Host shift and speciation in a coral-feeding nudibranch

Proc. R. Soc. B doi:10.1098/rspb.2006.3685 Published online Host shift and speciation in a coral-feeding nudibranch Anuschka Faucci1,*, Robert J. Toonen2 and Michael G. Hadfield1 1 2 Kewalo Marine Laboratory, University of Hawaii at Manoa, 41 Ahui Street, Honolulu, HI 96813, USA Hawaii Institute of Marine Biology, University of Hawaii at Manoa, PO Box 1346, Kaneohe, HI 96744, USA While the role of host preference in ecological speciation has been investigated extensively in terrestrial systems, very little is known in marine environments. Host preference combined with mate choice on the preferred host can lead to population subdivision and adaptation leading to host shifts. We use a phylogenetic approach based on two mitochondrial genetic markers to disentangle the taxonomic status and to investigate the role of host specificity in the speciation of the nudibranch genus Phestilla (Gastropoda, Opisthobranchia) from Guam, Palau and Hawaii. Species of the genus Phestilla complete their life cycle almost entirely on their specific host coral (species of Porites, Goniopora and Tubastrea). They reproduce on their host coral and their planktonic larvae require a host-specific chemical cue to metamorphose and settle onto their host. The phylogenetic trees of the combined cytochrome oxidase I and ribosomal 16S gene sequences clarify the relationship among species of Phestilla identifying most of the nominal species as monophyletic clades. We found a possible case of host shift from Porites to Goniopora and Tubastrea in sympatric Phestilla spp. This represents one of the first documented cases of host shift as a mechanism underlying speciation in a marine invertebrate. Furthermore, we found highly divergent clades within Phestilla sp. 1 and Phestilla minor (8.1–11.1%), suggesting cryptic speciation. The presence of a strong phylogenetic signal for the coral host confirms that the tight link between species of Phestilla and their host coral probably played an important role in speciation within this genus. Keywords: cryptic species; host shift; nudibranchs; Phestilla; speciation 1. INTRODUCTION The requirements for ecological speciation (i.e. reproductive isolation via selection for an alternative environment or food source) have been defined using mathematical models and simulations (Coyne & Orr 2004; Doebeli et al. 2005). The general consensus is that ecological speciation is plausible and may be more common than believed formerly, and that the most likely scenarios involve habitat or host preference, as previously stated by Bush (1969) and Mayr (1976). When host preference and local adaptation are combined with mate choice on the preferred habitat or host, assortative mating can lead to population subdivision, adaptation and divergence (Diehl & Bush 1989; Via 2001). Under these conditions, subdivision and reduction in gene flow arise from reproductive rather than geographic barriers (Bush 1994). Host preferences leading to host shifts have been most extensively investigated in phytophagous insects (reviewed in Coyne & Orr 2004). Such shifts may allow incipient species to escape from direct competition for resources or to enjoy benefits from enemy-free space (Gross & Price 1988; Lill et al. 2002). The assumption that widespread marine species with larval dispersal should show little spatial variation in genetic structure, given their high potential for dispersal via ocean currents, has been questioned (e.g. Hellberg 1998; Carlon & Budd 2002; Rocha et al. 2005). Recent research has also shown that many nominal marine species consist of a number of cryptic taxa (reviewed in Knowlton 2000; Hart et al. 2003; Landry et al. 2003; Meyer et al. 2005). The presence of many cryptic species (i.e. difficult or impossible to distinguish by morphology) in the oceans may be related to the role of chemical recognition systems, which are involved in many aspects of reproduction and settlement in the marine environment (Knowlton 1993). Differences in substrate specificity provide the potential for niche partitioning, adaptive shifts and the formation of marine sibling species complexes (Knowlton 1993, 2000). Host preference is a powerful factor in promoting ecological speciation in plant–insect associations (Bush 1969; Coyne & Orr 2004). The apple maggot fly Rhagoletis pomonella is the most studied and best understood example of speciation by host shift (Bush 1969, 1994; Via 2001). Recent work has revealed that R. pomonella has many characteristics that are thought to be necessary for sympatric speciation (Feder 1998; Via 2001). For example, fruit odour discrimination led to pre-mating reproductive isolation in R. pomonella, which then resulted in host-race formation and incipient sympatric speciation (Linn et al. 2003). However, evidence from genetic data suggests that allopatry and secondary introgression may have acted in conjunction with host shifts to facilitate sympatric speciation in R. pomonella (Feder et al. 2003). Although there are still few such studies in marine communities (reviewed in Sotka 2005), several marine organisms show local adaptation to their hosts, suggesting that host use could play a fundamental role in the differentiation and speciation of groups, such as herbivorous amphipods (Stanhope et al. 1992; Sotka et al. 2003), sponge-dwelling alpheid shrimps (Synalpheus brooksi; Duffy 1996), * Author for correspondence (anuschka@hawaii.edu). Received 2 July 2006 Accepted 20 July 2006 1 q 2006 The Royal Society 2 A. Faucci and others Host shifts in Phestilla spp. coral-inhabiting barnacles (Savignium milleporum; Mokady & Brickner 2001), a predatory whelk (Nucella caniculata; Sanford et al. 2003), herbivorous saccoglossan sea slugs ( Jensen 1997; Trowbridge & Todd 2001) and coral-dwelling gobies (Munday et al. 2004). Opisthobranch molluscs have undergone major evolutionary radiations in connection with habitat and diet ( Jensen 1997; Rudman 1998), and represent an excellent group to investigate speciation and host shifts. Opisthobranchs, including nudibranchs, are among the best examples of marine specialists; approximately, 50% of the species for which feeding