This paper reviews progress in developmental biology and phylogeny of the Nemertea, a common but poorly studied spiralian taxon of considerable ecological and evolutionary significance. Analyses of reproductive biology (including calcium dynamics during fertilization and oocyte maturation), larval morphology and development and developmental genetics have significantly extended our knowledge of spiralian developmental biology. Developmental genetics studies have in addition provided characters useful for reconstructing metazoan phylogeny. Reinvestigation of the cell lineage of Cerebratulus lacteus using fluorescent tracers revealed that endomesoderm forms from the 4d cell as in other spiralians and that ectomesoderm is derived from the 3a and 3b cells as in annelids, echiurans and molluscs. Studies examining blastomere specification show that cell fates are established precociously in direct developers and later in indirect developers. Morphological characters used to estimate the phylogenetic position of nemerteans are critically re-evaluated, and cladistic analyses of morphology reveal that conflicting hypotheses of nemertean relationships result because of different provisional homology statements. Analyses that include disputed homology statements (1, gliointerstitial cell system 2, coelomic circulatory system) suggest that nemerteans form the sister taxon to the coelomate spiralian taxa rather than the sister taxon to Platyhelminthes. Analyses of small subunit rRNA (18S rDNA) sequences alone or in combination with morphological characters support the inclusion of the nemerteans in a spiralian coelomate clade nested within a more inclusive lophotrochozoan clade. Ongoing evaluation of nemertean relationships with mitochondrial gene rearrangements and other molecular characters is discussed.
Phylum Nemertea contains about 1,150 species (Gibson, 1995) of unsegmented worms that possess an eversible proboscis contained in a fluid-filled cavity or rhynchocoel. The proboscis is a synapomorphy of the taxon and is used primarily in prey capture. These animals range in length from a few millimeters to about 30 m. The phylum has a global distribution and includes marine, freshwater and terrestrial species. Most nemerteans are epi- and endobenthic inhabitants of marine littoral and sublittoral zones, but a number of marine pelagic species also occur. These worms are primarily predators feeding on members of a variety of taxa, including annelids, arthropods, and molluscs (McDermott and Roe, 1985). Nemerteans have been identified as important predators in marine soft-bottom communities (e.g.,Nordhausen, 1988; Ambrose, 1991; Thiel, 1998), and recent investigations are leading to an evaluation of their role in structuring these communities (see Thiel and Kruse, 2001). The economic importance of certain nemerteans has been long recognized. For example, the impact on Dungeness crab fisheries by the ectocommensal crab egg predator Carcinonemertes has been well documented and the ecology of these economically important species continues to be an active area of investigation (e.g., see Wickham, 1986; Kuris, 1993; Messick, 1998). Studies also have indicated that Cerebratulus lacteus may have a negative impact on soft-shell clam (Mya arenaria) fisheries in Atlantic Canada (see Bourque et al., 1999).
Historically, nemerteans have been considered either acoelomate spiralians most closely related to flatworms (Platyhelminthes) or coelomate spiralians. Thus, nemerteans have played a pivotal role in traditional theories of body-plan evolution, prompting recent evaluations of their phylogenetic position (e.g.,Turbeville and Ruppert, 1985; Turbeville et al., 1992), embryology (e.g.,Henry and Martindale, 1998) and developmental genetics (Loosli et al., 1996; Kmita-Cunisse et al., 1998). The phylogenetic position of nemerteans remains controversial.
Despite the common occurrence of these animals and their ecological and evolutionary significance, nemerteans have been poorly studied in comparison to other spiralian protostome taxa (see Norenburg and Roe, 1998). Yet, during the past 20 yr, significant advances in nemertean biology have been made, and this is reflected in part by 5 conferences on nemertean biology held since 1983. Contributions to two active, and to an extent complementary areas of nemertean research will be addressed in this review: 1) development, and 2) phylogeny.
Historically, the focus of most research on developmental biology of nemerteans was limited to descriptive and experimental embryology and larval development (see Friedrich, 1979; Gibson, 1998). Recent investigations have provided further insight into these as well as more diverse aspects of nemertean development, including cell biology of gametogenesis, body-plan patterning, and developmental genetics.
