Introduction

Following the fundamental work of Darwin (1859), understanding the expression of variation in nature has become essential for the study of evolution, since variation is the basis for evolutionary change. The interpretation of variation, however, has changed in recent years to incorporate phenotypic (developmental) plasticity, in addition to genetic diversity, as important drivers of evolutionary change (West-Eberhard 1989, 2003, 2005; Price et al. 2003; Schlichting 2004; Pigliucci 2007; Pfennig et al. 2010). A major concept is that selection acts on phenotypes and, consequently, phenotypic variation is selectable variation, whether or not it is initially associated with genetic variation (West-Eberhard 1989, 2003, 2005). Phenotypic plasticity can, therefore, contribute to microevolutionary and macroevolutionary processes, emphasizing the importance of the study of variation at different evolutionary levels.

Cnidaria are known for their great morphological variation (e.g., De Weerdt 1981; Silveira and Migotto 1991; Dawson 2005a; Griffith and Newberry 2008; Forsman et al. 2009; Menezes et al. 2013; Ong et al. 2013). Modular growth, characteristic of the polyp stage of many cnidarians, enables a wide variability of colony form through increased regenerative capacity and varying growth rates, branching, number of hydranths, and annulations, contributing to morphological variation in response to differences in environmental conditions (Hughes 1989; Gili and Hughes 1995; Marfenin 1997). The expression of alternative life cycle stages in Medusozoa (Marques and Collins 2004) is another important source of variation in this group. At the microevolutionary level, intraspecific variation in life cycles (e.g., Stefani 1959; Bouillon et al. 1991) has been suggested to have a genetic basis, by means of a switch mechanism responsible for the expression of alternative phenotypes in accordance with environmental cues (West-Eberhard 1986; Boero and Sarà 1987; Boero and Bouillon 1989; Cornelius 1990a; Boero et al. 1997; Bavestrello et al. 2000). At the macroevolutionary level, consecutive suppression and reexpression of the medusa stage during the evolutionary history of the group account for the interspecific remarkable diversity of life cycles (e.g., Boero and Sarà 1987; Boero et al. 1992; Cornelius 1992; Boero and Bouillon 1993, 1994; Piraino et al. 1996; Holst et al. 2007; Miranda et al. 2010; Straehler-Pohl and Jarms 2011). The origin of a medusa in the lineage leading to the Medusozoa extends the levels of variation in this group and reiterates the characteristic modular developmental basis of Cnidaria, contributing to its widely known phenotypic plasticity (see West-Eberhard 2003; Table 1).

Table 1 Phenotypic plasticity in Cnidaria Medusozoa
Full size table

Difficulties in interpreting the morphological variations of Medusozoa have led to many taxonomic problems and are still a source of disagreement among taxonomists on the importance of morphological characters used to diagnose species (e.g. Boschma 1948; De Weerdt 1984 [Hydrozoa, Milleporidae]; Gershwin 2001; Dawson 2003 [Scyphozoa, Aurelia]; Cornelius 1982, 1990b; Cunha et al. 2015 [Hydrozoa, Campanulariidae]; Hirano 1997; Miranda et al. 2009 [Staurozoa, Haliclystus]; Miglietta et al. 2009 [Hydrozoa, Hydractiniidae]). These disagreements occur mainly because intraspecific morphological variations are often misinterpreted as interspecific variations or vice-versa, and consequently, the diversity of the group is frequently misestimated. In many taxa that show some degree of developmental plasticity, intraspecific variation of adaptive traits often parallels interspecific variation (e.g., Badyaev and Foresman 2000; West-Eberhard 2003; Gomez-Mestre and Buchholz 2006). For taxonomy, the definition of species’ diagnostic characters may be confounded by these factors if different levels of variation are not initially considered. For this reason, the variation of morphological diagnostic characters should be carefully examined, and generalizations should be treated with caution.

Considering the phenotypic plasticity of Medusozoa and its importance in the evolutionary history of the group, in this study we reviewed the patterns of morphological variation known for Medusozoa. We specifically evaluated how different interpretations of the levels of morphological variation may influence the understanding of the patterns of diversification of the group. We offered an overview of the different levels of morphological variation in Medusozoa based on information from the literature and unpublished results, including morphological and genetic data. In addition, we present an analysis based on data from GenBank in order to provide an estimate of the potential cryptic diversity in Hydrozoa.

