Lobster Phyl Ogen y
Lobster Phyl Ogen y
Lobster Phyl Ogen y
Abstract. Phylogenetic relationships among all described species and four subspecies (total of 21 taxa) of the spiny lobster genus Panulirus White, 1847, were examined with nucleotide sequence data from portions of two mitochondrial genes, large-subunit ribosomal RNA (16S) and cytochrome oxidase subunit I (COI). Multiple sequence alignments were subjected to maximum-parsimony, neighbour-joining, and maximum-likelihood analysis with Jasus edwardsii as the outgroup. Two major lineages within Panulirus were recovered by all three methods for both the 16S and COI alignments analysed separately and for the combined alignment. The first lineage included all species of Panulirus classified as Groups I and II by previous morphologically based definitions. The second included all species classified as Groups III and IV. Relationships within major lineages were not well resolved; the molecular phylogeny did not support separation of Group I from Group II or of Group III from Group IV. The degree of sequence divergence between different pairs of species was higher in pairwise comparisons between species in Group I/II (16S: 2.819.4%; COI: 12.431.8%) than in those between species in Group III/IV (16S: 5.313.2%; COI: 12.619.6%). This pattern suggests that the Group I/II lineage may represent an earlier radiation of species within Panulirus.
Introduction The spiny lobster genus Panulirus White, 1847, has long been of interest to evolutionary biologists because of its high level of species diversity, its wide geographic distribution, and the importance of many species to commercial fisheries. Among genera in the family Palinuridae, Panulirus has been the most successful in terms of species diversity; 19 species have been described to date, three of which are divided into seven recognized subspecies (Holthuis 1991; George 1997; Sarver et al. 1998). The key to this successful radiation is thought to be the invasion by species of Panulirus of shallow-water, tropical environments, which permits the occupation of varied habitats not accessible to congeners (George and Main 1967; George 1997). Because of morphological, ecological, and behavioural diversity, both within and among species of Panulirus, the phylogenetic history of this group has not been well understood. The group has undergone a number of taxonomic divisions based on interspecific diversity in morphological traits. George and Main (1967) proposed the first of these, in which they divided Panulirus into four informal groups (I IV) without giving these groups specific taxonomic
CSIRO 2001
rankings. These four groups were distinguished by comparative morphology of the exopods associated with the second and third maxillipeds of adults (Table 1). Species of Panulirus differ in the degree of development of these exopods. Within the genus Panulirus, species with the most reduced exopods (Group IV) are regarded as the most recently evolved, whereas the presence of exopods with flagella on either the second or third maxillipeds (or both) is considered the ancestral condition (Groups I and II; George and Main 1967). All other palinurid genera have fully developed exopods. More recently McWilliam (1995) proposed a revised taxonomic scheme based on interspecific diversity in morphological traits associated with phyllosoma larval and puerulus stages. This revised scheme is largely concordant with George and Mains (1967) classification, except in the placements of P. interruptus and P. guttatus. Comparative morphological data indicated two basic phyllosoma and puerulus forms. Members of the more recent radiation of tropical species (Groups III and IV) share a more specialized form of phyllosoma and puerulus than those of the earlier, subtropical radiation (Groups I and II). The phyllosoma and the puerulus of P . interruptus are of intermediate form, and the phyllosoma of P. guttatus shows
10.1071/MF01070 1323-1650/01/081037
1038
Table 1. Arrangement of species of Panulirus based on morphological and geographic differences adapted from George and Main (1967) and McWilliam (1995) Exopod Condition 3rd maxilliped I Present with flagellum 2nd maxilliped Present with flagellum Indo-West Pacific Panulirus sp. East Pacific West Atlantic East Atlantic
II III IV
cygnus japonicus longipes marginatus pascuensis penicillatus polyphagus homarus ornatus stimpsoni versicolor
interruptus
argus
echinatus regius
the more derived state of the specialized, spatulate antennae characteristic of species in Groups III and IV. The biogeographic history within the genus has been debated, mostly with respect to the timing of divergence among the four species groups (George and Main 1967; Baisre and Ruiz de Quevedo 1982; Pollock 1992, 1993; McWilliam 1995; Sekiguchi 1995; George 1997), but it is generally agreed that Groups I and II represent an early radiation of Panulirus species from an Indo-Pacific ancestor followed by a later radiation, Groups III and IV, also from an Indo-Pacific ancestor (Table 1). This historical interpretation is based on the retention of pleisiomorphic morphological traits by species in Groups I and II (fully developed exopods of the second and third maxillopeds in adults; George and Main 1967) and less specialized forms of phyllosoma and puerulus stages (McWilliam 1995). First attempts at dating the initial radiation of Groups I and II suggested an association with Pleistocene glaciation periods (George and Main 1967; Pollock 1992, 1993; Sekiguchi 1995). More recently, George (1997) has argued that genetic divergences of a subset of Panulirus species based on allozyme data (Chan and Chu 1996) suggest speciation of Groups I and II as early as the late Miocene. That geologic time is associated with the formation of new habitats after collisions of India and Australia with the Asian plate and associated alteration of circulation patterns and restricted larval transport between the Pacific and Indian Oceans, which favoured divergence between some western Pacific and Indian Ocean species (George 1997). Few studies have used a phylogenetic approach to test these biogeographic hypotheses or to examine the pattern of speciation among members of Panulirus . Sarver et al . (1998) examined a subset of species within Panulirus in
their description of a new Brazilian subspecies, P. argus westonii, based on nucleotide sequence divergence in two mitochondrial genes. Their analyses provided strong support for the genetic divergence of P. a. argus from P. a. westonii , but did not include enough of the remaining species of Panulirus to reveal phylogenetic relationships within the genus. Here, we report results of phylogenetic analyses based on nucleotide sequence data from two regions of the mitochondrial genome, the large-subunit ribosomal RNA (16S) and the cytochrome oxidase I subunit (COI). We included sequences of these gene regions for all species and four subspecies (total of 21 taxa) within the genus Panulirus in our phylogenetic analyses. Our specific aims in recovering the molecular phylogeny were to determine the phylogenetic relationships among all members of the genus and to test previously proposed taxonomic groupings and biogeographic hypotheses. The mitochondrial gene regions used in this study have proven useful for reconstructing detailed phylogenies that are independent of morphological characters and are free from the environmental influence that may bias morphologically based topologies (Simon et al. 1994). Methods
Specimens included in the phylogeny We examined 19 species of Panulirus for sequence variation at both the 16S and COI gene regions. For two species, P . homarus and P. longipes, two recognized subspecies of each, P. homarus homarus and P. h. megasculpta (Berry 1974) and P. longipes longipes and P. l. femoristriga , were sequenced for both gene regions. The P. l. femoristriga used in this analysis was the cross-banded form and has also been described as P. femoristriga (Chan and Chu 1996), P. longipes shirahige (George 1997), and most recently as P. l.