Norenburg and Roe (1998) recently presented the first modern findings on reproductive biology of the poorly studied pelagic nemerteans. Analyses of the ovaries in polystiliferous species strongly suggest that accessory ovarian cells translocate yolk precursors to the oocytes by way of cytoplasmic bridges. Such an association of cells and mechanism of precursor translocation have not been reported from nemerteans previously. Additional observations indicate that these pelagic species produce relatively few gametes, spawn in close proximity, exhibit iteroparous reproduction and are long-lived (Norenburg and Roe, 1998).
Stricker and Folsom (1998) conducted a comparative analysis of nemertean spermatozoa that included correlation of sperm morphology with modes of fertilization. Their findings suggest that rather than being exclusively associated with internal fertilization, elongate sperm of some nemertean species may represent an adaptation to penetrating vitelline envelopes of oocytes.
Other recent studies of nemertean reproduction include the chemical nature of oocyte maturation and elucidation of calcium dynamics during fertilization. In Cerebratulus lacteus fertilization effects a cortical flash (i.e., rapid increase) of free calcium within the oocyte that is followed by oscillating waves of calcium resulting from periodic release of internal calcium stores (Stricker, 1996). The calcium transients are presumably necessary to inhibit polyspermy. Stricker (1997) also provided evidence supporting the hypothesis that calcium oscillations in the oocyte can be elicited by a soluble sperm factor that need not bind to the oocyte plasma membrane. An analysis of oocyte maturation triggers in C. lacteus suggested that oocyte maturation involves either an external calcium influx or a serotonin (5-hydroxytryptamine) initiated signaling cascade (Stricker and Smythe, 2000). Stricker and Smythe (2001) also showed that in mature oocytes of C. lacteus and Micrura alaskensis, serotonin can lead to increased levels of cAMP, resulting in stimulation rather than inhibition of oocyte maturation. That cAMP stimulates (GVBD) in nemerteans is congruent with findings for some other nonvertebrate animals (e.g., cnidarians and some echinoderms) and together these results should contribute to the refinement of models of oocyte maturation in animals (see Stricker and Smythe, 2001).
Classical studies revealed that nemertean embryos exhibit spiral cleavage and that the mouth forms at or near the site of the blastopore, but aspects of cell lineage, especially origin of the mesoderm, proved problematic (see Friedrich, 1979). Henry and Martindale (1998) reinvestigated the cell lineage of the heteronemertean Cerebratulus lacteus using a fluorescent tracer and clarified the origin of mesoderm in this nemertean (Fig. 1). Their analysis revealed that ectomesoderm is derived from 3a and 3b blastomeres, and that endomesoderm is derived from the 4d cell. Mesodermal bands were also observed in the pilidium larva, suggesting that the 4d cell gives rise to teloblasts as in coelomate spiralians (see Nielsen, 1995). The origin of ectomesoderm from the 3a and 3b cells is also of potential phylogenetic significance, as these same two cells give rise to some of the ectomesoderm in annelids, echiurans and molluscs, whereas in polyclad flatworms ectomesoderm is derived from the 2b cell (see Boyer and Henry, 1998; Boyer et al., 1998). In acoel flatworms, which exhibit duet spiral cleavage, no ectomesoderm is produced (Henry et al., 2000). Analysis of cell lineage in additional spiralian taxa is warranted to further evaluate the phylogenetic utility of this embryological character.
Martindale and Henry (1995) investigated developmental potential of isolated blastomeres in direct- and indirect-developing nemerteans and observed variation in their regulative capacities. In the indirect-developing C. lacteus, blastomeres isolated at the two-cell stage developed into normal half-sized pilidium larvae, indicating regulative capacity. Blastomeres separated at the four-cell stage give rise to incomplete larvae, although limited regulation does occur. In the direct-developing Nemertopsis bivittata development of isolated two-celled stage blastomeres resulted in half “larvae” possessing a single ocellus instead of two. Thus, regulative capacity in this species is limited compared to C. lacteus. Development of C. lacteus isolated half-embryos, formed by bisecting four-celled embryos along the first cleavage plane, also revealed that axial properties (animal-vegetal and dorsoventral axes), which in C. lacteus are established before the first cleavage division (Henry and Martindale, 1996), are not fixed, but rather can be reorganized during embryonic regulation (Henry and Martindale, 1997). These findings for nemerteans are concordant with those for other spiralians, further indicating that early blastomere fate determination is not an absolute correlate of spiral cleavage (see Boyer and Henry, 1998).