preference is established appear to consume only one prey species, and many others prefer just one or two (Todd et al. 2001). Studies on opisthobranchs have also shown high specificity among settling larvae (Hadfield 1977; Havenhand 1991; RitsonWilliams et al. 2003), and that settlement cues originate from the host (Hadfield 1977; Lambert & Todd 1994; Krug & Manzi 1999). Parallels between terrestrial, herbivorous insects and ‘insect-like’ marine invertebrates (i.e. those that are relatively immobile, small in size compared to their biotic hosts and use hosts for both food and shelter) have been discussed previously by Hay et al. (1987) and Sotka (2005). Herein, we use the term ‘host’ to describe the association between the tropical nudibranch Phestilla and its coral prey in order to make the comparison between this association and that described extensively in the insect literature (reviewed in Coyne & Orr 2004). Species of the aeolid nudibranch genus Phestilla (class Gastropoda, subclass Opisthobranchia and family Tergipedidae) feed and reproduce on specific scleractinian corals (table 1). The planktonic larvae of Phestilla spp. require a coral-specific chemical cue to settle onto their host coral, metamorphose (Hadfield 1977; Hadfield & Pennington 1990) and complete their life cycle almost entirely on their specific coral host. Ritson-Williams et al. (2003) studied the host specificity of four species of Phestilla (Phestilla minor, Phestilla sibogae, Phestilla sp. 1 and Phestilla sp. 2) in Guam. They investigated host preference of adults in choice and no-choice feeding assays, and tested larval host specificity by measuring percentage of metamorphosis in response to different coral species. Their data reveal that larvae of Phestilla spp. can generally distinguish among coral species within a host genus and have highest rates of metamorphosis on their preferred coral hosts (Ritson-Williams et al. 2003). Furthermore, Hadfield & Koehl (2004) showed that larvae of P. siboage display distinct behavioural reactions to their specific host in the water column, which affects the location of their settlement. All species of Phestilla have planktonic larvae, but species differ in larval type (i.e. feeding or non-feeding larvae) and therefore in the minimum time spent in the plankton, ranging from a few hours for species with non-feeding larvae (Phestilla sp. 1 and P. minor; Ritson-Williams et al. 2003) to 8 days for species with feeding larvae (Phestilla melanobrachia; Harris 1975). Phestilla sp. 1 and P. minor are competent to metamorphose at hatching (Ritson-Williams et al. 2003). Reduced dispersal potential of Phestilla sp. 1 and P. minor could facilitate local adaptation and eventually, together with assortative mating on the host, ecological speciation on different coral hosts. Proc. R. Soc. B Thus, the genus Phestilla represents an excellent system to investigate host shifts and the potential for ecological speciation. The goal of this study was to construct a molecular phylogeny of the genus Phestilla using mitochondrial DNA (mtDNA) and to investigate the possible role of host shifts in speciation. 2. MATERIAL AND METHODS (a) Species studied, specimen collection and DNA extraction Nudibranchs were collected from corals in Hawaii, Guam and Palau (table 1) and were preserved in O70% ethanol. Species were identified according to Bergh (1905), Rudman (1981) and Ritson-Williams et al. (2003). In addition to the nominal species Phestilla sibogae Bergh, 1905 (type locality: Indonesia), P. lugubris Bergh, 1870 (Philippines), P. melanobrachia Bergh, 1874 (Philippines) and P. minor Rudman, 1981 (Tanzania), two undescribed species (Phestilla sp. 1 and sp. 2 of Ritson-Williams et al. 2003) and two different morphotypes for P. minor were distinguished. As P. sibogae and P. lugubris have been synonymized without rationale (Rudman 1981), we treat them as two different species due to their differences in larval development and adult morphology. Phestilla sp. 2 was identified based on distinct morphology, larval biology and food source (Ritson-Williams et al. 2003). Furthermore, we consider Phestilla sp. 1 and P. minor morphotype II (P. minor II) as distinct from P. minor morphotype I (P. minor) due to identifiably different cerata (Ritson-Williams et al. 2003). The only known species in the genus Phestilla not included in this study is the rare Phestilla panamica from Panama. Phestilla panamica resembles P. sibogae in morphology, but we were not able to obtain any specimens. For each specimen collected, the morphology, location and coral species on which it was found were recorded. Total genomic DNA was extracted from muscle (or whole animal if smaller than 2 mm in length) using the QIAGEN DNeasy tissue kit. (b) PCR, sequencing and sequence alignment A fragment of the mitochondrial cytochrome oxidase I gene (COI ) was amplified using the universal COI primers (Folmer et al. 1994) under the following PCR conditions: 2 min at 948C, 35 cycles of 948C for 30 s, 408C for 30 s and 728C for 30 s with a final extension at 728C for 7 min. Amplification of a portion of the mitochondrial 16S ribosomal RNA gene (16S ) was performed using the primers 16sar-L and 16sbr-H (Palumbi et al. 1991) using the same PCR conditions, but with an annealing temperature of 508C. PCR products were purified using a QIAquick PCR Purification Kit (Qiagen) prior to cycle sequencing. DNA sequencing was performed using an ABI 377 automated DNA sequencer. The sequences were aligned and edited using SEQUENCHER v. 2.4 (Gene Codes). Alignments were confirmed and edited by eye in MACCLADE v. 4.05 (Maddison & Maddison 2000). Sequences were deposited in GenBank under accession numbers DQ417228–DQ417325. (c) Phylogenetic analysis Phylogenetic analyses were conducted with maximumparsimony (MP), maximum-likelihood (ML), neighbourjoining (NJ) and Bayesian methods using all 49 sequences. MP, ML and NJ analyses were performed with PAUP v. 4.0b10 (Swofford 2002). Support for individual nodes was assessed using 100 (ML) or 1000 (MP and NJ) bootstrap Proc. R. Soc. B Table 1. Species investigated, with larval types, collection sites, host coral, number of specimens included, collector