1039
bispinosus (Chan and Ng 2001). The red rock lobster, Jasus edwardsii, a member of Silentes, the sister taxon to Stridentes (which includes Panulirus) (George and Main 1967), was used as the outgroup taxon in
these analyses. Locality information for each of the specimens included in the phylogenetic analyses is given in Table 2. A single individual of each species was sequenced.
Table 2. Sample locations and GenBank accession numbers for species used in COI and 16S sequence analyses Species abbreviations are included for data presented in Table 3 Gene COI Species Jasus edwardsii P. argus argus P. cygnus P. echinatus P. gracilis P. guttatus P. homarus homarus P. homarus megasculpta P. inflatus P. interruptus P. japonicus P. laevicauda P. longipes femoristriga1 P. longipes longipes P. marginatus P. ornatus P. pascuensis P. penicillatus P. polyphagus P. regius P. stimpsoni P. versicolor Jasus edwardsii P. argus argus P. cygnus P. echinatus P. gracilis P. guttatus P. homarus homarus P. homarus megasculpta P. inflatus P. interruptus P. japonicus P. laevicauda P. longipes femoristriga P. longipes longipes P. marginatus P. ornatus P. pascuensis P. penicillatus P. polyphagus P. regius P. stimpsoni P. versicolor Abbreviation jas arg cyg ech gra gut hoh hom inf int jap lae lof lol mar orn pas pen pol reg sti ver jas arg cyg ech gra gut hoh hom inf int jap lae lof lol mar orn pas pen pol reg sti ver Location Fiordland, New Zealand Long Key, Florida, USA Western Australia Brazil Mazatln, Mexico Long Key, Florida, USA Marquesas Is. Sadh, Oman Mazatln, Mexico Goleta Bay, Calif., USA Shikoki Is., Japan Brazil Torres Strait, Australia Philippines Hawaii Torres Strait, Australia Easter Is. Palau Is. Singapore Pointe-Noire, Congo Hong Kong Palau Is. Fiordland, New Zealand Miami, Florida, USA Western Australia Brazil Mazatln, Mexico Miami, Florida, USA Singapore Sadh, Oman Mazatln, Mexico Southern California, USA Japan Brazil Singapore Philippines Hawaii Singapore Easter Is. Singapore Singapore Pointe-Noire, Congo Hong Kong Philippines Submitted by R. W. George M. J. Childress J. Silberman J. Silberman R. W. George M. J. Childress R. W. George R. Mohan R. W. George S. Anderson R. W. George J. Silberman C. R. Pitcher F. Parrish C. R. Pitcher R. W. George P. & L. Colin P. Ng Dr. Crosnier Ka Hou Chu P. & L. Colin R. W. George J. Silberman J. Silberman J. Silberman M. Hendrikx J. Silberman P. Ng R. Mohan M. Hendrikx J. Silberman M. Katoh J. Silberman P. Ng F. Parrish P. Ng R. W. George P. Ng P. Ng Dr. Crosnier Ka Hou Chu GenBank Accession No. AF339473 AF339452 AF339453 AF339454 AF339455 AF339456 AF339457 AF339458 AF339459 AF339460 AF339461 AF339462 AF339463 AF339464 AF339465 AF339467 AF339466 AF339468 AF339469 AF339470 AF339471 AF339472 AF337979 AF337966 AF337967 AF337965 AF337964 AF337963 AF337962 AF337961 AF337960 AF337959 AF337968 AF337969 AF339156 AF337970 AF337972 AF337971 AF337973 AF337974 AF337975 AF337976 AF337977 AF337978
16S
Cross-banded form called P. femoristriga by Chan and Chu (1996), P. longipes shirahige by George (1997) and P. longipes bispinosus by Chan and Ng (2001).
1040
Molecular Techniques Genomic DNA was extracted from muscle tissue of each individual with the GNOME DNA extraction kit (BIO 101, Vista, California) for polymerase chain reaction (PCR; Saiki et al. 1988) amplification of the COI gene region. Standard phenol/chloroform extraction techniques were used for DNA isolation from specimens used to amplify the 16S region (Hillis et al. 1990). Portions of these two mitochondrial genes (550 base pairs (bp) of 16S and 650 bp of COI) were amplified by PCR with conserved primers: 16Sar-L and 16Sbr-H for 16S (Palumbi et al. 1991), L-CO1490 and H-CO2198 for COI (Folmer et al. 1994). Conditions for symmetric PCR amplification of each gene region followed standard conditions (Palumbi 1996). Each set of primers was used for both PCR amplification and sequencing. A ~650-bp segment of COI was amplified, and PCR products were purified with Qiaquick PCR purification columns (Qiagen, Inc., Valencia, California). This purified product was used as template in sequencing reactions with the Taq DyeDeoxy Terminator Cycle Sequencing Kit (Applied Biosystems Inc., Foster City, California). These were run on the Applied Biosystems Model 377 Automated DNA Sequencing System located in the Molecular Core Research Facility at Idaho State University. Sequences were determined for both DNA strands for all individuals. A ~550-bp segment of 16S was amplified, and PCR products were purified as indicated above. The purified double-stranded amplification products were used as templates for DNA sequencing with the Taq cycle sequencing kit (U.S. Biochemical Corp., Cleveland, Ohio). Cycle sequencing reactions used 33P-dATP endlabelled primers and approximately 0.1 g of DNA template, followed by manual sequencing. Sequences were determined for both DNA strands for all individuals. Alignment and Phylogenetic Analyses Alignments of sequences for both gene regions were first generated with Clustal V (Higgins and Sharp 1988) and then adjusted by eye to produce the final alignments. Two ambiguous regions containing multiple indels (total of 27 characters) were detected in the 16S alignment and were deleted before phylogenetic analyses. The 22 taxa for the 16S and COI alignments were analysed by maximum parsimony, neighbour joining (Saitou and Nei 