The organization and development of nemertean larvae have been the focus of several studies in recent years. Lacalli and West (1985) and Hay-Schmidt (1990) elucidated the organization of the nervous system of the heteronemertean pilidium larvae and found that in contrast to other spiralian larvae, the pilidium lacks an apical ganglion. The results of the cell lineage-tracing study of Henry and Martindale (1998) are also in accord with this observation. Hay-Schmidt (1990, 2000) revealed further that in the pilidium larva, most serotonergic cell bodies are associated with nerves innervating the ciliary bands and that these neurons are not retained by the juvenile. The juvenile, which forms within the larval ectoderm, independently develops a serotonergic system.
Analyses of larval development in the hoplonemertean Tetrastemma candidium revealed that larval ectoderm is replaced by definitive ectoderm and that the apical plate degenerates, leading to speculation that hoplonemertean larvae metamorphose, and thus that hoplonemerteans exhibit indirect development rather than direct development as traditionally held (Maslakova and Malakhov, 1999). Recently, Maslakova and Norenburg (2001; S. A. Maslakova, personal communication), examined early larval development in Carinoma tremaphoros and found that the morphology, position and cell lineage of larval ectoderm forming a preoral belt supports its homology to the prototroch of trochophore larvae. These findings, along with preliminary observations on hoplonemerteans, suggest that a trochophore larva may be plesiomorphic for nemerteans (see below).
Tarpin et al. (1999) artifically constructed homeotic Lineus viridis by grafting an ocellar (eye) region from a donor individual in place of the post ocellar region of a recipient individual. The manipulation resulted in a worm lacking a postocellar region but instead, possessing two ocellar regions anterior to the cerebral ganglia. During a 14-wk interval, a new postocellar region was regenerated while the anteriormost ocellar region degenerated, restoring the original anterior- posterior morphological pattern. The results indicate that regression or maintenance of ocelli in L. viridis depends on their position along the anterior-posterior axis. This innovative study sets the stage for analyses of genetic mechanisms responsible for reverse homeosis, which will contribute additional insight into bilaterian body-plan patterning.
Developmental genetics: Hox and paired-class genes
Kmita-Cunisse et al. (1998) characterized Hox genes in the nemertean Lineus sanguineus to enhance understanding of Hox gene evolution. The investigation revealed a single cluster of Hox genes consisting of minimally two anterior-class genes (LsHox1, LsHox3), three middle-class genes (Ls Hox 4, Ls Hox 6, Ls Hox 7) and one posterior-class gene (Ls Hox 9 = Abd-B). Two additional homeobox genes also were identified: Lscdx (caudal-type gene) and an NK-1-type gene (LsNK). Such findings are contributing to an understanding of homeobox gene evolution and have provided data for inferring metazoan phylogeny (see Peterson and Eernisse, 2001).
The paired-class gene, Pax-6, also has been identified in L. sanguineus, a nemertean possessing numerous eyes (=ocelli; Loosli et al., 1996). Pax-6 occurs in a broad range of metazoan taxa, and experiments strongly suggest that it is the principal control gene for eye morphogenesis (Gehring, 2002).
PHYLOGENETIC POSITION: MORPHOLOGICAL EVIDENCE
In the English-language zoological literature nemerteans are treated conventionally as “primitive” acoelomate bilaterians that are most closely related to flatworms (Platyhelminthes), and, consequently, they are sometimes selected as a study organism to gain insight into the “early” evolution of genes and developmental programs (e.g., see Loosli et al., 1996; Kmita-Cunisse et al., 1998). However, the phylogenetic position of nemerteans remains controversial, and correct evolutionary interpretation of developmental genetics studies will require reliable placement of nemerteans among the Bilateria.
In this section I critically assess characters used to estimate nemertean relationships and consider alternative phylogenies. Morphological features cited in precladistic analyses suggesting a close relationship of nemerteans and flatworms include multiciliated, glandular epidermis, rod-shaped secretory bodies or rhabdites, frontal glands or organs, protonephridia, and acoelomate body organization (e.g.,Bürger, 1897–1907; Hyman, 1951). However, a comparative evaluation does not support the phylogenetic potential of these features.