Material and methods

Levels of morphological variation in Medusozoa were assessed by compiling information from the literature on intraspecific and interspecific variation, as well as the occurrence of cryptic species. Additionally, the range of variation of several morphological characters on a case study with Orthopyxis sargassicola (Nutting, 1915) were evaluated (Online Resource 1) in order to illustrate some particular points in the discussion regarding intraspecific variation (comparison between intracolony and population-level variation). This species was chosen because of the large availability of data from different populations in Brazil, which were for the most part included in the molecular phylogeny of the genus Orthopyxis previously published (Cunha et al. 2015).

In order to estimate the potential cryptic diversity in Hydrozoa based on genetic data, we compiled information on COI haplotype sequences deposited in GenBank from several species (Online Resource 2). Genetic distance among haplotypes was calculated as the inverse of the minimum value of similarity recorded by pairwise comparisons of the sequences of each species, using BLAST (Altschul et al. 1990). We also calculated the geographical distance among haplotypes based on geographical coordinate information provided with the sequences metadata. When geographical coordinates were not provided, they were estimated based on the name of the sampling location. Species were then classified in accordance with their life cycles (holoplanktonic/pleustonic or benthic/meroplanktonic) based on literature information. With this data, we fitted a Linear Model (LM) including geographical distance and life cycle strategy as explanatory variables in relation to genetic distance among haplotypes, using R programming language (R Core Team 2015). Continuous variables were log10(x + 1) transformed to meet the assumptions of the LM. All data for each species included in this analysis are available in Online Resource 2.

Microevolutionary morphological variation: intraspecific variation

Intraspecific variation is the variation found within a species, including not only variation between conspecific populations but also individual variation, such as ontogenetic variation and polymorphisms (see Mayr 1973). Intraspecific variation may result from phenotypic plasticity when more than one phenotypic alternative is produced in response to environmental factors (West-Eberhard 1989, 2003). Genotype-specific alternatives are usually referred to polymorphisms (Mayr 1973; West-Eberhard 1989). The term polymorphism, however, has been used in many different contexts, sometimes including aspects of phenotypic plasticity (see Clark 1976 and discussion by West-Eberhard 2003: 378). In the case of marine colonial organisms, polymorphisms are defined as discontinuous variations in the morphology of zooids within a colony (Boardman and Cheetham 1973; Harvell 1994). They are considered an important evolutionary innovation in the Hydrozoa, which evolved independently in multiple lineages within the Hydroidolina (Cartwright and Nawrocki 2010; Maronna et al. 2016).

The extent of functional specialization of polyps varies among species but generally involves feeding (gastrozooids), reproduction (gonozooids), and defense (dactylozooids) (Millard 1975; Bouillon et al. 2004; Mills et al. 2007). Additional types of polymorphisms are found in particular groups (e.g., Namikawa et al. 1992; Gravier-Bonnet 2004, 2008), reaching their highest complexity in siphonophores (Pugh 1999; Bouillon et al. 2004; Dunn and Wagner 2006). Being distinctive features, many polymorphisms (and their absence) are important diagnostic characters (e.g., Calder 1988; Boero et al. 1998; Boero et al. 2000; Schuchert 2008 [Hydractiniidae, Milleporidae, Porpitidae, Zancleidae]; Calder 1997; Gravier-Bonnet 2004 [Plumulariidae]). However, their use as diagnostic characters of supraspecific taxa may cause some taxonomic inconsistencies, since their occurrence varies among species (e.g., Clava and Hydractiniidae, Boero et al. 1998; Schuchert 2004).

Another source of morphological variation is related to ontogeny. Many species, especially in the medusa stage, have been described based on ontogenetic differences (Mayer 1910; Bouillon and Boero 2000). This type of variation may be responsible for considerable differences in bell size and shape or in the number of tentacles and statocysts, and for the appearance or disappearance of morphological characters during development (Russell 1953; Zamponi and Girola 1989; Cornelius 1990b; Lindner and Migotto 2002; Widmer 2004). In the polyp stage of colonial species, such as many hydrozoans, ontogenetic variation is responsible for changes in zooid morphology during its development (Boardman and Cheetham 1973). Variations in internode length, number and orientation of branches are typical ontogenetic changes found in polyps within a colony, particularly in species with upright colonies (e.g., Cornelius 1975, 1990b; Kosevich 2006). Developmental changes may continue throughout the colony’s life because of its characteristic modular growth and may affect many levels of colony organization (Hughes 1989, 2005; Marfenin 1997). Differences associated with colony development (growth and senescence), such as hydranth budding and stolonal growth, branching and regression are usually referred to astogeny (Boardman and Cheetham 1973; Hughes 1989) and may be responsible for spatial and temporal changes in colony morphology (Braverman 1974; McFadden et al. 1984; Vogt et al. 2011). Although they appear during the course of development, variations in colony morphology may also result from variations in physiological processes (Dudgeon and Buss 1996; Vogt et al. 2008; Bumann and Buss 2008) and environmental factors (e.g., Vogt et al. 2011; Miglietta and Cunningham 2012).