1987), and maximum likelihood (Felsenstein 1981) as implemented in version 4.0b4a of PAUP* (Swofford 1998). The data from the two gene regions were analysed in a two-step process. First, the two sets of data were separately subjected to maximum parsimony, neighbour-joining, and maximum-likelihood analyses. The maximum-parsimony analysis employed a heuristic search, with the following options: keep minimal trees only, collapse zero branch lengths, random stepwise addition of taxa with 100 replications, tree bisection and reconnection for branch swapping, and MULPARS option on. Character states were unordered, and all characters were equally weighted. The neighbour-joining searches assumed the generalized nucleotide substitution model of Hasegawa et al. (1985), HKY85, for calculating distances (Swofford et al. 1996). To calculate phylogenies under the maximum-likelihood optimality criterion, we used an iterative strategy (Swofford et al. 1996). Initially, we used a maximum-parsimony tree as a starting tree for the analysis. We then performed a heuristic search, assuming the HKY85 substitution model, with a gamma correction for among-site rate variation of 0.50, and simultaneously estimating model parameters. The likelihood of the maximum-parsimony tree was significantly improved by the addition of these parameters to the model at each step (Huelsenbeck and Crandall 1997). We used the resulting phylogeny as the starting point for another search and repeated this process until the analysis converged on a single phylogeny. The following model of evolution for the maximum-likelihood algorithm was therefore used
for the 16S data set: maximum-likelihood estimates of the base-pair frequencies under the HKY85 model, transition to transversion ratio of 3.175, and gamma distribution shape factor of 0.379. The model of evolution for the maximum-likelihood algorithm used for the COI data set was: maximum-likelihood estimates of the base-pair frequencies under the HKY85 model, transition to transversion ratio of 7.639, and gamma distribution shape factor of 0.512. Models used for both data sets had four categories of substitution rate. The next step in analysis involved combining the alignments from 16S and COI into a single data file and subjecting this combined data set to maximum-parsimony, neighbour-joining, and maximumlikelihood analyses. Maximum-parsimony and neighbour-joining analyses used the same parameters employed in the analyses of the individual data sets. The model of evolution for the maximumlikelihood algorithm used for the combined data set was: maximumlikelihood estimates of the base-pair frequencies under the HKY85 model, transition to transversion ratio of 4.097, and gamma distribution shape factor of 0.517. The model had four categories of substitution rate. To evaluate the degree of phylogenetic signal in both data sets, 16S and COI, we performed two tests for phylogenetic structure. First, we performed the cladistic permutation tail probability (PTP) test (Archie 1989; Faith and Cranston 1991) as implemented in PAUP*. This test uses the original data set to produce 100 randomly generated data sets, randomizing only the data among the ingroup taxa. Parsimony analyses using heuristic searches were then performed on each of the random data sets. The observed minimum length of the trees resulting from analyses of the random data sets was then compared to tree length of the minimal-length trees observed from a heuristic search performed on the original data set. The null hypothesis of no difference between tree lengths of original and random data sets can be rejected at the 0.05 level if 5% or fewer of the random data sets have a tree length equal to or less than that of the original most-parsimonious tree(s) (PTP 0.05). A second test for phylogenetic signal in both 16S and COI data sets employed an estimation of skewness of tree-length distributions from maximum-parsimony analyses (Huelsenbeck 1991; Hillis and Huelsenbeck 1992), again as implemented in PAUP*. The shape of a tree-length distribution provides a good indication of the presence of phylogenetic signal in a data set (Huelsenbeck 1991). Distributions of tree lengths with a strong left skew indicate that relatively few solutions exist near the optimal solution compared to elsewhere in the distribution. This pattern, in turn, is an indication of correlation among characters beyond that expected at random and therefore of significant phylogenetic signal in the data set (Hillis and Huelsenbeck 1992). For our analyses, we evaluated 10,000 random trees each for the 16S and COI data sets. Finally, we calculated bootstrap values (Felsenstein 1985) to examine the degree of support for the nodes of trees resulting from maximum parsimony and neighbour joining for the separate and combined data sets. For each data set we perfomed 1000 bootstrap iterations using heuristic searches and the same parameters as those used in the original analyses.
Results Nucleotide Composition of COI and 16S Alignments The alignment of COI sequences for all 22 taxa was 643 bp in length. Of these, 341 characters were constant, 64 were variable but uninformative, and 238 were parsimonyinformative. All sequence differences among taxa consisted of single-base-pair substitutions except for two 3-bp deletions found in P. gracilis. Sequences have been deposited with GenBank (Table 2).