Multicilated, glandular epidermis
Comparative studies indicate no unique similarities of the nemertean and platyhelminth epidermis (Norenburg, 1985; Turbeville, 1991). Multiciliated epidermal cells and epidermal gland cells are not exclusive to nemerteans and platyhelminths. They are found in many animal taxa including Ctenophora (Hernandez-Nicaise, 1991), Echiura (Pilger, 1993), Sipuncula, (Rice, 1993) Annelida (Gardiner, 1992) Mollusca (e.g.,Scheltema et al., 1994; Haszprunar and Schaeffer, 1997) among others. Thus, these features are likely symplesiomorphies of nemerteans and flatworms.
The epidermis of some nemerteans contains discrete rod-shaped inclusions that were traditionally interpreted as homologues of platyhelminth rhabdites (e.g.,Bürger, 1897; Hyman, 1951; Gontcharoff and Lechenault, 1966). Ultrastructural studies of these secretory bodies reveal, however, that they have a substructure unlike rhabdites (Norenburg, 1985; Stricker and Cavey, 1988; Turbeville, 1991). Specifically, they lack the characteristic lamellate cortex of true rhabdites (Smith et al., 1982). The proboscis epithelium of several palaeo- and hoplonemerteans contains cells that produce structures that superficially resemble cnidocysts (Turbeville, 1991; Montalvo et al., 1998). They have been called pseudocnids or rhabdites in earlier literature and considered platyhelminth rhabdite homologues (e.g.,Hyman, 1959; Gontcharoff, 1957; Ling, 1971). These “pseudocnids” are rod-shaped or conical and have an electron lucent cortex, an electron dense medulla and a filament-like core. The cortex and medulla are distinctly different from those of turbellarian rhabdites (cf., figures in Turbeville  and Smith et al. ). Evidence of core eversion is conflicting (cf.,Gerner  and Montalvo et al. ), but these structures are likely involved in gripping or puncturing prey (Jennings and Gibson, 1969). The organization of these secretory bodies suggests that they are unique to nemerteans, and their similarity to cnidarian cnidocysts and turbellarian rhabdites is superficial at best.
Frontal glands have been cited in the classical literature as probable homologues of Platyhelminthes and nemerteans (e.g.,Bürger, 1897–1907; Hyman, 1951). These are simple structures consisting of an anterior concentration of gland cells opening terminally by way of pores. In nemerteans the frontal glands typically open into a protrusible structure consisting of ciliated sensory cells called a frontal organ (Bürger, 1897–1907). Anteriorly situated glandular complexes are not restricted to nemerteans and flatworms, however. For example, they occur in some meiofaunal annelids (e.g., “prostomial adhesive glands,” Kristensen and Nilonen, 1982), the meiofaunal (annelid?) worm Labotocerebrum (Rieger, 1980, 1981), and in entoproct larvae (see Haszprunar, 1996). Complicating evaluation of this character is the fact that they are simple in organization and vary among taxa. Even within the Platyhelminthes, the organization and content of the glands and the position of gland cell emergence is variable (Rieger et al., 1991; Haszprunar, 1996). Detailed histochemical and ultrastructural analyses of these structures in nemerteans are desirable to further assess their phylogenetic potential, but it is important to note that when coded and included in a cladistic analysis, this character does not support monophyly of Nemertea and Platyhelminthes (see Haszprunar, 1996 and Zrzavy et al., 1998).
The anteriorly situated, eversible ciliated pits (frontal organs) resemble nuchal organs of polychaete annelids. These structures in polychaetes consist minimally of ciliated sensory cells and ciliary support cells. In some species unciliated support cells and retractor muscles are also present (see Verger-Bocquet, 1992). Although limited available evidence suggests that similarity is superficial, nemertean frontal organs and annelid nuchal organs have been considered homologous by some (e.g.,Zrzavy et al., 1998). Further detailed comparison of these structures is warranted.