with year of collection and GenBank accession numbers for sequences from COI and 16S. (Larval types: feeding (planktotrophic), non-feeding (lecithotrophic); collectors: AF, Anuschka Faucci; GP, Gustav Paulay; MH, Michael Hadfield and RR, Raphael Ritson-Williams.) GenBank accession no. larval type collection locality host coral n collection COI 16S Phestilla lugubris planktotrophic Phestilla melanobrachia planktotrophic Phestilla minor, morphotype I lecithotrophic Phestilla minor, morphotype II (ZPhestilla minor II) Phestilla sibogae lecithotrophic Guam, Pago Bay Guam, Apra Harbor Guam Oahu/HI, Kaneohe Bay Guam, Apra Harbor Palau, Western Channel Oahu/HI, Kaneohe Bay Guam, Pago Bay Guam, Tangrissan Reef Palau, Ngel Channel Palau, N of Ngel Channel Palau, Western Channel Oahu/HI, Kaneohe Bay Porites rus Porites rus Porites rus Tubastrea coccinea Tubastrea coccinea Tubastrea micrantha Porites compressa Porites annae Porites annae Porites lutea Porites lutea Porites lutea Porites compressa 1 1 1 3 3 3 3 2 3 2 2 1 3 RR, 2002 GP, 1996 MH, 1991 AF, 2002 RR, 2002 RR, 2003 AF, 2002 RR, 2003 RR, 2003 RR, 2003 RR, 2003 RR, 2003 AF, 2002 DQ417298 DQ417299 DQ417300 DQ417274-76 DQ417277-79 DQ417280-82 DQ417301-03 DQ417304-05 DQ417306-08 DQ417309-10 DQ417312-13 DQ417311 DQ417287-89 DQ417252 DQ417253 DQ417254 DQ417228-30 DQ417231-33 DQ417234-36 DQ417255 DQ417256-57 DQ417258-59 DQ417260-61 DQ417263-64 DQ417262 DQ417241-43 French Frigate Shoals/HI Guam, Pago Bay Palau, N of Ngel Channel Palau, Western Channel Guam, Luminao Reef Palau, N of Ngel Channel Palau, N of Ngel Channel Guam Palau, Lighthouse Reef Oahu/HI, Kaneohe Bay Porites compressa Porites lutea Porites rus Porites cylindrica Porites cylindrica Porites cylindrica Porites rus Goniopora fruticosa Goniopora djiboutiensis 1 3 2 2 3 4 4 3 1 1 AF, 2002 RR, 2003 RR, 2003 RR, 2003 RR, 2003 RR, 2003 RR, 2003 RR, 2003 RR, 2003 AF, 2002 DQ417290 DQ417291-93 DQ417294-95 DQ417296-97 DQ417314-16 DQ417319-20,23-24 DQ417317-18,21-22 DQ417283-85 DQ417286 DQ417325 DQ417244 DQ417245-47 DQ417248-49 DQ417250-51 DQ417265-67 DQ417269-72 DQ417268 DQ417237-39 DQ417240 DQ417273 facultative planktotrophic Phestilla sp. 1 lecithotrophic Phestilla sp. 2 lecithotrophic Caloria indica (outgroup) Host shifts in Phestilla spp. A. Faucci and others taxon 3 4 A. Faucci and others Host shifts in Phestilla spp. replicates. Bootstrap replicates for the ML tree were obtained using a reduced dataset (all replicate haplotypes were removed) with 32 sequences for computational reasons. The overall tree topology of the ML tree from this reduced dataset was identical to the ML tree including all 49 sequences. For ML and NJ analyses, the model for best nucleotide substitution was selected using the Akaike Information Criterion as implemented in MODELTEST v. 3.7 (Posada & Crandall 1998). The best fit to our combined dataset was provided by the general time reversible (GTR) model of substitution, with a gamma parameter (G) of 0.466 and a proportion of invariant sites (I) of 0.488. Bayesian analyses were conducted with MRBAYES v. 3.1.1p (Huelsenbeck & Ronquist 2001). Analyses were performed with uninformative priors. Four chains were used per run (three heated and one cold), and each analysis was repeated at least three times, twice with two million generations, and a final analysis running for 10 million generations. The COI and combined datasets were run as unpartitioned and partitioned datasets (i.e. the combined dataset was partitioned into the two genes and the COI dataset was partitioned according to the three different codon positions). All runs of the same dataset (partitioned or unpartitioned) produced nearly identical tree topologies, with the only exceptions being some minor rearrangements of sequences within a species or among clades with low support in all analyses. Analysis of the two datasets (COI and 16S) produced almost identical tree topologies when analysed individually; differences were all minor rearrangements of sequences within a species or among clades with low support. Therefore, the 16S and COI data were combined, and the combined dataset was used for all analyses presented here. Several nudibranch species within the Aeolidaea were included as outgroup taxa based on the most recent nudibranch phylogeny in Wollscheid et al. (2001); Eubranchus exiguus of the family Tergipedidae and the aeolids Caloria indica, Flabellina verrucosa and Flabellina pedata (for phylogeny and GenBank accession number, see Wollscheid et al. 2001 and table 1 for C. indica). Genetic distances were calculated through a pairwise distance matrix in MEGA v. 3.1 (Kumar et al. 2004). Monophyly of the nominal species Phestilla sp. 1 and P. minor were tested using the Shimodaira–Hasegawa test (SHT). An unconstrained ML tree was compared to an ML tree where monophyly was constrained for Phestilla sp. 1 or P. minor. The SHT was performed under the likelihood model resulting from Modeltest and using 1000 Resampling of Estimated Log Likelihood (RELL) replicates. To test the presence of a phylogenetic signal in the characters host-coral specificity and geographic location, we used the difference in the number of steps for a character on random trees compared to the observed number of steps on the MP tree. If the number of steps for a character is less in the observed data than in at least 95% of the randomized trees, we conclude that the evolution of this character is most likely associated with this tree (i.e. there is a phylogenetic signal in this character). MACCLADE was used to determine the presence of a phylogenetic signal in both host-coral specificity and geographic location. Taxa were coded according to host genus and species as well as geographic location. All four multistate characters: (i) host genus, (ii) host species, (iii) island and (iv) locality were then mapped onto 1000 MP and 1000 randomized trees. The thousand most parsimonious trees (created in PAUP) were randomized in MACCLADE using random joining and splitting. For all Proc. R. Soc. B characters, the distribution of the total number of steps (i.e. host switches or dispersal among locations) on each tree was then compared between the MP and the randomized trees. 