1041
The COI region for these 22 taxa was 57% A + T rich, which is similar to the 59.5% A/T bias reported among species of Jasus (Ovenden et al. 1997) and slightly less than that reported for insects (range of 6875%, Lunt et al. 1996). Nucleotide content averaged 24.8% A, 32.2% T, 22.6% C, and 20.4% G, with very little deviation among taxa. The transition/transversion ratio estimated from the maximum parsimony analysis was 6.97. This bias favouring transitions is similar to that reported by other studies of substitutions in mtDNA (Moritz et al. 1987). The alignment of 16S sequences for all 22 taxa was 487 bp in length. Twenty-seven characters were excluded because of ambiguities in the alignment of these regions. Of the remaining 460 characters, 269 were constant, 69 were variable but uninformative, and 122 were parsimonyinformative. Sequence differences among taxa consisted of single-base-pair substitutions and 29 indel mutations. Of these 29 gaps, two regions had a 2-bp indel and one region had a 4-bp indel. The remainder of these mutations consisted of single-base-pair indels. Gaps were treated as missing data in the analyses. Sequences have been deposited with GenBank (Table 2). The 16S region for these 22 taxa was 64.2% A + T rich, which is similar to the 62% A/T bias reported for species of Jasus (Ovenden et al. 1997) but less than that reported for insects (range of 7782%, Simon et al. 1994). Nucleotide content averaged 31.1% A, 33.1% T, 13.6% C, and 22.2% G. The transition/transversion ratio estimated from the maximum-parsimony analysis was 2.99. The pairwise sequence divergences based on Kimuras (1980) two-parameter model, which corrects for the 2:1 transition bias in mtDNA nucleotide substitutions, were consistently lower in the 16S than in the COI region (Table 3). This result suggests that the 16S gene is more conserved than the COI gene in Panulirus, a finding similar to that reported for species of Jasus (Ovenden et al. 1997) and several insect genera (Simon et al. 1994). A total of 1104 characters was included for analysis in the combined COI/16S data set. Of these, 610 characters were constant, 134 were variable but uninformative, and 360 were parsimony-informative. Phylogenetic Relationships A significant level of phylogenetic signal was detected in both 16S and COI data sets. The randomization technique of Faith and Cranston (1991), the PTP test, confirmed that both 16S and COI original data sets were phylogenetically informative (PTP = 0.01 for 16S and PTP = 0.001 for COI). Critical values of g1 (test statistic for skewness of treelength distributions) were less (i.e. more negative) than tabled values (Hillis and Huelsenbeck 1992) for both 16S and COI data sets, indicating that each was significantly more structured than were random data. The value of g1 for the 16S data set was 0.47, which is less than the tabled
value of g1 = 0.20 (no. of taxa = 15, no. of characters = 100, P = 0.01). The value of g1 for the COI data set was 0.45, which is less than the tabled value of g1 = 0.16 (no. of taxa = 15, no. of characters = 250, P = 0.01). Thus parsimony trees generated by phylogenetic analyses of both 16S and COI data sets should show more structure than expected at random. Maximum-parsimony, neighbour-joining, and maximum-likelihood analyses of the 16S data set all yielded trees with the same overall topology. Two well-supported major lineages of Panulirus species were recovered by all three methods of phylogenetic analysis. The first major lineage included all species of the morphologically based Groups I and II; the second lineage contained all species included in Groups III and IV (George and Main 1967). Maximum-parsimony analysis of the 16S data set recovered two minimal-length trees (Fig. 1A). Neighbour joining recovered a topology similar to the maximumparsimony trees with some minor differences in relationships among species within the Group III/IV lineage. The maximum-likelihood tree (Fig. 1B) also recovered the same two major lineages among species of Panulirus, but its topology differed from those of maximum parsimony and neighbour joining in its placement of P. interruptus as the sister to the remaining Group I/II species and P. polyphagus as the sister to the remaining Group III/IV species. In contrast, the maximum-parsimony and neighbour-joining trees placed P. interruptus as the second lineage to branch off (after P. argus) in Group I/II, and P. polyphagus was nested within the Group III/IV lineage. Placements of these two species were not strongly supported by bootstrap analyses, however (Fig. 1A). Trees resulting from the different analysis methods also differed in relationships among Group III/IV species. Maximum-parsimony analysis of the COI data set yielded three minimal-length trees, again showing two major lineages of species of Panulirus, a Group I/II lineage and a Group III/IV lineage (Fig. 2A). Neighbour joining recovered a similar overall topology but differed with respect to some of the sister relationships recovered among species within each of the two major lineages. Bootstrap support for the Group I/II lineage was lower than that from the 16S data set (Fig. 2A). Bootstrap support for the Group III/IV lineage was strong for both 16S and the COI data sets. The maximum-likelihood analysis also recovered the same two major lineages of Panulirus as were recovered by maximum-parsimony and neighbour-joining analyses (Fig. 2B). The maximum-likelihood tree differed from the parsimony and neighbour-joining trees in its placement of P. argus as the sole sister taxon to the remaining Group I/II species. In contrast, the maximum-parsimony trees placed P. guttatus, P. echinatus and P. penicillatus along with P. argus as a sister clade to the remaining species in Group I/II. The neighbour-joining tree placed P. guttatus as the
1042