Nemerteans possess protonephridia consisting of multiciliated terminal cells, a ciliated nephridial duct and a nephridiopore opening at the epidermis (Bartolomaeus and Ax, 1992; Ruppert and Smith, 1988). The terminal cells of nemertean nephridia are situated in the body wall often in close proximity to the walls of the lateral vessels of the circulatory system (see below). In Tubulanus annulatus (Jespersen and Lützen, 1987) and Tubulanuscf.T. pellucidus (Turbeville, 1991) the terminal cells are separated at points along the vessel only by extracellular matrix (ECM), as the vessel lining is fenestrated. In the acoelomate platyhelminths flame cells are situated in the intercellular space of the body wall. In adult coelomate spiralians with protonephridia the terminal cells typically possess a single cilium (solenocytes) and extend into the coelomic cavity (e.g., polychaete annelids, Smith, 1992), but there are exceptions. For example, in annelids with reduced coeloms terminal cells are multiciliated and are in open communication with the coelom (e.g.,Westheide, 1985, 1986). Partially acoelomate annelids also possess multiciliated terminal cells, and they are necessarily situated in the intercellular space of the peritoneal cells (=coelenchyme) that occlude the coelom in certain regions of the body (e.g.,Dinophilus, see Smith, 1992).
Although long considered a character informative for linking nemerteans and platyhelminths, rigorous comparative analyses of protonephridial morphology have revealed no synapomorphies of nemertean and platyhelminth nephridia (see Bartolomaeus, 1988; Bartolomaeus and Ax, 1992; Turbeville, 1991). It is also noteworthy that arguments against homology of the nemertean circulatory system and the spiralian coelom cite the non-correspondence in position and type of nephridia in nemerteans and spiralian coelomates (e.g.,Bartolomaeus, 1988; Nielsen, 2001; see below).
The organization of the body compartment between the epidermis and gut in nemerteans and flatworms also does not reveal a uniquely similar pattern. Even within the Platyhelminthes organization of this tissue compartment is variable. For example, in acoel flatworms the connective tissue (parenchyma) of this space is entirely cellular, with the cells situated between the body-wall musculature and the digestive syncytium, whereas in platyhelminths such as polyclads and triclads the parenchyma is markedly different, being comprised of extracellular matrix (ECM) surrounding the muscles, epidermal replacement cells, fixed parenchymal cells, neoblasts (totipotent cells), and pigment cells (Rieger et al., 1991). Monophyly of the Platyhelminthes is in dispute based on morphology and molecular data, thus the plesiomorphic condition of parenchymal organization is uncertain based on available evidence.
In nemerteans the compartment between the epidermis and the gut is occupied primarily by a well-developed musculature embedded in the noncellular component of the connective tissue (ECM, see Turbeville, 1991). Several cell types have been described for nemertean connective tissue but the function of few has been substantiated (Turbeville and Ruppert, 1985; Turbeville, 1991). The overall pattern of muscle and connective tissue organization corresponds most closely to that found in larger flatworms such as polyclads and triclads, but a similar organization consisting of a body-wall musculature embedded in an ECM is widespread among the Spiralia, occurring for example in, sipunculans (Rice, 1993) echiurans (Pilger, 1993) and many annelids (Bartolomaeus, 1994; Gardiner, 1992; Jamieson, 1992). Thus, comparative analysis suggests that this character is not a synapomorphy of nemerteans and flatworms.
Informative morphological characters
None of the above characters traditionally used to evaluate the phylogenetic position of nemerteans can be interpreted as a provisional shared-derived homologue of nemerteans and platyhelminths or nemerteans and other spiralian taxa. However, comparative ultrastructural analyses have revealed two characters that are potentially informative for placing nemerteans among the Spiralia: 1) gliointerstitial cell system and 2) coelomic circulatory system.
A system of granule-containing cells closely associated with nervous systems and ECM occurs in all nemertean higher taxa (Turbeville and Ruppert, 1985; Turbeville, 1991). The position and composition of these cells suggests that they are homologous to the gliointerstitial cell system of molluscs (Nicaise, 1973; Turbeville and Ruppert, 1985; see also Haszprunar, 1996), sipunculans (Rice, 1993), annelids (Gardiner, 1992; Jamieson, 1992) echiurans (Pilger, 1993). Such cells are absent in Platyhelminthes, deuterostomes and radiate outgroups (see Haszprunar, 1996 for an earlier consideration of this character). Thus, this feature is interpreted as a provisional synapomorphy of nemerteans and spiralian coelomates. Proposed functions of these granule-containing cells include formation and regulation of extracellular matrix (Rieger, 1981) and insulation of nerves from excess acetylcholine (Nicaise, 1973).
