. The P. strobi species group displays a uniquely high level of mtDNA divergence and polymorphism, which contrasts markedly with low levels of allozyme and morphological divergence, suggesting that the rate of change in the nuclear genome of Pissodes has been substantially different than that in the mitochondrial genome.

The extensive polymorphism for restriction sites and size in mtDNA also suggests that evolution in Pis- sodes mtDNA may be accelerated compared with that in other insects examined to date, especially in comparison with Lepidoptera (Brower and Boyce 199 1; Pashley and Ke 1992). The surprising mtDNA phylogeny, which indicates that P. strobi is the sister lineage to all others in the group, could be a reflection of extremely high rates of sequence evolution, which might distort or bias an mtDNA phylogeny based on restriction-site

polymorphism alone. The high level of polymorphism and the resulting low level of phylogenetic resolution of haplotypes within P. strobi are suggestive of such an acceleration.

Acknowledgments

We thank the many people who helped in the collection of weevils, including S. G. and H. A. Boyce, J. Dewey, T. Koerber, J. McLean, R. Clower, E. L. Cuebas- Incle, L. A. Martinez, L. McLauchlin, and W. W. Upton. We especially thank T. W. Phillips and G. N. Lanier for their advice and help, W. Noon provided technical and computing assistance. We thank G. C. Eickwort, R. G. Harrison, W. W. Anderson, and J. G. Oakeshott for their advice and comments. This work was supported by a grant from the U.S. Department of Agriculture.

LITERATURE CITED

AVISE, J. C. 1989. Gene trees and organismal histories: a phylogenetic approach to population biology. Evolution 43: 1192-1208.

AVISE, J. C., J. ARNOLD, R. M. BALL, E. BERMINGHAM, T. LAMB, J. E. NEIGEL, C. A. REEB, and N. C. SAUNDERS. 1987. Intraspecific phylogeography: the mitochondrial DNA bridge between population genetics and systematics. Annu. Rev. Ecol. Syst. l&489-522.

AVISE, J. C., R. M. BALL, and J. ARNOLD. 1988. Current versus historical population sizes in vertebrate species with high gene flow: a comparison based on mitochondrial DNA lineages and inbreeding theory for neutral mutations. Mol. Biol. Evol. 5:33 l-344.

BAKER, W. L. 1972. Insects of eastern forests. U.S. Department of Agriculture, Forest Service, Washington, D.C.

BELLA, I. E. 1985. Western gall rust (Endocronartium hark- nessi) and insect leader damage in relation to tree size in young lodgepole pine (Pinus contorta var. latifolia) in Al- berta Canada. Can. J. Forestry Res. 15: 1008- 10 10.

B~VING, A. G. 1929. Taxonomic characters for the identification of the mature larvae of Pissodes strobi Peck and Pissodes approximatus Hopkins (Fam. Curculionidae ) .

Proc. Entomol. Sot. Wash. 31:182-186. BOYCE, T. M. 1990. Molecular evolutionary genetics of the

bark weevils: speciation and mitochondrial DNA evolution. Ph.D. dissertation, Cornell University, Ithaca, N.Y.

BOYCE, T. M., M. E. ZWICK, and C. F. AQUADRO. 1989. Mi- tochondrial DNA in the bark weevils: size, structure and heteroplasmy. Genetics 123:825-836.

BROWER, A. V. Z., and T. M. BOYCE. 199 1. Mitochondrial DNA variation in monarch butterflies. Evolution 45: 128 l- 1286.

BROWN, W. M., E. M. PRAGER, A. WANG, and A. C. WILSON. 1982. Mitochondrial DNA sequences of primates: tempo and mode of evolution. J. Mol. Evol. l&225-239.

DENSMORE, L. D., J. W. WRIGHT, and W. M. BROWN. 1985. Length variation and heteroplasmy are frequent in mitochondrial DNA from parthenogenetic and bisexual lizards (genus Cnemidophorus). Genetics 110:689-707.


194 Boyce et al.

DESALLE, R., and L. V. GIDDINGS. 1986. Discordance of nuclear and mitochondria DNA phylogenies in Hawaiian Drosophila. Proc. Natl. Acad. Sci. USA 83:6902-6906.

DROUIN, J. A., C. R. SULLIVAN, and S. G. SMITH. 1963. Oc- currence of P&odes terminalis Hopp. (Coleoptera: Cur- culionidae) in Canada: life history, behaviour, and cytogenetic identification. Can. Entomol. 95:70-76.

FELSENSTEIN, J. 1985. Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39:783-79 1.

- 1993. PHYLIP (phylogeny inference package) version . 3.5~. University of Washington, Seattle.

FURNISS, R. L., and V. M. CAROLIN. 1977. Western forest insects. U.S. Department of Agriculture, Forest Service, Washington, D.C.

GODWIN, P. A., and T. M. ODELL. 1967. Experimental hybridization of Pissodes strobi and P. approximatus (Co- leoptera: Curculionidae). Ann. Entomol. Sot. Am. 60:55- 58.

GODWIN, P. A., H. T. VALENTINE, and T. M. ODELL. 1982. Identification of Pissodes strobi, P. approximatus and P. nemorensis (Coleoptera: Curculionidae) using descriminant analysis. Ann. Entomol. Sot. Am. 75:599-604.

HARRISON, R. G., D. M. RAND, and W. C. WHEELER. 1987. Mitochondrial DNA variation across a narrow hybrid zone. Mol. Biol. Evol. 4: 144- 158.