Table 3. Interspecific sequence divergence estimates for pairs of Panulirus species for 16S (above diagonal) and COI (below diagonal) Sequence divergence estimates are based on Kimuras (1980) two-parameter model. See Table 2 for species abbreviations hoh 0.175 0.170 0.184 0.071 0.165 --0.140 0.156 0.288 0.241 0.170 0.275 0.270 0.248 0.127 0.268 0.270 0.173 0.186 0.183 0.196 0.264 0.179 0.174 0.189 0.080 0.166 0.007 --0.148 0.297 0.251 0.162 0.231 0.215 0.220 0.145 0.256 0.259 0.170 0.173 0.173 0.171 0.256 0.166 0.167 0.199 0.065 0.171 0.051 0.053 --0.332 0.280 0.154 0.272 0.266 0.256 0.154 0.268 0.243 0.175 0.181 0.173 0.159 0.258 0.142 0.126 0.148 0.156 0.118 0.145 0.149 0.153 --0.280 0.310 0.297 0.315 0.287 0.318 0.297 0.292 0.296 0.322 0.301 0.305 0.323 0.139 0.079 0.127 0.169 0.121 0.146 0.145 0.160 0.121 --0.263 0.232 0.221 0.211 0.253 0.238 0.260 0.255 0.263 0.259 0.261 0.279 0.168 0.178 0.187 0.080 0.178 0.075 0.073 0.067 0.155 0.165 --0.243 0.233 0.253 0.188 0.290 0.288 0.173 0.170 0.184 0.181 0.265 0.187 0.062 0.145 0.203 0.134 0.188 0.190 0.180 0.134 0.106 0.196 --0.060 0.140 0.254 0.198 0.234 0.258 0.266 0.258 0.254 0.286 0.164 0.068 0.158 0.201 0.150 0.183 0.186 0.155 0.127 0.109 0.203 0.078 --0.124 0.236 0.171 0.207 0.239 0.259 0.237 0.258 0.279 0.157 0.047 0.133 0.194 0.131 0.170 0.175 0.177 0.131 0.087 0.185 0.065 0.076 --0.247 0.169 0.230 0.228 0.250 0.251 0.253 0.264 0.181 0.193 0.190 0.079 0.168 0.051 0.054 0.067 0.170 0.181 0.080 0.200 0.207 0.197 --0.276 0.250 0.179 0.185 0.173 0.178 0.281 0.185 0.065 0.140 0.205 0.134 0.196 0.197 0.185 0.130 0.109 0.198 0.028 0.081 0.067 0.199 --0.212 0.247 0.280 0.246 0.284 0.268 0.194 0.144 0.076 0.212 0.153 0.206 0.211 0.189 0.159 0.138 0.209 0.150 0.151 0.149 0.215 0.148 --0.273 0.262 0.252 0.246 0.257 0.173 0.176 0.200 0.100 0.172 0.071 0.072 0.080 0.154 0.170 0.111 0.207 0.195 0.192 0.079 0.206 0.201 --0.185 0.164 0.179 0.250 0.189 0.205 0.224 0.089 0.200 0.097 0.103 0.087 0.181 0.199 0.093 0.238 0.232 0.228 0.099 0.236 0.225 0.118 --0.186 0.189 0.256 0.175 0.173 0.194 0.081 0.169 0.074 0.075 0.068 0.163 0.161 0.086 0.200 0.183 0.179 0.071 0.202 0.191 0.090 0.089 --0.147 0.263 0.175 0.194 0.185 0.105 0.183 0.076 0.080 0.092 0.170 0.164 0.078 0.207 0.208 0.197 0.069 0.202 0.196 0.103 0.132 0.089 --0.240 hom inf int jap lae lof lol mar orn pas pen pol reg sti ver jas 0.286 0.253 0.240 0.257 0.256 0.234 0.237 0.261 0.256 0.260 0.257 0.280 0.290 0.268 0.238 0.268 0.267 0.253 0.264 0.233 0.257 ---
arg
cyg
ech
gra
gut
arg cyg ech gra gut hoh hom inf int jap lae lof lol mar orn pas pen pol reg sti ver jas
--0.211 0.248 0.253 0.218 0.262 0.269 0.237 0.307 0.248 0.263 0.218 0.209 0.223 0.257 0.236 0.226 0.242 0.266 0.233 0.246 0.287
0.157 --0.238 0.265 0.209 0.242 0.221 0.243 0.289 0.210 0.253 0.136 0.135 0.139 0.242 0.175 0.219 0.245 0.258 0.242 0.244 0.255
0.176 0.145 --0.272 0.239 0.268 0.280 0.230 0.318 0.261 0.301 0.237 0.236 0.261 0.264 0.222 0.131 0.288 0.291 0.275 0.260 0.251
0.190 0.172 0.203 --0.255 0.126 0.152 0.172 0.329 0.269 0.159 0.282 0.272 0.264 0.147 0.304 0.251 0.175 0.162 0.164 0.174 0.289
0.160 0.122 0.149 0.177 --0.252 0.216 0.242 0.271 0.233 0.234 0.212 0.198 0.218 0.247 0.234 0.227 0.234 0.277 0.249 0.276 0.293
1043
A.
22 81,83 11
30 20 19 16 21 12 99,100 16
B.
P. argus (I) P. interruptus (I) P. guttatus (II) P. penicillatus (II) P. echinatus (II)
11 2 0.030 0.038
0.180 0.091 0.062 0.111 0.038 0.035 0.003 0.023 0.108 0.034 0.018 0.017 0.033 0.014 0.058 0.047 0.063 0.068 0.014 0.019 0.006 0.086 0.008 0.006 0.007 0.048 0.059 0.025 0.006 0.018 0.063 0.001 0.024 0.008 0.665 0.003 0.019 0.066
P. argus (I) P. guttatus (II) P. penicillatus (II) P. echinatus (II) P. marginatus (I) P. cygnus (I) P. l. longipes (I) P. pascuensis (I) P. l. femoristriga (I) P. japonicus (I) P. interruptus (I) P. polyphagus (III) P. laevicauda (III) P. inflatus (III) P. regius (III) P. gracilis (III) P. stimpsoni (IV) P. ornatus (IV) P. versicolor (IV) P. h. homarus (IV) P. h. megasculpta (IV) Jasus edwardsii
P. marginatus (I)
6
14 4 12 9 56,89 7 18 14
12 99,100 6
P. pascuensis (I) P. l. femoristriga (I) P. cygnus (I) P. l. longipes (I) P. japonicus (I)
88,100
17
P. polyphagus (III)
18 7 6 5 8 1 7 3 10 99,100 2 14
P. laevicauda (III) P. versicolor (IV) P. ornatus (IV) P. h. homarus (IV) P. h. megasculpta (IV) P. inflatus (III)
0.083
19
96,100
13 22 4 15 70
Fig. 1. (A) Strict consensus of two minimal-length trees based on maximum-parsimony analysis of the 16S data set for the 22 taxa examined (tree length = 533; consistency index = 0.552; homoplasy index = 0.448; retention index = 0.644). Numbers on the branches are branch lengths measured as number of nucleotide substitutions. Bootstrap support (>50%) for 1000 iterations for maximum parsimony and neighbour joining, respectively, is given in circles at the nodes. (B) Optimal tree topology from maximum-likelihood analysis with branch lengths indicated (-Ln = 2978). (I) = Group I; (II) = Group II; (III) = Group III; (IV) = Group IV, all based on the morphological groups of George and Main (1967).
sister to P. argus and with this clade as sister to all the remaining Group I/II species except P. interruptus and P. japonicus, the first clade coming off of the Group I/II lineage. A P. echinatus, P. penicillatus lineage was nested within the Group I/II clade in the neighbour-joining tree. Bootstrap support was weak ( 38%), however, for all of these arrangements in either the maximum-parsimony trees or the neighbour-joining tree. All three topologies differed slightly in their relationships among species within the Group III/IV lineage, but these differences were also not strongly supported by bootstrap analyses. The combined COI/16S data set yielded seven minimallength trees from maximum-parsimony analysis (Fig. 3A). The same two major lineages (Group I/II and Group III/IV) were again recovered and strongly supported by bootstrap analyses for both maximum-parsimony and neighbourjoining analyses (Fig. 3A). Neighbour-joining analysis produced a topology nearly identical to the maximumlikelihood tree (Fig. 3B). The major difference between the maximum-likelihood tree and the maximum-parsimony tree
topology was in the placement of P. argus. Similarly to the separate analyses of each data set, the maximum-likelihood analysis of the combined data placed P. argus as sister to the remaining Group I/II species. In contrast, maximumparsimony analysis placed P. interruptus as sister to the remaining Group I/II species and P. argus as sister to a P. echinatus/P. penicillatus clade within the Group I/II lineage. Bootstrap support for these arrangements in the maximum-parsimony trees was weak, however (< 50% for each node; Fig. 3A). Again, maximum-parsimony, neighbour-joining, and maximum-likelihood trees differed slightly in relationships among species within the Group III/ IV lineage. Discussion All methods of phylogenetic reconstruction used on the two gene regions either separately or combined consistently recovered two major evolutionary lineages among species of Panulirus that correspond to the morphologically based Group I/II and Group III/IV, but the molecular phylogeny
1044
A.