In contrast to platyhelminths, nemerteans possess a circulatory system, comprised of fluid-filled, cell-lined channels situated between the gut and body-wall musculature. In basic organization the system consists of two lateral vessels situated between the digestive tract and body wall joined by anterior and posterior connectives, forming a loop. A common modification occurring in many hetero- and hoplonemerteans is a dorsal vessel and transverse connectives. Nemertean vessels are similar in anatomy and ontogeny to coeloms of spiralian coelomates; the major vessels 1) are situated laterally between the gut and body wall, 2) are lined by an epithelium derived from mesoderm (mesothelium), and 3) form as a split in a band of mesodermal cells3 (see Turbeville, 1986, 1991; Turbeville and Ruppert, 1985; Jespersen and Lützen, 1988). The similarity test (a priori test of homology) supports the hypothesis that nemertean vessels are coelom homologues.4
However, some systematists interpret the nemertean circulatory system as an autapomorphy of the taxon rather than as a coelom homologue, based on the lack of complete correspondence in morphology of nemertean vessels to coeloms of polychaete annelids, echiurans and sipunculans. For example, Nielsen (1995, 2001) cites the nonsegmental nature of these cavities in nemerteans and the lack of associated metanephridia or otherwise open nephridial connections as evidence against homology. Bartolomaeus (1994) notes that, unlike the situation in nemerteans, in annelids, echiurans and sipunculans, the mesodermal bands give rise to the coelomic lining as well as to the body musculature, which may be myoepithelial, pseudostratified or stratified (Bartolomaeus, 1994). In echiurans, sipunculans and in most polychaete annelids at least portions of the coelomic lining form the somatic and visceral (=splanchnic) musculature. A noncontractile peritoneum overlying the musculature occurs in echiurans (Pilger, 1993) sipunculans (Rice, 1993) clitellate annelids (e.g.,Fernández et al., 1992) and in certain polychaete annelids (e.g., capitellids; see Bartolomaeus, 1994 for a survey). When present, the annelid peritoneum either rests directly on the muscle cells or is separated from them by a layer of extracellular matrix (see Bartolomaeus, 1994).
In the interstitial nemertean Cephalothrix sp. the lining of the anteriormost region of the circulatory system (anterior lacuna) exhibits characteristics of a myoepithelium and like myoepithelia in many spiralian coelomates, it forms a portion the body-wall musculature (Fig. 2; unpublished data, J.M.T.). In the palaeonemerteans Procephalothrix spiralis (see Fig. 5F in Turbeville and Ruppert, 1985) and Cephalothrixcf.C. rufifrons (unpublished data, J.M.T.) the mesothelium (peritoneum) of the vessel wall adjacent to the gut directly overlies muscle cell processes containing either circular or longitudinally oriented myofilaments. There is no intervening extracellular matrix, and the underlying muscles appear to form part of a weak visceral musculature (unpublished data, J.M.T.). This condition (i.e., a noncontractile peritoneum directly overlying muscles) is similar to that found in some polychaete annelids (Gardiner, 1992; Fransen, 1980; Bartolomaeus, 1994). In hetero- and hoplonemerteans investigated there is an ECM layer situated between the vessel lining mesothelium and underlying muscle, which may be part of the general body-wall musculature or an intrinsic musculature (Jespersen and Lützen, 1988; Turbeville, 1986, 1991; Turbeville and Ruppert, 1985). Such an intrinsic musculature disassociated from the major body-wall muscles parallels the condition characteristic of coelomic channels in some leeches (Fernández et al., 1992; unpublished data, J.M.T.).
Provisional results for the cephalothricid nemerteans (Turbeville and Ruppert, 1985; this paper) suggest that at least some of their body musculature may be derived from the vessel rudiment, but analyses of vessel ontogeny and supplemental investigation of vessel anatomy in these species are needed to further clarify the situation. The plesiomorphic condition for nemertean vessel organization cannot be inferred without a phylogenetic framework for the group. However, if the cephalothricid condition (vessel lining forming a portion of the musculature) turns out to be plesiomorphic, this would further support the hypothesis that nemertean vessels are coelom homologues and suggest that the condition found in other nemertean taxa (noncontractile lining cells resting on ECM) is derived.