Intraspecific variation may be triggered by many different factors, especially in different populations. Notably, individual morphological variation may sometimes parallel variations found among populations. For instance, experiments with replicated and transplanted colonies of Millepora spp. and Bougainvillia muscus (Allman, 1863) have shown that colony growth patterns change from branched forms to more robust, solid forms with variations in water flow (De Weerdt 1981; Griffith and Newberry 2008). Similar morphological variations are found among populations living in habitats with contrasting water movement conditions (Kaandorp 1999). In addition, in the family Campanulariidae, variations in colony size, perisarc thickness, length of the hydrotheca and gonotheca, and number of branches and annulations are found among populations subjected to contrasting water flow, temperature, and substrate type (Naumov 1969; Ralph 1956; Cornelius 1975, 1982, 1990b; Lindner and Migotto 2002). Many of these variations, however, may also occur within a single colony (Fig. 1). Although the amplitude of variation of morphological characters at the colony level is not the same as at the population level (Fig. 2), there may be an overlap of morphological variation produced by different levels of intraspecific variation.

Fig. 1
figure 1

Intracolony variation in Orthopyxis sargassicola (Nutting, 1915) based on three polyps randomly sampled from a single colony (MZUSP4079, see Online Resource 1). a Measurements (in μm) of total length of trophosome (Tr), length of pedicel (Pd), hydrothecal length (Hd), and diameter at margin (Diam); b measurements (in μm) of maximum perisarc (Ps) thickness (Thick) of hydrotheca and pedicel at medial portion, as well as maximum number of sinuosities (NS) in pedicel and number of hydrothecal cusps (NC ); c, d polyps of O. sargassicola from a single colony (both polyps are at the same position of maximum perisarc thickness). Note the differences in size and shape of the pedicels (Pd) and hydrotheca (Hd), as well as the perisarc thickness (Ps) and sinuosities of the pedicel (S)

Full size image
Fig. 2
figure 2

Mean values (μm) (±standard deviation) of a total length of trophosome (Tr), length of pedicel (Pd), hydrothecal length (Hd), and diameter at margin (Diam); and b maximum perisarc (Ps) thickness (Thick) of hydrotheca and pedicel at medial portion, maximum number of sinuosities in pedicel (NS), and number of hydrothecal cusps (NC). Measurements were taken from polyps of the same colony of Orthopyxis sargassicola (intracolony variation, MZUSP4161, see Online Resource 1) and from polyps of different populations of the same species from the states of São Paulo, Espírito Santo, Rio de Janeiro and Santa Catarina, Brazil (variation among populations, Online Resource 1)

Full size image

Assessing morphological variations between populations may be a difficult task since it involves the presupposition that populations are conspecific, when they might, in fact, represent different species. Although many studies have reported intraspecific morphological variation in the Medusozoa, only a few of them have attempted to assess the identity of the populations studied, using phylogenetic inferences or other methods for detecting reproductive isolation (e.g. Dawson 2005a; Galea and Leclère 2007). For instance, many studies have reported variations in the symmetry of medusae (Scyphozoa and Hydrozoa, Navas-Pereira 1984; Gershwin 1999; Silva et al. 2003; Nogueira and Haddad 2006) and stauromedusae (Zagal 2008). Although these variations are known to originate at the clonemate level in Aurelia (Gershwin 1999) and may also be a response to variation in physical factors (e.g., temperature and salinity, Zamponi and Genzano 1989), the underlying causes of these variations are still unclear, and they may be different depending on the group and populations studied. As a result, studies of morphological variation may end in questioning the taxonomic affinities of the populations sampled (e.g., Bolton and Graham 2004). Considering the complicated taxonomic history of many groups within the Medusozoa, it is not always easy to delimit conspecific individuals or populations based on morphological characters alone. Individuals from different species may frequently be regarded as conspecific, particularly since intraspecific morphological variation may extend from individuals to populations. Obviously, in order to be confident of the taxonomic level investigated, it is important to know whether one is dealing with intraspecific or interspecific variations.