51 23 44 44 97,100 32
B.
P. argus (I) P. guttatus (II) P. echinatus (II) P. penicillatus (II) P. japonicus (I)
27 31 57,63 23 28 100,100 33 58,90 44 75 14 0.447 0.044 0.101
0.405 0.204 0.220 0.148 0.449 0.505 0.178 0.023 0.073 0.067 0.218 0.084 0.146 0.064 0.144 0.044 0.318 1.714 0.155 0.029 0.234 0.086 0.045 0.328 0.214 0.239 0.133 0.120 0.089 0.007 0.214 0.298
P. argus (I) P. echinatus (II) P. penicillatus (II) P. guttatus (II) P. japonicus (I) P. l. femoristriga (I) P. l. longipes (I) P. cygnus (I) P. marginatus (I) P. pascuensis (I) P. interruptus (I) P. gracilis (III) P. laevicauda (III) P. regius (III) P. inflatus (III) P. h. megasculpta (IV) P. h. homarus (IV) P. ornatus (IV) P. polyphagus (III)
46
P. cygnus (I) P. l. femoristriga (I) P. l. longipes (I) P. marginatus (I) P. pascuensis (I) P. interruptus (I)
41
28 12 15 20 20 58 35 14 10 54,67 39 34 38
P. gracilis (III) P. inflatus (III) P. laevicauda (III) P. regius (III) P. h. homarus (IV) P. ornatus (IV) P. h. megasculpta (IV)
45 0.061 0.010
26
36
21 27 87,99 33 97
P. stimpsoni (IV)
0.207
63,74
40
0.217
P. stimpsoni (IV)
0.265
P. versicolor (IV)
0.994
Jasus edwardsii
Fig. 2. (A) Strict consensus of three minimal-length trees based on maximum-parsimony analysis of the COI data set for the 22 taxa examined (tree length = 1394; consistency index = 0.367; homoplasy index = 0.633; retention index = 0.420). Numbers on the branches are branch lengths measured as number of nucleotide substitutions. Bootstrap support (>50%) for 1000 iterations for maximum parsimony and neighbour joining, respectively, is given in circles at the nodes. (B) Optimal tree topology from maximum-likelihood analysis with branch lengths indicated (-Ln = 6032). (I) = Group I; (II) = Group II; (III) = Group III; (IV) = Group IV, all based on the morphological groups of George and Main (1967).
does not support the monophyly of any of the individual Groups IIV (George and Main 1967). The degree of genetic divergence between species suggests that species in Group I/II may have radiated earlier than species in Group III/IV. The level of sequence divergence was consistently higher among species in Groups I and II (16S: 2.819.4%; COI: 12.431.8%) than that among species in Groups III and IV (16S: 5.313.2%, COI: 12.619.6%). However, the higher level of sequence divergence among species in Groups I and II may also be due to a higher rate of molecular evolution at these mitochondrial genes for species in this lineage. The observed pattern of sequence divergence is consistent with previous hypotheses suggesting that species in Groups I and II represent an early radiation of Panulirus. These hypotheses were based on the retention of presumably more primitive or ancestral morphological traits in adult (George and Main 1967), phyllosoma, and puerulus forms (McWilliam 1995) for species in Groups I and II. The bootstrap support for a major division between Panulirus species of Group I/II and Group III/IV is reasonably strong, especially for the analyses of the 16S and
combined data sets (Fig. 1A and 3A), but within each of the two major lineages, support for relationships among species is much weaker. Within the Group I/II lineage, support is strong for a monophyletic lineage containing Group I species P. cygnus, P. marginatus, P. longipes and P. pascuensis. This group was consistently recovered by all three methods of phylogenetic analysis for each data set separately and for the combined data set. In addition, all analyses placed P. japonicus as sister to this lineage, although bootstrap support for this node is weak (e.g. Fig. 3A). However the remaining Group I species, P. argus and P. interruptus, are well outside of this Group I species clade and are clearly distantly related. Little support exists for Group II of George and Main (1967). Group II species P. echinatus and P. penicillatus are well-supported sister taxa, but they clearly do not share a recent common ancestor with P. guttatus, the third species in Group II. In addition, maximum-parsimony and maximum-likelihood trees differ considerably in the placement of these two taxa within the Group I/II lineage (Fig. 3A and 3B). McWilliam (1995) also argued that P. guttatus did not fit with other Group II
1045
A.
66
91 57 100,10 0 51
B.
P. argus (I) P. echinatus (II) P. penicillatus (II) P. japonicus (I) P. cygnus (I) P. marginatus (I)
0.144 0.058 0.077
0.215
P. argus (I)
0.062 0.010 0.019 0.077 0.128 0.037 0.025 0.048 0.167 0.078 0.025 0.041 0.133 0.078 0.161 0.364 0.072 0.010 0.095 0.024 0.020 0.140 0.079 0.082
P. cygnus (I) P. marginatus (I) P. pascuensis (I) P. l. femoristriga (I) P. l. longipes (I) P. japonicus (I) P. echinatus (II) P. penicillatus (II) P. guttatus (II) P. interruptus (I) P. gracilis (III) P. laevicauda (III) P. regius (III) P. inflatus (III) P. ornatus (IV)
0.033
76,81 62
59
25 26 22 41
60 74
18 18 49
0.018
67,72 45
45
0.025
P. h. homarus (IV) P. h. megasculpta (IV) P. stimpsoni (IV) P. versicolor (IV) P. polyphagus (III)
99,100
56 0.039 58
P. versicolor (IV)
64
P. polyphagus (III)
170
Jasus edwardsii
Jasus edwardsii
Fig. 3. (A) Strict consensus of seven minimal-length trees based on maximum-parsimony analysis of the combined 16S and COI data sets for the 22 taxa examined (tree length = 1962; consistency index = 0.411; homoplasy index = 0.589; retention index = 0.473). Numbers on the branches are branch lengths measured as number of nucleotide substitutions. Bootstrap support (>50%) for 1000 iterations for maximum parsimony and neighbour joining, respectively, is given in circles at the nodes. (B) Optimal tree topology from maximum-likelihood analysis with branch lengths indicated (-Ln = 9208). (I) = Group I; (II) = Group II; (III) = Group III; (IV) = Group IV, all based on the morphological groups of George and Main (1967).