Contrasting interpretations of morphology as outlined in the foregoing discussion necessarily lead to differing provisional homology statements and illustrate the basis of conflicting hypotheses of nemerteans and, by extension, metazoan phylogeny. Because of the relative paucity of recognized informative morphological characters for inferring metazoan phylogeny, alternative interpretations of even 1 or 2 characters can lead to alternative placement of taxa in a phylogeny. The effect of conflicting a priori character interpretations on the placement of nemerteans can be illustrated by considering the following example. Nielsen's (2001) analysis of a data set containing 64 morphological characters places nemerteans as the sister taxon to the Platyhelminthes, forming the basal subclade of the Spiralia (Fig. 3A). This hypothesis is supported by a single unambiguous synapomorphy, reduced hyposphere of polyclad flatworm larvae and pilidium larvae of heteronemerteans. A priori interpretation of this character as a homologue assumes that the ground-pattern of the nemerteans and flatworms exhibits indirect development and secondly, that the larvae of polyclad flatworms and heteronemerteans are modified trochophores (Nielsen, 2001). Although platyhelminth phylogeny is unsettled, none of the published alternative phylogenies (e.g.,Ax, 1995; Haszprunar, 1996; Ruiz-Trillo et al., 1999; Peterson and Eernisse, 2001) supports indirect development as the plesiomorphic condition. A preliminary phylogenetic analysis of nemerteans suggest that the pilidium larva is a derived feature of a Heteronemertea + Hubrechtella clade (Norenburg, 1993; J. L. Norenburg, personal communication) rather than a primitive feature of the phylum. Furthermore, data supporting homology of the trochophore prototroch and the ciliary bands of the pilidium larvae are ambiguous (Henry and Martindale, 1998), but it is important to note that Maslakova and Norenburg (2001) recently provided preliminary evidence from a “direct developing” species suggesting that a trochophore may be plesiomorphic for nemerteans (see page 694). However, even if Nielsen's hypothesis is accepted, the inclusion of just two potential homologues, schizocoelous coelom and gliointerstitial cell system in his original matrix results in radically different placement of the nemerteans, namely as the sister taxon to the coelomate spiralians (Fig. 3B). The estimated phylogeny suggests that the larval similarities are homoplastic. Other cladistic analyses of morphology that included one or both of these characters are congruent with the result presented herein (Eernisse et al., 1992; Zrzavy et al., 1998; Peterson and Eernisse, 2001).
PHYLOGENETIC POSITION: MOLECULAR EVIDENCE
Analyses of 18S rRNA genes place nemerteans within the “Lophotrochozoa” more closely related to spiralian coelomates than to the acoelomate platyhelminths. Original analyses of partial 18S rRNA sequences that included a limited number of taxa supported inclusion of nemerteans in a coelomate spiralian clade (Turbeville et al., 1992). Subsequent investigations using the entire 18S rDNA sequence and denser taxon sampling confirmed the earlier result (e.g.,Winnepenninckx et al., 1995, 1996; Carranza et al., 1997), although when multiple nemertean taxa were used Nemertea was not inferred as monophyletic (Giribet et al., 2000; unpublished data). In a simultaneous analysis of 18S rDNA and morphological characters, the nemerteans are monophyletic and form the sister taxon to a clade of spiralian coelomates nested within the “Lophotrochozoa” (=“Trochozoa” Giribet et al., 2000; see also Peterson and Eernisse, 2001; Fig. 4).
Analyses of partial elongation factor 1-alpha (EF1a) sequences suggest that nemerteans are closely related to molluscs. However, given that taxon sampling was limited (e.g., no flatworms included) and only a small fragment of the gene was analyzed (McHugh, 1997), this result must be regarded as preliminary.
The mitochondrial genome also holds promise for elucidating relationships of animal phyla. The mitochondrial genome is a small circular genome about 15–20 kb in the length. The arrangement of 37 genes on the genome is, with some exceptions, conserved within phyla but varies between phyla. Uniquely shared gene arrangements are considered to be strong indicators of genealogical descent, as the number of possible rearrangements is so great that homoplastic similarity is improbable (Boore and Brown, 1998). A number of recent studies indicate that gene rearrangements contain phylogenetically informative characters (e.g.,Boore et al., 1995; Blanchette et al., 1999; Boore and Brown, 2000). However, hypervariability does occur in some taxa (e.g., Nematoda, Platyhelminthes; Mollusca); thus as with sequence data, sampling multiple taxa will be essential to obtain accurate representation of the phylogenetic signal (see Le et al., 2000; Curole and Kocher, 1999).
