RESULTS AND DISCUSSION

Thirteen primers yielded a total of 154 consistently well-amplified DNA fragments with an average of 11.85 bands per primer (Table 2). Overall, 23 of these bands were monomorphic (15%) and 131 were polymorphic (85%). Primers 5 and 10 had the highest percentage of polymorphic fragments between samples (both 100%). Cophenetic correlation analysis strongly supported the reliability of the phenogram based on the original distance matrix (r = 0.845). The phenogram (Figure 1) separated clearly C. dilatata and C. andreae from the populations of C. penantis, supporting the validity of these species, traditionally considered altogether under the old ‘acutifrons’ complex. Populations of C. penantis from Spain, Portugal, Morocco, Japan and Brazil were clustered together in the RAPD analysis, indicating that, probably, all the specimens of C. penantis could belong to the same species. From the eight populations of this species included in the analysis, only the population of Punta Carnero (Cp-Car) and Ceuta (Cp-Ceu) belonged to the form simulatrix (without proximal projection on male gnathopod 2 propodus). The remaining populations were all provided with a proximal projection belonging, at least, to the forms testudo and lusitanica. Interestingly, the cluster did not show the populations with projection within the same group. Population Cp-Ceu was more similar to the Japanese and Brazilian specimens, that were provided with proximal projection, and Cp-Car was more similar to the population of Mindelo (Cp-Min) and to the population of Torreguadiaro (Cp-Tor), also provided with projection in male gnathopod 2 propodus. According to these results, it seems that the presence/absence of this proximal projection should not be strong enough as morphological difference for separating species, and that the different forms of C. penantis could be explained by intraspecific variation. The only population which showed higher genetic differentiation within the C. penantis complex was the form gibbosa from Coquimbo, Chile (Mayer, Reference Mayer1890, p. 52). This form was referred as Caprella verrucosa Boeck, 1871, in Guerra-García & Thiel (Reference Guerra-García and Thiel2001) and Thiel et al. (Reference Thiel, Guerra-García, Lancellotti and Vásquez2003) due to the presence of abundant dorsal tubercles on pereonites, especially on pereonites 5, 6 and 7, and the robustness of antenna 1 (figure 4 in Guerra-García & Thiel, Reference Guerra-García and Thiel2001). However, the morphology of the specimens changes considerably along the coast of Chile (Thiel et al., Reference Thiel, Guerra-García, Lancellotti and Vásquez2003). Further molecular and morphological phylogenetic studies are necessary to elucidate if the material from Coquimbo is really belonging to Caprella penantis, C. verrucosa or to a different undescribed species. One of the limitations of the RAPD molecular approach is that we cannot conclude which genetic difference as measured by the Nei & Li coefficient is necessary to establish specific differences. Consequently, this method should be used with care to make definite statements about the limits considered to distinguish species, and future studies based on mtDNA, 18S rRNA, ISSR should be necessary to consider if Caprella penantis f. gibbosa should be erected as a new species differing from C. penantis and C. verrucosa. Unfortunately, material of the real C. verrucosa from California or Japan were not available in the present study to be compared with the ‘C. verrucosa’ (C. penantis f. gibbosa) from Coquimbo.

Fig. 1. Phenogram of the fourteen populations studied, based on similarities generated by UPGMA clustering of RAPD band scores.

Table 2. Number of amplification products (bands) obtained per primer in the random amplified polymorphic DNA (RAPD) analysis of the cosmopolitan Caprella penantis and related species.