species on the basis of its possession of more derived phyllosoma characters. Within the Group III/IV major lineage of Panulirus, there is little support for any groupings among species; bootstrap values are low for most nodes within this lineage (Fig. 3A). Trees resulting from the different analysis methods also produced different relationships among species in Group III/IV (e.g. Fig. 3A and 3B), indicating poor resolution within this lineage. The molecular phylogeny therefore provides no support for a separation of Group III from Group IV whether these groupings are based on adult (George and Main 1967), phyllosoma, or puerulus traits (McWilliam 1995). Because most of the relationships within each of the two major lineages are only weakly supported, we can only begin to make tentative predictions concerning the pattern of speciation of Panulirus. Our findings based on mtDNA sequence divergences seem to agree with allozyme-based genetic distances (Chan and Chu 1996) in suggesting that species in Groups I and II radiated much earlier than the
Pleistocene, the time previously proposed by George and Main (1967), Pollock (1992, 1993), and Sekiguchi (1995). Indeed, George (1997) has recently revised his hypotheses regarding the timing of the initial radiation of Panulirus in light of the recent genetic evidence and suggests a late Miocene radiation of species in Group I/II. George and Main (1967) proposed an evolutionary hypothesis for speciation in Panulirus that in many ways is consistent with the pattern recovered by our molecular phylogeny. They proposed that the ancestral species of Panulirus was able to inhabit the sublittoral of an extensive geographic area in the equatorial region with free access to all oceans because of its wide environmental tolerance in both the larval and adult condition. They further suggested that, among modern species, those that fit this description most closely are P. longipes and P. argus. George and Main (1967) chose P. longipes to represent the ancestral species because it is better known to us. Our maximum-likelihood phylogeny of the combined data suggests a radiation of Group I/II species from an argus-like ancestor from which
1046
P. argus and P. interruptus diverged (Fig. 3B). Indeed, all methods of phylogenetic analysis of all three data sets (16S, COI, and combined) show an early split between a P. argus lineage and a P. interruptus lineage, followed by the radiation of the Indo-West Pacific species in Group I: P. cygnus, P. marginatus, P. longipes, P. pascuensis, and P. japonicus. Interoceanic connections and westward equatorial currents present in the late Miocene would support this pattern of evolutionary diversification between Group I taxa in different oceans (Pollock 1992; George 1997). A more recent radiation of species in Groups III and IV has been hypothesized on the basis of their smaller genetic distances from allozyme studies (Chan and Chu 1996; Macaranas et al. personal communication, cited by George 1997). This pattern is also supported by our mtDNA analyses, which show lower levels of sequence divergence and shorter branch lengths among species in Groups III and IV than among species in Groups I and II (Table 3, Fig. 3A). George and Main (1967) hypothesized that an Indo-West Pacific ancestor gave rise to the more recent radiation of species in Group III/IV, probably speciating in response to temporary isolation within restricted parts of the vast region of the Indo-West Pacific. Species in Group III/IV with restricted distributions such as P. regius, P. polyphagus, P. stimpsoni, and two subspecies of P. homarus, P. h. megasculpta and P. h. rubellus are good examples (George 1997). Our molecular phylogeny provides some support for George and Mains (1967) hypothesis of allopatric speciation within the Group III/IV lineage. For example, P. polyphagus is the first species to split off in the Group III/ IV clade in maximum-likelihood trees for both the 16S and combined data sets and in the maximum-parsimony trees for the COI data set, suggesting an early divergence of this species. The overall pattern of speciation recovered by our mtDNA phylogenies supports earlier hypotheses, based on morphological characters (George and Main 1967; McWilliam 1995), suggesting the early divergence of two major lineages in Panulirus. Further phylogenetic analyses are required to test relationships among species within these two major lineages. For example, the taxonomic status of the subspecies of P. homarus and P. longipes warrants further investigation. In both instances, subspecies are sister taxa within trees generated from the combined analyses, but levels of sequence divergence at the two genes suggest genetic differentiation has occurred between subspecies (Table 3). Future studies should address questions of genetic differentiation both within and between species of Panulirus. In summary, our molecular phylogeny has begun to clarify the relationships within the genus Panulirus. Our phylogenetic results represent a first step toward understanding the pattern of speciation in Panulirus solely
on the basis of molecular characters. We hope that future researchers will add to our molecular data set by including sequences of additional genes from both the nuclear and mitochondrial genomes for species and subspecies of Panulirus. In addition, we encourage others to use our phylogenetic framework as a starting point in attempts to uncover the origin and likely direction of change in key morphological, ecological, and behavioural traits possessed by members of this fascinating and highly divergent genus. Acknowledgments We thank the following people, who collected lobsters and sent tissue samples included in this study: S. Anderson, M. Bianchini, K. H. Chu, L. Colin, P. Colin, Dr. Crosnier, P. Davies, H. Fikiyuchi, R. George, S. Hane, M. Hendrikx, M. Katoh, K. Lavalli, J. Martin, R. Mohan, P. Ng, F. Parrish, C. R. Pitcher, D. Pollock, and J. Silberman. We are grateful to the following for laboratory assistance: J. Loxterman, T. McKay, J. Silberman, K. Smith, T. Spears, and J. Wiklund. The manuscript has been greatly improved by comments from J. Cook, J. Xiang, and two reviewers, R. George and T. Spears. Funding for this project was made possible by a COFERS grant from Florida State University to W.F.H. and a NSF EPSCoR in Idaho award to M.B.P. This paper is dedicated to Ray W. George, whose lifetime of research on palinurid lobsters has served as an inspiration to us all. References
Archie, J. W. (1989). A randomization test for phylogenetic information in systematic data. Systematic Zoology 38, 23952. Baisre, J. A., and Ruiz de Quevedo, M. E. (1982). Two phyllosome larvae of Panulirus laevicauda (Latrielle, 1817) (Decapoda, Palinuridae) from the Caribbean Sea with a discussion about larval groups within the genus. Crustaceana 43, 14753. Berry, P. D. (1974). A revision of the Panulirus homarus group of spiny lobsters (Decapoda, Palinuridae). Crustaceana 27, 3142. Chan, T. Y., and Chu, K. H. (1996). On the different forms of Panulirus longipes femoristriga (von Martens, 1872) (Crustacea: Decapoda: Palinuridae), with description of a new species. Journal of Natural History 30, 36787. Chan, T. Y., and Ng, P. K. L. (2001). On the nomenclature of the commercially important spiny lobsters Panulirus longipes femoristriga (Von Martens, 1872), P. bispinosus Borradaile, 1899, and P. albiflagellum Chan and Chu, 1996 (Decapoda, Palinuridae). Crustaceana 74(1), 1237. Faith, D. P., and Cranston, P. S. (1991). Could a cladogram this short have arisen by chance alone? On permutation tests for cladistic structure. Cladistics 7, 128. Felsenstein, J. (1981). Evolutionary trees from DNA sequences: a maximum likelihood approach. Journal of Molecular Evolution 17, 36876. Felsenstein, J. (1985). Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39, 78391. Folmer, O., Black, M., Hoeh, W., Lutz, R., and Vrijenhoek, R. (1994). DNA primers for amplification of mitochondrial cytochrome c oxidase subunit I from diverse metazoan invertebrates. Molecular Marine Biology and Biotechnology 3, 29499.