3. RESULTS A 615 base pair (bp) fragment of the mitochondrial gene COI and a 404 bp fragment of the ribosomal 16S gene were sequenced in each of six species of the nudibranch genus Phestilla from Guam, Palau and Hawaii. Overall, 45 specimens of Phestilla from nine localities and nine different species of host corals (table 1) and four outgroups were included. Within the 16S gene sequences, two indels were present. Specifically, all sequences of Phestilla sp. 1 from Guam showed a 1 bp deletion at position 192 and all sequences of P. melanobrachia showed a 1 bp deletion at position 194. The combined dataset included 1019 bp out of which 649 sites were constant. Of the remaining variable sites, 321 were parsimony informative. The gross tree topology from the ML analysis of the combined dataset (figure 1) is identical to the most parsimonious tree, the NJ tree, as well as the 50% majority consensus tree resulting from the Bayesian analysis. All trees differed only in a few minor rearrangements of sequences within a species or among clades with low support (bootstrap supports (BS) and Bayesian posterior probabilities (PP) lower than 60). Phestilla melanobrachia and Phestilla sp. 2 are well distinguished by COI and 16S sequence comparisons, as shown by the highly supported and reciprocally monophyletic clades. Therefore, the mitochondrial phylogeny agrees with the morphological taxonomy of these species. In contrast, P. lugubris and P. sibogae, which occur on different host species of Porites, cannot be distinguished by COI and 16S sequence comparisons. They form a single highly supported monophyletic clade with no obvious structure corresponding to the two putative species. Phestilla minor morphotype II forms a monophyletic clade distinct from other sequences of P. minor. However, Phestilla sp. 1 and P. minor form a single monophyletic clade, where individuals of the two nominal species are intermixed. Support for some branches is low within this clade (lower than 60 for BS and PP) and monophyly was rejected for either species (SHT: p!0.001). Each of the sub-clades (figure 1) within the clade including P. minor I and Phestilla sp. 1 is highly supported (both BS and PP higher than 84) and represents specimens of one species from one location and from a single species of host coral. A total of six specimens, three from Phestilla sp. 1 (from Palau on Porites rus) and three from P. minor (one from Guam and two from Oahu), did not produce unambiguous 16S sequences and were therefore included only in the COI dataset. The inclusion or exclusion of these individuals did not change the overall tree topology. Both sub-clades, Phestilla sp. 1 from Palau on Porites rus and P. minor from Oahu on Porites compressa, were highly supported (BS and PP higher than 89) in all COI trees (tree not shown). For the combined dataset, the average pairwise sequence divergence (uncorrected p; table 2) between the outgroup taxa and each of the nominal species of Phestilla (considering P. minor II as a distinct species) was between 17.2 and 19.9%. Among nominal species of Phestilla, the divergence ranged from 7.3 to 15.8%, with the exception of the divergence between P. sibogae and Host shifts in Phestilla spp. A. Faucci and others 5 Figure 1. Maximum-likelihood tree (GTRCICG) of the combined COI and 16S dataset. Branch support is as follows: MP and NJ bootstrap support of 1000 replications are above; Bayesian posterior probabilities and ML bootstrap support of 100 replicates are below branches. Species of Phestilla are in bold italics, host coral species in italic and number of sequences in parentheses. Table 2. Average genetic distances (uncorrected p) of combined COI and 16S sequences for within (diagonal) and among (below diagonal) species of Phestilla and outgroup. 1 2 3 4 5 6 7 8 P. melanobrachia P. sp. 2 P. sibogae P. lugubris P. minor P. minor II P. sp. 1 outgroup 1 2 3 4 5 6 7 8 0.005 0.102 0.148 0.148 0.134 0.149 0.132 0.183 0.044 0.14 0.139 0.136 0.141 0.139 0.187 0.009 0.008 0.153 0.154 0.158 0.199 0.006 0.153 0.153 0.158 0.199 0.055 0.14 0.073 0.172 0.061 0.138 0.181 0.059 0.174 0.164 P. lugubris, which was only 0.8%. The average divergence within nominal species of Phestilla was between 0.5 and 0.9% for P. melanobrachia, P. sibogae and P. lugubris, and between 4.4 and 6.1% for Phestilla sp. 2, P. minor II, Proc. R. Soc. B P. minor and Phestilla sp. 1. The average divergence for COI (uncorrected p; data not shown) among locations within P. melanobrachia and P. sibogae/lugubris is 0.2–2.3%. In comparison, the average divergence among locations 6 A. Faucci and others Host shifts in Phestilla spp. within Phestilla sp. 1 and P. minor is much higher, ranging from 8.1 to 11.1%, with the exception of Phestilla sp. 1 from Palau on Porites cylindrica, which was only 0.8% divergent from P. minor from Palau on Porites lutea. However, out of the 10 variable sites of the six combined COI and 16S sequences between the two latter clades, five sites are fixed (i.e. sites at which all of the sequences in one clade are different from all of the sequences in the other clade). The most surprising result was the high sequence divergence (11.1%) between specimens of Phestilla sp. 1, which were collected on the same day and at the same locality in Palau but from different species of coral (Porites rus and P. cylindrica). The high divergence was also reflected in the high number of fixed differences between the eight COI sequences of these two clades (66 fixed sites). Randomization tests in MACCLADE found a strong phylogenetic signal of both host-coral genus and species use by species of Phestilla. For the combined dataset, the character ‘host genus’ involved two steps in the MP trees and 8–13 steps in the randomized trees, whereas ‘host species’ involved 15–18 steps in the MP trees and 26–35 steps in the randomized trees. Therefore, the number of steps for the character ‘host’ (i.e. changes in host genus or species within the trees) was roughly two to six times less and nonoverlapping for the MP trees as compared to the randomized trees (data not shown). There was also a phylogenetic signal for geographic location. For the combined dataset, the character ‘island’ involved 8–12 steps in the MP trees and 14–25 steps in the randomized trees, and ‘locality’ involved 16–19 steps in the MP trees and 25–35 steps in the randomized trees. Thus, the number of steps for the character ‘geography’ was roughly 1/2 for the MP trees as compared to the randomized trees, and for the character ‘host coral genus’ was always less than one-quarter of the randomized trees (data not shown). 