Macroevolutionary morphological variation: interspecific variation

Difficulties related to variations in morphological characters have led taxonomists to search for additional characters that could contribute to species delimitation. Characters of the cnidome in many hydroid species (e.g., Östman 1982, 1987; Marques 1995, 1996; Morandini and Marques 2010), as well as ecological and behavioral patterns in medusae (e.g., Dawson and Martin 2001; Dawson 2005b) have contributed to the diagnosis of species in these groups, but the difficulties in assessing and describing these characters have limited their use in species delimitation. Similarly, advances in molecular techniques have introduced many new approaches for studying species relationships and have improved our understanding of the evolutionary history and diversification of the Medusozoa (Collins et al. 2006; Leclère et al. 2007, 2009; Cartwright and Nawrocki 2010; Kayal et al. 2013; Cunha et al. 2015). Species delimitation, however, still remains difficult, and many studies adopt integrative approaches, recognizing species based not only on genetic divergence but also on additional characters, such as morphological, ecological, and behavioral, that could contribute evidence for species boundaries (Dayrat 2005; Padial et al. 2010).

The combination of morphological and molecular data for studying species boundaries contributed to the reevaluation of several morphological diagnostic characters in Medusozoa, leading to the description of many new species (e.g., Schierwater and Ender 2000; Collins and Daly 2005; Bayha and Dawson 2010; Collins et al. 2011; Cunha et al. 2015) and the revalidation of formerly synonymized species (e.g., Dawson 2003, 2005c; Schuchert 2005; Miglietta et al. 2007, 2009; Fritz et al. 2009; Lindner et al. 2011; Moura et al. 2012). Reassessment of morphological characters is showing that many “species” previously considered cosmopolitan are in fact geographically isolated lineages, which often can be delimited morphologically (Dawson 2003; Miglietta et al. 2007; Bentlage et al. 2010). This means that the underestimation of species diversity in the Medusozoa, in most cases, results from misinterpretations of species diagnostic characters, which may explain the historical splitting and lumping of species that have been common in several groups. The paucity of morphological characters and poor descriptions in some groups, as well as the wide morphological variation in others, have certainly contributed to these misinterpretations.

Morphological variation can be misleading when there is an overlap between intraspecific and interspecific variations. Considering the phenotypic plasticity in colony form shown by species of Millepora in different water movement conditions (e.g., De Weerdt 1981; Kaandorp 1999, see previous section), molecular and morphological data proved that this variation is also interspecific and resulted in the delimitation of two different lineages based on colony growth form (Meroz-Fine et al. 2003). Moreover, branched and unbranched forms of species of Aglaophenia, commonly thought to be a result of phenotypic plasticity (e.g., Andrade and Migotto 1999), were shown to represent different species in the North Atlantic (Thorpe et al. 1992). It is clear from these findings that interspecific variations may easily be misinterpreted as intraspecific variation.

Once again, the family Campanulariidae is a good example of the historical splitting and lumping of species owing to misinterpretations of morphological characters. The validity of the genus Orthopyxis L. Agassiz, 1862, for instance, is a frequent source of disagreement among taxonomists, since the perisarc thickness, regarded by some authors as one of the diagnostic characters of the genus (Calder 1991; Cornelius 1995; Bouillon et al. 2004), is also thought to be phenotypically plastic (Millard 1975; Galea et al. 2009). This common belief prevents the use of perisarc thickness as a diagnostic character in Orthopyxis, although it may have taxonomic value for delimiting other species of the family (e.g., Obelia geniculata (Linnaeus, 1758), Cornelius 1975). Molecular and morphological data clearly show that intraspecific variation in perisarc thickness occurs, but the perisarc also shows interspecific variation, which makes it a reliable character for species delimitation within Orthopyxis (Cunha et al. 2015). Additionally, this approach supported the validity of the species Orthopyxis caliculata (Hincks, 1853) (Cunha et al. 2015), which was long regarded as a synonym of the widespread species Orthopyxis integra (MacGillivray, 1842) (Cornelius 1982, 1995). The evidence that characters previously regarded as intraspecifically variable may be diagnostic of different species in Orthopyxis, support the idea that O. integra might not have as wide a geographic range as presently thought, and that the morphological variation assumed for this species is overestimated (e.g., shape of the gonotheca; see Cornelius 1995; Cunha et al. 2015).

Macroevolutionary phylogenetic signal with no morphological variation: cryptic species

The existence of species that are morphologically indistinguishable has always intrigued taxonomists. These species were originally termed “sibling species” and defined as “sympatric forms which are morphologically very similar or indistinguishable, but which posses specific biological characteristics and are reproductively isolated” (Mayr 1964: 200). This morphological indistinctness, however, may prove to be a result of previously overlooked morphological differences (Mayr 1976). This may explain the majority of the cryptic diversity found among the Medusozoa (e.g., Aurelia, Dawson and Jacobs 2001; Schroth et al. 2002; Dawson 2003; Nemertesia, Moura et al. 2008, 2012; Acryptolaria, Lafoea, Moura et al. 2011), although in some cases, a reassessment of morphological characters has not proved useful for delimiting species (e.g., Cassiopea, Holland et al. 2004; Stylactaria, Miglietta et al. 2009; Cryptolaria pectinata, Moura et al. 2011).