HOPKINS, A. D. 19 11. Contributions toward a monograph of the bark-weevils of the genus Pissodes. Entomol. Bull. 20 (Tech. Ser.). U.S. Department of Agriculture, Washington, D.C.

HOPPING, R. 1920. A new species of the genus Pissodes (Co- leoptera). Can. Entomol. 52: 132- 134.

KINGMAN, J. F. C. 1982. On the genealogy of large populations. J. Appl. Prob. 19A:27-43.

MORITZ, C., T. E. DOWLING, and W. M. BROWN. 1987. Evo- lution of animal mitochondrial DNA: relevance for population biology and systematics. Annu. Rev. Ecol. Syst. 18: 269-292.

NEI, M., and L. JIN . 1989. Variances of the average numbers of nucleotide substitutions within and between populations. Mol. Biol. Evol. 6:290-300.

NEI, M., and W.-H. LI. 1979. Mathematical model for studying genetic variation in terms of restriction endonucleases. Proc. Natl. Acad. Sci. USA 76:5269-5273.

NEI, M., and F. TAJIMA. 198 1. DNA polymorphism detectable by restriction endonucleases. Genetics 97: 145- 163.

OLLIEU, M. M. 197 1. Damage to southern pines in Texas by Pissodes nemorensis. J. Econ. Entomol. 64: 1456- 1459.

PASHLEY, D. P., and L. D. KE. 1992. Sequence evolution in mitochondrial ribosomal and ND-l genes in Lepidoptera: implications for phylogenetic analyses. Mol. Biol. Evol. 9: 1061-1075.

PHILLIPS, T. W. 1984. Ecology and systematics of Pissodes sibling species (Coleoptera: Curculionidae) . Ph.D. dissertation, State University of New York, Syracuse.

PHILLIPS, T. W., and G. N. LANIER. 1983. Biosystematics of Pissodes Germar (Coleoptera: Curculionidae ) : feeding preference and breeding site specificity of P. strobi and P. approximatus. Can. Entomol. 115: 1627- 1636.

- 1986. Interspecific activity of semiochemicals among . sibling species of Pissodes (Coleoptera: Curculionidae ) . J. Chem. Ecol. 12:1587-1601.

PHILLIPS, T. W., S. A. TEALE, and G. N. LANIER. 1987. Bio- systematics of Pissodes (Germar): seasonality, morphology and synonymy of P. approximatus and P. nemorensis. Can. Entomol. 119:465-480.

SAITOU, N., and M. NEI. 1987. The neighbor-joining method: a new method for reconstructing phylogenetic trees. Mol. Biol. Evol. 4:406-425.

SATTA, Y., H. ISHIWA, and S. I. CHIGUSA. 1987. Analysis of nucleotide substitutions of mitochondrial DNAs in Dro- sophila melanogaster and its sibling species. Mol. Biol. Evol. 4:638-650.

SATTA, Y., and N. TAKAHATA. 1990. Evolution of Drosophila mitochondrial DNA and the history of the melanogaster subgroup. Proc. Natl. Acad. Sci. USA 87:9558-9562.

SILVER, G. T. 1968. Studies on the sitka spruce weevil, Pissodes sitchensis, in British Columbia. Can. Entomol. 10:93- 110.

SMITH, S. G. 1956. Chromosomal polymorphism in a bark weevil. Nature 177:386.

- 1962. Cytogenetic pathways in beetle speciation. Can. . Entomol. 94:94 l-955.

SMITH, S. G., and B. A. SUGDEN. 1969. Host trees and breeding sites of native North American Pissodes bark weevils, with a note on synonymy. Ann. Entomol. Sot. Am. 62: 146- 148.

SMITH, S. G., and Y. TAKENOUCHI. 1962. Unique incompatibility system in a hybrid species. Science 138:36-37.

-. 1969. Chromosomal polymorphisms in Pissodes weevils: further on incompatibility of P. terminalis. Can. J. Genet. Cytol. 11:76 l-782.

SOLIGNAC, M., M. MONNEROT, and J. MOUNOLOU. 1986. Mitochondrial DNA evolution in the melanogaster species subgroup of Drosophila. J. Mol. Evol. 23:3 l-40.

SWOFFORD, D. 1993. PAUP: phylogenetic analysis using parsimony, version 3.1.1. Smithsonian Institution, Washing- ton, D.C.

TAJIMA, F. 1983. Evolutionary relationship of DNA sequences in finite populations. Genetics 105:437-460.

WALLACE, D. R., and C. R. SULLIVAN. 1985. The white pine weevil, Pissodes strobi (Coleoptera: Curculionidae ) : a re- view emphasizing behavior and development in relation to physical factors. Proc. Entomol. Sot. Ont. Suppl. 116:39- 62.

WILSON, A. C., R. L. CANN, S. M. CARR, M. GEORGE, U. B. GYLLENSTEN, K. M. HELM-BYCHOWSKI, R. G. HIGUCHI, S. R. PALUMBI, E. M. PRAGER, R. D. SAGE, and M. STONE- KING. 1985. Mitochondrial DNA and two perspectives on evolutionary genetics. Biol. J. Linnean Sot. 26:375-400.

WITCOSKY, J. J., T. D. SCHOWALTER, and E. M. HANSEN. 1986. The influence of time of precommercial thinning on the colonization of Douglas-fir (Pseudotsuga menziesii) by three species of root-colonizing insects. Can. J. Forestry Res. 16:745-749.

DANIEL HARTL, reviewing editor

Received August 11, 1992

Accepted October 20, 1993


mitochondrial dna in the bark weevils: phylogeny and evolution in

Documents