4. DISCUSSION Few host shifts have been reported for marine invertebrates. Some of these adaptations to different hosts or different feeding preferences evolved in different geographic populations within a species (Sanford et al. 2003; Sotka et al. 2003; Sotka 2005). In other cases, there is no genetic evidence for a host shift (Jensen 1997; Trowbridge & Todd 2001), or host races have not yet diverged, and differences may be due to phenotypic plasticity (Duffy 1996). Our data (figure 1) show a host switch from the coral genus Porites to Goniopora and Tubastrea in sympatric Phestilla spp. Host shift seems to be the likely mechanism underlying speciation in this nudibranch genus. The differentiation apparent in the mtDNA sequence data agrees with the data on host use in the field and host preference and specificity of adults and larvae in laboratory experiments (Ritson-Williams et al. 2003). For example, adults of Phestilla sp. 2 ate only polyps of species of Goniopora and its larvae had by far the highest rate of metamorphosis in response to Goniopora species (Ritson-Williams et al. 2003). Likewise, adults of Phestilla sp. 1, P. minor and P. sibogae ate only polyps of Porites spp., preferring the species on which they have been found most frequently in the field, and on which the larvae had the highest rate of metamorphosis (Ritson-Williams et al. 2003). In Phestilla, both host coral and island location show a Proc. R. Soc. B significant phylogenetic signal. However, the number of additional steps between observed and randomized host trees was at least twice (and often much more) than between observed and randomized location trees. Therefore, we hypothesize that speciation in allopatry with subsequent dispersal, which results in currently overlapping ranges, played a smaller role than adaptation to different hosts in driving the speciation of Phestilla. Shifts to a new host may provide release from direct competition for resources and/or enemy-free space (Gross & Price 1988; Lill et al. 2002). Most nudibranchs have diverse chemical and other defence mechanisms and are rarely victims of predation (Todd 1981). However, besides camouflage and hiding, P. sibogae has no recognized defence mechanism (Haramaty 1991). Phestilla sibogae is rarely encountered in the field, but becomes very abundant when coral heads of Porites spp. are kept in predator-free seawater tanks (Haramaty 1991). Further, it has been shown that P. sibogae is heavily preyed upon by reef fish and crustaceans in the field (Gochfeld & Aeby 1997). Harris (1975) suggested that rapid growth and shorter time to sexual reproduction in P. sibogae compared to P. melanobrachia evolved as a result of intense predation pressure. Therefore, looking at our phylogeny of Phestilla, where feeding on the coral genus Porites is ancestral, it appears that the slower growth and longer time to reproduction of P. melanobrachia may be the result of a host switch from Porites to Tubastrea and release from predation pressure. This host switch is remarkable considering the substantial chemical and morphological differences between the solitary cupcoral Tubastrea and the reef building colonial coral Porites. Differences in larval host preference can promote hostrace formation (Linn et al. 2003). There can be differences among batches of larvae in the proportion of metamorphosis of larvae of Phestilla spp. in response to settlement cues (Hadfield 1984; Ritson-Williams et al. 2003), which might suggest a genetic component (Toonen & Pawlik 2001) or plasticity (Hadfield & Strathmann 1996) to the response of settlement cues. Furthermore, larval response to host corals appears to be less specific than adult feeding preference (Ritson-Williams et al. 2003), suggesting that host shifts are most likely to occur at the larval stage. Since the larval stage is also the stage at which dispersal occurs for Phestilla species, it seems most plausible for this to be the stage at which the host shift occurs. However, these hostswitching events must be rare to prevent hybridization between already isolated and specialized populations or new host races or species could not arise. Phestilla sp. 1 and P. minor appear to comprise a cryptic species complex. The short or non-existent planktonic period of these species implies reduced potential for dispersal, which could be one reason for this divergence, because restricted gene flow via restricted dispersal or active habitat choice can lead to local adaptation (Kawecki & Ebert 2004). Phestilla lugubris and P. sibogae cannot be distinguished using COI and 16S sequence comparisons. In Guam, where both the species occur, they are distinguished by morphological differences (i.e. shape of adult head, size of adults and size of egg masses) as well as by the mode of larval development. However, P. sibogae has been observed to occasionally lay eggs with feeding (planktotrophic) larvae on Guam, but when these feeding larvae were raised to adults, they laid eggs containing non-feeding Host shifts in Phestilla spp. A. Faucci and others (lecithotrophic) larvae (V. Paul 2005, personal communication). However, we (MGH) have never seen planktotrophic larvae emerge from egg masses of P. sibogae, despite 37 years of studying this species in Hawaii. The opisthobranch mollusc, Alderia modesta, can switch from producing only lecithotrophic larvae to producing a mix of lecithotrophic and planktotrophic larvae under starvation (Krug 1998). In the same study, Krug (1998) showed that the two reproductive morphs were not distinct based on COI sequences. Furthermore, even though larvae of P. sibogae are lecithotrophic, they retain all feeding structures and are actually able to feed; i.e. the larvae are facultative planktotrophic (Kempf & Hadfield 1985). Thus, the