Current estimates of the total global species richness indicate that the Hydrozoa as a group has an increasing rate of species discovery and a high proportion of cryptic species, probably due to the paucity of morphological diagnostic characters (Appeltans et al. 2012). Indeed, our estimate showed that the amount of cryptic diversity within nominal species of Hydrozoa is significant (R 2 adj = 27.2 %, p < 0.001, Fig. 3). The positive association between geographical and genetic distances (F = 21.0417, p < 0.01) indicate that many species of Hydrozoa may contain cryptic lineages, especially if samples from different geographical localities are considered. Schuchert (2014), for instance, showed that Plumularia setacea is a species complex that is mostly composed of geographically circumscribed lineages, and the same is true for other species of hydrozoans (e.g., Obelia geniculata, Govindarajan et al. 2005; Clytia gracilis and Obelia dichotoma, A.F. Cunha pers. obs.).

Fig. 3
figure 3

A Linear Model (LM) showing the relationship between haplotype genetic distance (n ≥ 2) and geographical distance (Km) of specimens with COI sequences deposited in GenBank (up to January 2013), in accordance with their life cycle strategy (holoplanktonic/pleustonic or benthic/meroplanktonic). Both geographical distance and life cycle strategy were included as explanatory variables. Similarity was calculated using BLAST by pairwise comparisons of all haplotypes of a single species, and the inverse of the minimum value of similarity recorded was used as its genetic distance. The geographical distance was calculated based on geographical information provided with the metadata for the sequences. See Online Resource 2 for more details

Full size image

Additionally, the occurrence of species complexes in the Hydrozoa has frequently been associated with limited dispersal abilities of species that lack a long-lasting pelagic phase (Moura et al. 2011, 2012; Schuchert 2005, 2014). Hydrozoans with holopelagic life cycle stages were shown to be more widely distributed and have lower species richness than benthic and meroplanktonic species (Gibbons et al. 2009), corroborating the prediction that a relatively short period in the plankton is associated with limited dispersal (Palumbi 1992; Bradbury et al. 2008). Following these predictions, we also found a significant relationship when considering different life cycle strategies, with meroplanktonic/benthic species showing higher genetic distance between haplotypes than did holoplanktonic species (F = 4.5027, p = 0.0379, Fig. 3). This is evidence that benthic/meroplanktonic taxa have potential for the discovery of cryptic lineages even over short geographical distances, probably because of their limited dispersal ability. Studies investigating species boundaries among taxa with different life cycle strategies are important to corroborate this hypothesis. Nevertheless, our results provide evidence that the number of species in the Hydrozoa, and probably in all the Medusozoa, is underestimated. Increased sampling, integrative approaches, and careful investigations of morphological variations will inevitably uncover this hidden species diversity.

Conclusion

It is clear that morphological characters have many different levels of variation or may not vary at all in some cases (summarized in Fig. 4). When morphological variation is present, it may frequently overlap between microevolutionary and macroevolutionary scales, hampering their use at different hierarchical and inclusive taxonomic levels. The widespread morphological variation of the Medusozoa, as well as the frequent overlapping between intraspecific and interspecific variation, indicate that phenotypic plasticity may play an important role in the diversification of the group (see West-Eberhard 1989; Pfennig et al. 2010). However, whether alternative phenotypes are, indeed, environmentally induced or are genetically controlled is an important question (see Schwander and Leimar 2011) which needs further investigation. Nevertheless, at the taxonomic level, morphological variation leads to misinterpretations of diagnostic characters and difficulties in species delimitation. Importantly, however, the level of variation and amount of overlap may be different depending on the group studied and its general biology and life history. In order to minimize the possibility of misinterpretations of morphological characters, generalizations should be avoided, and morphological variation should be interpreted within the context of each taxon, taking into account its phylogenetic relationships and evolutionary history.

Fig. 4
figure 4

Schematic summary of levels of morphological variation found in medusozoans, including the absence of variation (cryptic species). Clades indicate different lineages, colors and shades represent the phenotype in current time, and the circles represent individuals. Note that there is individual variation (arrows) and it can parallel intraspecific variation. The same occurs with interspecific variation, which can parallel intraspecific variation in b

Full size image