1047
George, R. W. (1997). Tectonic plate movements and the evolution of Jasus and Panulirus spiny lobsters (Palinuridae). Marine and Freshwater Research 48, 112130. George, R. W., and Main, A. R. (1967). The evolution of spiny lobsters (Palinuridae): a study of evolution in the marine environment. Evolution 21, 80320. Hasegawa, M., Kishino, H., and Yano, T. (1985). Dating of the humanape splitting by a molecular clock of mitochondrial DNA. Journal of Molecular Evolution 22, 16074. Higgins, D. G., and Sharp, P. M. (1988). Clustal: a package for performing multiple sequence alignment on a microcomputer. Genetics 73, 23744. Hillis, D. M., and Huelsenbeck, J. P. (1992). Signal, noise, and reliability in molecular phylogenetic analyses. Journal of Heredity 83, 18995. Hillis, D. M., Larson, A., Davis, S. K., and Zimmer, E. A. (1990). Nucleic acids III: sequencing. In Molecular Systematics, 1st ed. (Eds D. M. Hillis and C. Moritz.) pp. 318370. (Sinauer: Sunderland, Massachusetts.) Holthuis, L. B. (1991). FAO species catalogue. Vol. 13. Marine lobsters of the world: an annotated and illustrated catalogue of species of interest to fisheries known to date. FAO Fisheries Synopsis No. 125 Vol. 13. 292 pp. Huelsenbeck, J. P. (1991). Tree-length distribution skewness: an indicator of phylogenetic information. Systematic Zoology 40, 257 70. Huelsenbeck, J. P., and Crandall, K. A. (1997). Phylogeny estimation and hypothesis testing using maximum likelihood. Annual Review of Ecology and Systematics 28, 43766. Kimura, M. (1980). A simple method for estimating evolutionary rate of base substitutions through comparative studies of nucleotide sequences. Journal of Molecular Evolution 16, 11120. Lunt, D. H., Zhang, J. M., and Hewitt, G. M. (1996). The insect cytochrome oxidase I gene: evolutionary patterns and conserved primers for phylogenetic studies. Insect Molecular Biology 5, 153 65. McWilliam, P. S. (1995). Evolution in the phyllosoma and puerulus phases of the spiny lobster genus Panulirus White. Journal of Crustacean Biology 15, 54257. Moritz, C., Dowling, T. E., and Brown, W. M. (1987). Evolution of animal mitochondrial DNA: relevance for population biology and systematics. Annual Review of Ecology and Systematics 18, 269 92.
Ovenden, J. R., Booth, J. D., and Smolenski, A. J. (1997). Mitochondrial DNA phylogeny of red and green rock lobsters (genus Jasus). Marine and Freshwater Research 48, 113136. Palumbi, S. R. (1996). Nucleic acids. II. The polymerase chain reaction. In Molecular Systematics, 2nd ed. (Eds D. M. Hillis, C. Moritz, and B. K. Mable.) pp. 20548. (Sinauer: Sunderland, Massachusetts.) Palumbi, S. R., Martin, A., Romano, S., McMillan, W. O., Stice, L., and Grabowski, G. (1991). A simple fools guide to PCR, v2. 0. Special publication of the University of Hawaii Department of Zoology and Kewalo Marine Laboratory, pp. 123. Pollock, D. E. (1992). Palaeoceanography and speciation in the spiny lobster genus Panulirus in the Indo-Pacific. Bulletin of Marine Science 51, 13546. Pollock, D. E. (1993). Speciation in spiny lobsters-clues to climatically-induced changes in ocean circulation patterns. Bulletin of Marine Science 53, 93744. Saiki, R. K., Gelfand, D. H., Stoffel, S., Scharf, S. J., Higuchi, R., Horn, G. T., Mullis, K. B., and Erlich, H. A. (1988). Primerdirected enzymatic amplification of DNA with a thermostable DNA polymerase. Science 239, 48791. Saitou, N., and Nei, M. (1987). The neighbor-joining method: A new method for reconstructing phylogenetic trees. Molecular Biology and Evolution 4, 40625. Sarver, S. K., Silberman, J. D., and Walsh, P. J. (1998). Mitochondrial DNA sequence evidence supporting the recognition of two subspecies or species of the Florida spiny lobster Panulirus argus. Journal of Crustacean Biology 18, 17786. Sekiguchi, H. (1995). A hypothesis on the speciation of the Panulirus japonicus group (Crustacea, Decapoda, Palinuridae) based on geographic distributions. Proceeding of the Japanese Society for Systematic Zoology 54, 4459. Simon, C., Frati, F., Deckenbach, A., Crespi, B., Lui, H., and Flook, P. (1994). Evolution, weighting, and phylogenetic utility of mitochondrial gene sequences and a compilation of conserved polymerase chain reaction primers. Entomological Society of America 87, 651701. Swofford, D. L. (1998). PAUP*. Phylogenetic Analysis Using Parsimony (* and other methods). Version 4. (Sinauer: Sunderland, Massachusetts.) Swofford, D. L., Olsen, F. J., Waddell, P. J., and Hillis, D. M. (1996). Phylogenetic inference. In Molecular Systematics, 2nd ed. (Eds D. M. Hillis, C. Moritz, and B. K. Mable.) pp. 407514. (Sinauer: Sunderland, Massachusetts.)
http://www.publish.csiro.au/journals/mf