main difference between the larvae is presence or absence of yolk. Consequently, two possibilities emerge: (i) the divergence between P. lugubris and P. sibogae is relatively recent and not yet reflected in the mtDNA sequences or (ii) the two nominal species represent different morphs of the same species and the differences in morphology, adult and egg mass sizes and mode of larval development may be triggered by environmental factors. However, we do not want to base species designation solely on 1019 bp of mtDNA (Rubinoff 2006). Therefore, until additional investigation adds clarity to these questions, it is useful to retain the names P. lugubris for the form with planktotrophic larvae and P. sibogae for the better studied form with lecithotrophic development. In general, planktotrophic development is considered to be ancestral and easier to lose than gain (Strathmann 1978). Strathmann (1978) stated that the potential for gaining a feeding from a non-feeding stage depends on the original mechanism of larval feeding and the degree of reorganization at metamorphosis and adult structure. Some gastropods appear to have more flexibility in reacquiring planktotrophic larvae because encapsulated or lecithotrophic larvae often retain the structures used in feeding (Strathmann 1978). Most phylogenetic studies on the evolution of different larval types confirm the ancestral character state of planktotrophy and its irreversibility once lost (reviewed in Hart 2000). However, Reid (1989) argued that reversal from non-feeding to feeding larvae was most parsimonious for littorinid snails based on ancestral (feeding larvae) and derived (shell form) traits. Likewise, within calyptraeid gastropods, planktotrophy has been regained possibly three times from direct development with nurse eggs (Collin 2004). Some species with nurse eggs appear to have fewer embryonic modifications than those with large yolky eggs, and therefore retain the greatest possibility of subsequent evolution of a different mode of development (Collin 2004). In Phestilla, our results suggest that planktotrophy arose twice (in both P. melanobrachia and P. lugubris/ sibogae) from a lecithotrophic ancestor (figure 1). However, it also appears that lecithotrophic species examined to date (P. sibogae and P. minor) retain the ancestral feeding structures. Furthermore, larvae of P. sibogae have the capacity to feed on phytoplankton (Kempf & Hadfield 1985), and other lecithotrophic species of Phestilla may share this ability. In order to clarify the evolution of larval types within the genus Phestilla, more experiments and observations of all species within the genus and its closest relatives would be needed. Proc. R. Soc. B 7 Our data indicate that a taxonomic revision of the genus Phestilla is needed. The two nominal species Phestilla sp. 1 and P. minor form a polyphyletic clade, where individuals of the two species from different locations and host corals are intermixed (figure 1). Furthermore, P. minor II from Palau on Porites lutea is very distinct from the rest of the specimens. In addition to the molecular divergence, this morphotype also has identifiably different cerata, suggesting that it may be a cryptic species. We call this species P. minor II, but further studies are clearly needed to resolve its taxonomic status. Although Phestilla sp. 1 and P. minor occur in sympatry and are morphologically similar, they feed on different species of host corals. In Guam, Phestilla sp. 1 has been observed to feed on Porites cylindrica and P. rus in the field as well as in the laboratory, whereas P. minor feeds on Porites annae and P. lutea in the field, but only P. annae in the laboratory (Ritson-Williams et al. 2003). Each of the highly divergent sub-clades within the P. minor/sp. 1 clade includes individuals of a species from one location and from a single species of host coral. Even between the least divergent sub-clades, Phestilla sp. 1 on Porites cylindrica and P. minor on Porites lutea from Palau (figure 1), there are fixed genetic differences. In the most extreme case, specimens of Phestilla sp. 1 from Palau collected from two different species of Porites are as divergent (11.1%) as recognized nominal species within the genus of Phestilla (7.3–15.8%) and show a high number of fixed differences (66 sites within COI ). The most parsimonious explanation for this divergence appears to be speciation by host shift. We suggest that each of these distinct lineages probably represents a cryptic species; if true, we would have to rename four of these sub-clades and retain P. minor for just one of them. The relationship between clades of Phestilla sp. 1 and P. minor and their specific hosts may be obscured by the unresolved taxonomy of the host-coral genus Porites. Taxonomy at the species level is still problematic in most coral genera due to plasticity and absence of fixed morphological characters among species and the lack of appropriate molecular markers for fine-scale phylogenies (Forsman et al. 2006). The genus Porites is no exception and its species taxonomy is still largely unresolved (Forsman et al. 2006), which raises the possibility that cryptic species within Porites are confusing the relationship between nudibranchs and their host species. Since larvae of Phestilla are better able to distinguish slight chemical differences among species of Porites than we are, the phylogeny of Phestilla may also shed light on the species classification of Porites corals. Our data comparing COI and 16S sequences, and the presence of a strong phylogenetic signal for the coral host, confirm that the tight link between species of Phestilla and their host coral likely played an important role in the speciation of this genus. To further resolve the relationships within Phestilla, we would need to expand collection sites to encompass the entire range of the genus. Furthermore, it would be most useful to compare the complete phylogeny of Phestilla with a similarly detailed phylogeny of the respective host corals. Rocha et al. (2005) argue that ecological speciation can explain high diversity in marine habitats where physical barriers to gene flow are rare, but many more studies are