A preliminary comparison of metazoan mitochondrial gene orders including that of the nemertean Cerebratulus lacteus (Fig. 5; Turbeville, in preparation) reveals a provisional synapomorphy of the protostomes (Fig. 5; e.g.,D-A8). The arrangement ND2-COI is shared by brachiopods and spiralian taxa (personal observations), but additional ingroup and outgroup data are needed to more critically evaluate its phylogenetic potential. No arrangement synapomorphies of the nemertean and the parasitic, cestode platyhelminths Echinococcus (GenBank) and Hymenolepis (see von Nickisch-Rosenegk et al., 2001) have been recognized (personal observations), but it will be essential to obtain gene arrangement data from a broad sampling of platyhelminth taxa to allow for a meaningful comparison (Turbeville, in preparation).
Because of character non-independence, standard coding can be problematic (see Swofford et al., 1996), but a promising method for comparative analysis of gene arrangements that helps obviate this pitfall in coding is the minimum breakpoint method recently described by Sankoff and Blanchette (1998) and Blanchette et al. (1999). This method calculates the minimum number of breakpoints necessary to relate a set of gene orders. A breakpoint is equivalent to nonadjacency of identical genes in separate genomes. Preliminary breakpoint analyses of protein-encoding and rRNA genes of a data set including lophotrochozoan, ecdysozoan and deuterostome taxa support the inclusion of the nemertean in a spiralian clade, but monophyly of coelomate spiralians is not supported (Fig. 6). Although data from a denser sample of nemertean and flatworm taxa will be required for a rigorous assessment of the utility of this approach for phylogeny reconstruction, the initial results are encouraging. As currently implemented (exhaustive search), only small data sets can be analyzed, but software allowing more rapid breakpoint analysis of larger data sets is currently under development (Moret et al., 2001) and will provide a powerful means of gene arrangement data analysis.
In addition to gene arrangement comparisons, the nucleotide sequences of the entire mitochondrial genome and the amino acid sequences of the protein-encoding genes will be used to reconstruct phylogeny. We also are currently assessing the phylogenetic potential of ribosomal proteins, and analyses of these nuclear, protein-encoding genes in combination with morphological and available molecular characters will allow a more comprehensive estimate of nemertean and metazoan phylogeny.
Recent years have witnessed a growing interest in the anatomy, embryology, larval development, genomes, and phylogenetic position of nemerteans, and this trend is leading to a substantial increase in our knowledge of the biology of these important animals. Some recent investigations of nemertean development have contributed informative characters for phylogenetic analysis (page 695), whereas others have provided insight into developmental mechanisms and patterns (pages 693–695). Analyses of nemertean relationships have been in part contradictory, but the weight of the present morphological and molecular evidence analyzed separately or simultaneously supports the inclusion of nemerteans in a coelomate-spiralian subclade nested within the “Lophotrochozoa.” Character support for this placement is weak, given the relative deficit of phylogenetically informative characters available for analysis; however, assessment of the phylogenetic utility of additional molecular and morphological characters is ongoing and should result in a solid inference of nemertean relationships, which will be crucial for understanding body-plan evolution within the “Lophotrochozoa.”
From the Symposium Lesser-Known Protostome Taxa: Evolution, Development, and Ecology presented at the Annual Meeting of the Society for Integrative and Comparative Biology, 3–7 January 2001, at Chicago, Illinois.
Present address: Department of Biology, Virginia Commonwealth University, Richmond, Virginia 23284 E-mail: firstname.lastname@example.org
Whether the mesodermal bands (vessel rudiments) that form the circulatory system are derived from the 4d cell or other sources has not yet been established.
I would like to thank Jim Garey for organizing the symposium and the opportunity to participate. I am grateful to Drs. Thomas Stach, J. L. Norenburg, Alan Kohn and an anonymous reviewer for critically reviewing the manuscript and to Seth Tyler for providing specimens of Cephalothrix sp. This investigation was supported in part by NSF grant DEB-0089654.