Caprella penantis is regarded as one of the most problematic caprellids throughout the world, since this species has been recorded under several species or subspecies names from the temperate regions of the world and the need for genetic studies to determine its nomenclatural status at each locality has been pointed out in most of the taxonomic caprellids studies (McCain, Reference McCain1968; Laubitz, Reference Laubitz1972). In Mayer's monographs (Reference Mayer1890, Reference Mayer1903) he described nineteen forms of the ‘acutifrons’ group (forms typica, minor, tabida, tibada, neglecta, gibbosa, andreae, carolinensis, virginia, lusitanica, natalensis, porcellio, simulatrix, testudo, angusta, incisa, verrucosa, borealis and cristibrachium). Several of these forms have already been given specific rank. Forms typica and minor have been assigned to Caprella dilatata (McCain, Reference McCain1968) and probably forms tabida and tibada also belong to C. dilatata, although its taxonomic status under Caprella penantis is still under discussion (Guerra-García et al., Reference Guerra-García, Redondo-Gómez, Espina, Castillo, Luque, García-Gómez and Figueroa2006). Caprella neglecta Mayer, Reference Mayer1890, was also considered as a valid species (Vassilenko, Reference Vassilenko1967; Laubitz, Reference Laubitz1972), form andreae was assigned to Caprella andreae (McCain, Reference McCain1968), forms natalensis and porcellio to Caprella natalensis Mayer, Reference Mayer1903 (Laubitz, Reference Laubitz1972), forms incisa, verrucosa, borealis and cristibrachium to Caprella incisa Mayer, Reference Mayer1903, C. verrucosa, C. borealis Mayer, Reference Mayer1903, and C. cristibrachium Mayer, Reference Mayer1903, respectively (Utinomi, Reference Utinomi1943; Dougherty & Steinberg, Reference Dougherty and Steinberg1953; McCain, Reference McCain1968; Laubitz, Reference Laubitz1972). The form angusta was considered as Caprella angusta Mayer, Reference Mayer1903, by Dougherty & Steinberg (Reference Dougherty and Steinberg1953) and Laubitz (Reference Laubitz1970); but Laubitz (Reference Laubitz1972) questioned the validity of this species. Consequently, its position is still unclear. The remaining forms gibbosa, carolinensis, virginia, lusitanica, testudo (all these forms with proximal projection in the propodus of male ganthopod 2) and simulatrix (without projection) are, consequently, still classified under the species Caprella penantis (McCain, Reference McCain1968; Laubitz, Reference Laubitz1970, Reference Laubitz1972; Krapp-Schickel, Reference Krapp-Schickel and Ruffo1993). The present study shows that, using RAPD, all C. penantis forms seem to be very similar from the molecular point of view, suggesting that there are no clear evidences to separate them as different species. The presence/absence of the projection seems not to be related with a species differentiation but with phenotypical intraspecific variation. The morphological variation among populations of C. penantis could be mainly influenced by the ecological characteristics of the habitat. In this sense, Bynum (Reference Bynum1980) conducted a morphometric study using C. penantis from coastal and estuarine sites in North Carolina and found a gradient of forms related to the degree of exposure to turbulence. Body parts and appendages associated with grasping the substrate of the same caprellid species are capable of modification depending on the degree of wave exposure, thus exhibiting ecological plasticity or ‘ecotopic variation’ (Caine, Reference Caine1989; Guerra-García, Reference Guerra-García2001). Caprellid species in environments with strong wave action tend to have a more robust body than those living in calmer places (Hirayama & Kikuchi, Reference Hirayama and Kikuchi1980).

The RAPD analysis (Welsh & McClelland, Reference Welsh and McClelland1990; Williams et al., Reference Williams, Kubelik, Livak, Rafalski and Tingey1990) requires no prior knowledge of DNA sequence (Hadrys et al., Reference Hadrys, Balick and Schierwater1992) and is faster and cheaper than many other molecular techniques (Costa et al., Reference Costa, Neuparth, Theodorakis, Costa and Shugart2004b), but this method also has some limitations and low reproducibility (Williams et al., Reference Williams, Kubelik, Livak, Rafalski and Tingey1990; Pérez et al., Reference Pérez, Albornoz and Domínguez1998). In fact, in spite of using the same methodology and similar laboratory conditions, the number of bands obtained per primer in the study carried out by Guerra-García et al. (Reference Guerra-García, Redondo-Gómez, Espina, Castillo, Luque, García-Gómez and Figueroa2006, p. 102; Table 2) differs from the number of the present study (see Table 2). However, although RAPD analysis must be interpreted with caution, the results of the present study based on a wide range of samples throughout the world support the preliminary observations made by Guerra-García et al. (Reference Guerra-García, Redondo-Gómez, Espina, Castillo, Luque, García-Gómez and Figueroa2006) based exclusively on material from the Strait of Gibraltar. Additional sampling and future studies based on a larger set of molecular markers and techniques (e.g. mtDNA, 18S rRNA, ISSR, etc.) are necessary to confirm the patterns obtained with RAPD analyses.