still needed to understand the mechanisms of speciation in the sea. 8 A. Faucci and others Host shifts in Phestilla spp. We thank G. Paulay and R. Ritson-Williams for providing specimens from Palau and Guam, and the staff of the Greenwood Molecular Biology Facility (University of Hawaii at Manoa) and A. Koga of the University of Hawaii Dell Cluster system (Department of Information and Computer Sciences) for logistic support. Discussions with B. W. Bowen, D. B. Carlon, Z. H. Forsman, S. A. Karl, M. P. Miglietta, J. P. Wares and the members of the Toonen/Bowen Laboratory, as well as valuable comments from three anonymous reviewers greatly improved the manuscript. Funding was provided through NSF Grant no. OCE-9907545 to M. G. Hadfield and an Edmondson Research Grant to A. Faucci. REFERENCES Bergh, L. S. R. 1870 Malacologische untersuchungen. In Reisen im Archipel der Philippinen Wissenschaftliche Resultate (ed. C. Semper), 2, 1–30. Bergh, L. S. R. 1874 Neue Nacktschecken der Sudsee malacologische Untersuchungen 2. Journal des Museum Goddeffroy 4, 91–116. Bergh, L. S. R. 1905 Die Opisthobranchiata der Siboga Expedition. Siboga-Expedite 50, 1–248. Bush, G. L. 1969 Sympatric host race formation and speciation in frrugivorous flies of the genus Rhagoletis (Diptera: Tephritidae). Evolution 23, 237–251. (doi:10. 2307/2406788) Bush, G. L. 1994 Sympatric speciation in animals: new wine in old bottles. Trends Ecol. Evol. 9, 285–288. (doi:10.1016/ 0169-5347(94)90031-0) Carlon, D. B. & Budd, A. F. 2002 Incipient speciation across a depth gradient in a scleractinian coral? Evolution 56, 2227–2242. (doi:10.1554/0014-3820(2002)056[2227: ISAADG]2.0.CO;2) Collin, R. 2004 Phylogenetic effects, the loss of complex characters, and the evolution of development in calyptraeid gastropods. Evolution 58, 1488–1502. (doi:10.1554/ 03-038) Coyne, J. A. & Orr, H. A. 2004 Speciation. Sunderland, MA: Sinauer Associates. Diehl, S. R. & Bush, G. L. 1989 The role of habitat preference in adaptation and speciation. In Speciation and its consequences (ed. D. Otte & J. A. Endler), pp. 345–365. Sunderland, MA: Sinauer Associates. Doebeli, M., Dieckmann, U., Metz, J. A. J. & Tautz, D. 2005 What we have also learned: adaptive speciation is theoretically plausible. Evolution 59, 691–695. (doi:10. 1554/04-154) Duffy, J. E. 1996 Resource-associated population subdivision in a symbiotic coral-reef shrimp. Evolution 50, 360–373. (doi:10.2307/2410807) Feder, J. L. 1998 The apple maggot fly, Rhagoletis pomonella: flies in the face of conventional wisdom about speciation? In Endless forms: species and speciation (ed. D. J. Howard & S. H. Berlocher), pp. 130–144. Oxford, UK: Oxford University Press. Feder, J. L. et al. 2003 Allopatric genetic origins for sympatric host-plant shifts and race formation in Rhagoletis. Proc. Natl Acad. Sci. USA 100, 10 314–10 319. (doi:10.1073/ pnas.1730757100) Folmer, O., Black, M., Hoeh, W., Lutz, R. & Vrijenhoek, R. 1994 DNA primers for amplification of mitochondrial cytochrome c oxidase subunit I from diverse metazoan invertebrates. Mol. Mar. Biol. Biotechnol. 3, 294–299. Forsman, Z. H., Hunter, C. L., Fox, G. E., Wellington, G. M. 2006 Is the ITS region the solution to the ‘species problem’ in corals? Intragenomic variation and alignment permutation in Porites, Siderastrea and outgroup taxa. In Proc. 10th Int. Coral Reef Symp. Okinawa, Japan 28 July– 2 July 2004 (ed. Y. Suzuki, T. Nakamori, M. Hikada, Proc. R. Soc. B H. Kayanne, B. E. Casareto, K. Nadaoka, H. Yamano & M. Tsuchiya), pp, 14–23. Tokyo, Japan: Japanese Coral Reef Society. Gochfeld, D. J. & Aeby, G. S. 1997 Control of populations of the coral-feeding nudibranch Phestilla sibogae by fish and crustacean predators. Mar. Biol. 130, 63–69. (doi:10. 1007/s002270050225) Gross, P. & Price, P. W. 1988 Plant influences of parasitism of two leafminers: a test of enemy-free space. Ecology 69, 1506–1516. (doi:10.2307/1941648) Hadfield, M. G. 1977 Chemical interactions in larval settling of a marine gastropod. In Marine natural products chemistry (ed. D. J. Faulkner & W. H. Fenical), pp. 403–413. New York, NY: Plenum Publishing Company. Hadfield, M. G. 1984 Settlement requirements of molluscan larvae: new data on chemical and genetic roles. Aquaculture 39, 283–298. (doi:10.1016/0044-8486(84)90272-2) Hadfield, M. G. & Koehl, M. A. R. 2004 Rapid behavioral responses of an invertebrate larva to dissolved settlement cue. Biol. Bull. 207, 28–43. Hadfield, M. G. & Pennington, J. T. 1990 The nature of the metamorphic signal and its internal transduction in larvae of the nudibranch Phestilla sibogae. Bull. Mar. Sci. 46, 455–464. Hadfield, M. G. & Strathmann, M. F. 1996 Variability, flexibility and plasticity in life histories of marine invertebrates. Oceanol. Acta 19, 323–334. Haramaty, L. 1991 Reproduction effort in the nudibranch Phestilla sibogae: calorimetric analysis of food and eggs. Pac. Sci. 45, 257–262. Harris, L. G. 1975 Studies on the life history of two coraleating nudibranchs of the genus Phestilla. Biol. Bull. 149, 539–550. Hart, M. W. 2000 Phylogenetic analyses of mode of larval development. Sem. Cell Dev. Bio. 11, 411–418. (doi:10. 1006/scdb.2000.0194) Hart, M. W., Byrne, M. & Johnson, S. L. 2003 Patiriella pseudoexigua (Asteroisea: Asterinidae): a cryptic species complex revealed by molecular and embryological analyses. J. Mar. Biol. Ass. UK 83, 1109–1116. (doi:10. 1017/S002531540300835Xh) Havenhand, J. N. 1991 On the behaviour of opisthobranch larvae. J. Moll. Stud. 57, 119–131. Hay, M. E., Duffy, J. E. & Pfister, C. A. 1987 Chemical defence against different marine herbivores: are amphipods insect equivalents? Ecology 68, 1567–1580. (doi:10. 2307/1939849) Hellberg, M. E. 1998 Sympatric sea shells along the sea’s shore: the geography of speciation in the marine gastropod Tegula. Evolution 52, 1311–1324. (doi:10.2307/2411301) Huelsenbeck, J. P. & Ronquist, F. 2001 MRBAYES: Bayesian inference of phylogenetic trees. Bioinformatics 17, 754–755. (doi:10.1093/bioinformatics/17.8.754) Jensen, K. R. 1997 Evolution of the Sacoglossa (Mollusca, Opisthobranchia) and the ecological associations to their food plants. Evol. Ecol. 11, 301–335. (doi:10.1023/ A:1018468420368) Kawecki, T. J. & Ebert, D. 2004 Conceptual issues in local adaptation. Ecol. Lett. 7, 1225–1241. (doi:10.1111/j.14610248.2004.00684.x) Kempf, S. C. & Hadfield, M. G. 1985 Planktotrophy by the lecithotrophic larvae of a nudibranch, Phestilla sibogae (Gastropoda). Biol. Bull. 169, 119–130. Knowlton, N. 1993 Sibling species in the sea. Annu. Rev. Ecol. Syst. 24, 189–216. (doi:10.1146/annurev.es.24. 110193.001201) Knowlton, N. 2000 Molecular genetic analyses of species boundaries in the sea. Hydrobiologia 420, 73–90. (doi:10. 1023/A:1003933603879) Host shifts in Phestilla spp. A. Faucci and others Krug, P. 1998 Peocilogony in an estuarine opisthobranch: planktotrophy, lecithotrophy, and mixed clutches in a population of the saccoglossan Alderia modesta. Mar. Biol. 132, 483–494. (doi:10.1007/s002270050414) Krug, P. & Manzi, A. E. 1999 Waterborne and surfaceassociated carbohydrates as settlement cues for larvae of the specialist marine herbivore Alderia modesta. Biol. Bull. 197, 94–103. Kumar, S., Tamura, K. & Nei, M. 2004 MEGA3: integrated software for molecular evolutionary genetic analysis and sequence alignment. Brief. Bioinform. 5, 150–163. (doi:10. 1093/bib/5.2.150) Lambert, W. J. & Todd, C. D. 1994 Evidence for a water-borne cue inducing metamorphosis in the dorid nudibranch mollusk Adalaria proxima (Gastropoda:Nudibranchia). Mar. Biol. 120, 267–271. (doi:10.1007/BF00349687) Landry, C., Geyer, L. B., Arakaki, Y., Uehara, T. & Palumbi, S. R. 2003 A recent speciation event in the Indo-West Pacific: rapid evolution of gamete recognition and sperm morphology in cryptic species of sea urchin. Proc. Biol. Sci. 270, 1839–1847. (doi:10.1098/rspb.2003.2395) Lill, J. T., Marquis, R. J. & Ricklefs, R. E. 2002 Host plants influence parasitism of forest caterpillars. Nature 417, 170–173. (doi:10.1038/417170a) Linn, C., Feder, J. L., Nojima, S., Dambroski, H. R., Berlocher, S. H. & Roelofs, W. 2003 Fruit odor discrimination and sympatric host race formation in Rhagoletis. Proc. Natl Acad. Sci. USA 100, 11 490–11 493. (doi:10.1073/pnas. 1635049100) Maddison, D. R. & Maddison, W. P. 2000 MACCLADE 4: analysis of phylogeny and character evolution, version 4.03. Sunderland, MA: Sinauer Associates. Mayr, E. 1976 Evolution and the diversity of life: sympatric speciation. Cambridge, MA: Harvard University Press. Meyer, C. P., Geller, J. B. & Pauley, G. 2005 Fine scale endemism on coral reefs: archipelagic differentiation in Turbinid Gastropods. Evolution 59, 113–125. (doi:10. 1554/04-194) Mokady, O. & Brickner, I. 2001 Host-associated speciation in a coral-inhabiting barnacle. Mol. Biol. Evol. 18, 975–981. Munday, P. L., van Herwerden, L. & Dudgeon, C. L. 2004 Evidence for sympatric speciation by host shift in the sea. Curr. Biol. 14, 1498–1504. (doi:10.1016/j.cub. 2004.08.029) Palumbi, S. R., Martin, A. P., Romano, S., McMillan, W. O., Stice, L. & Grabowski, G 1991 The simple fool’s guide to PCR. Honolulu, HI: Special Publication Department of Zoology, University of Hawaii. Posada, D. & Crandall, K. A. 1998 MODELTEST: testing the model of DNA substitution. Bioinformatics 14, 817–818. (doi:10.1093/bioinformatics/14.9.817) Reid, D. G. 1989 The comparative morphology, phylogeny and evolution of the gastropod family Littorinidae. Phil. Trans. R. Soc. B 324, 1–110. Ritson-Williams, R., Shjegstad, S. & Paul, V. 2003 Host specificity of four corallivorous Phestilla nudibranchs (Gastropoda: Opisthobranchia). Mar. Ecol. Prog. Ser. 255, 207–218. Proc. R. Soc. B 9 Rocha, L. A., Robertson, D. R., Roman, J. & Bowen, B. W. 2005 Ecological speciation in tropical reef fishes. Proc. R. Soc. B 272, 573–579. (doi:10.1098/2004.3005) Rubinoff, D. 2006 Utility of mitochondrial DNA barcodes in species conservation. Cons. Biol. 20, 1026–1033. Rudman, W. B. 1981 Further studies on the anatomy and ecology of opisthobranch mollusks feeding on the scleractinian coral Porites. Zool. J. Linn. Soc. 71, 373–412. Rudman, W. B. 1998 6. Subclass Opisthobranchia. general introduction: history of discovery, morphology & physiology, natural history. Fauna of Australia, 5. Mollusca. Canberra, Australia: ABRS. Sanford, E., Roth, M. S., Johns, G. C., Wares, J. P. & Somero, G. N. 2003 Local selection and latitudinal variation in a marine predator–prey interaction. Science 300, 1135–1137. (doi:10.1126/science.1083437) Sotka, E. E. 2005 Local adaptation in host use among marine invertebrates. Ecol. Lett. 8, 448–459. (doi:10.1111/j.14610248.2004.00719.x) Sotka, E. E., Wares, J. P. & Hay, M. E. 2003 Geographic variation in feeding preferences for chemically defended seaweeds. Evolution 57, 2262–2276. (doi:10.1554/03-076) Stanhope, M. J., Connelly, M. M. & Hartwick, B. 1992 Evolution of a crustacean chemical communication channel: bevahioral and ecological genetic evidence for a habitat-modified, race-specific pheromone. J. Chem. Ecol. 18, 1871–1887. Strathmann, R. R. 1978 The evolution and loss of feeding larval stages of marine invertebrates. Evolution 32, 894–906. (doi:10.2307/2407502) Swofford, D. L. 2002 PAUP . Phylogenetic analysis using parsimony ( and other methods). Version 4.0b10. Sunderland, MA: Sinauer Associates. Todd, C. D. 1981 The ecology of nudibranch molluscs. Oceanogr. Mar. Biol. Annu. Rev. 19, 141–234. Todd, C. D., Lambert, W. J. & Davies, J. 2001 Some perspectives on the biology and ecology of nudibranch mollusks: generalisations and variations on the theme prove the rule. Boll. Mala. 37, 105–120. Toonen, R. J. & Pawlik, J. R. 2001 Foundations of gregariousness: a dispersal polymorphism among the planktonic larvae of a marine invertebrate. Evolution 55, 2439–2454. (doi:10.1554/0014-3820(2001)055[2439: FOGADP]2.0.CO;2) Trowbridge, C. D. & Todd, C. D. 2001 Host-plant change in marine specialist herbivores: ascoglossan sea slugs on introduced macroalgae. Ecol. Monogr. 71, 219–243. (doi:10.2307/2657217) Via, S. 2001 Sympatric speciation in animals: the ugly duckling grows up. Trends Ecol. Evol. 7, 381–390. (doi:10. 1016/S0169-5347(01)02188-7) Wollscheid, E., Boore, J., Brown, W. & Waegele, H. 2001 The phylogeny of Nudibranchia (Opisthobranchia, Gastropoda, Mollusca) reconstructed by three molecular markers. Org. Divers. Evol. 1, 241–256. (doi:10.1078/ 1439-6092-00022)