Hidden diversity under morphology – based identifications of widespread invasive species : the case of the ' well – known ' hydromedusa Craspedacusta sowerbii Lankester 1880

Hidden diversity under morphology–based identifications of widespread invasive species: the case of the 'well– known' hydromedusa Craspedacusta sowerbii Lankester 1880. A relatively scarce number of morphological features available for delimiting closely related species and an increasingly worrisome scenario on Global Climate Change causing the rapid dispersion of invasive alien species can lead to the rapid spread of reports of a given species around the world. Craspedacusta sowerbii Lankester, 1880 is considered the most widespread freshwater jellyfish species and has been reported in numerous locations on all continents except Antarctica. Recently, a few medusae attributed to C. sowerbii were collected from a water reservoir (Bin El Ouidan) in Morocco, this being the first confirmed record of the species from North Africa. The morphology of these newly collected specimens agrees well with previous descriptions, but mitochondrial (Cox1 and 16S) and nuclear ITS (ITS1–5,8S–ITS2) molecular data lead to a discussion of a more complex general view concerning the number of species, synonyms and nomenclatural problems hidden behind the reports of Craspedacusta sowerbii.


Introduction
Among the recommendations suggested by a wide panel of specialists to ensure progress in the management of aquatic NIS (non-indigenous species), the first was the availability of taxonomic expertise (see Ojaveer et al., 2014). Taxonomists produce the basic knowledge for understanding biodiversity (Agnarsson and Kuntner, 2007;Linse, 2017). The second was the need to use molecular tools, as classical taxonomy often requires additional sources of information for species description and identification (Goldstein and DeSalle, 2011), leading to an integrative taxonomy (DeSalle et al., 2005;Rubinoff et al., 2006aRubinoff et al., , 2006bPires and Marinoni, 2010;Chen et al., 2011).
Despite this, reliable identification of organisms to species level is one of the greatest constraints. The lack of specialists and the inaccuracy of species identifications often result in an erroneous interpretation of the actual biodiversity and inadequate conservation policies, from local to global levels. This is the so-called 'taxonomic impediment' (Hoagland, 1996;Giangrande, 2003;Dar et al., 2012). Avoiding this problem by limiting OTUs data matrices to higher taxonomic levels or considering functional biodiversity (Cernansky, 2017) is not a viable solution, especially when working on NIS, whose influence at different levels on native ecosystems is well documented (e.g. Bax et al., 2003;Wallentinus and Nyberg, 2007;Walther et al., 2009;Poulin et al., 2011;González-Duarte et al., 2016;among others).
The life cycle of Craspedacusta sowerbii includes both polyp (assuming asexual reproduction) and free-swimming stages (involved in sexual reproduction) (Bekleyen et al., 2011;Gasith et al., 2011). The appearance of the active medusa stage is related to an increase in water temperature (Bekleyen et al., 2011). Occurrences of this pelagic stage are sporadic, lasting only a few weeks, usually in the late summer and autumn (Minchin et al., 2016). The polyp stage is often overlooked because of its small size, having a wide capacity to tolerate different temperature and light conditions (see Payne, 1924;Boulenger and Flower, 1928;Acker, 1976;Acker and Muscat, 1976). The polyp and medusa stages are rarely reported together (see Failla-Siquier et al., 2017). Duggan and Eastwood (2012) established a protocol to find polyp stages that would be usable even in water reservoirs where the medusa stage had not been previously observed. These authors reached the conclusion that C. sowerbii is more common and widespread than is apparent from observations of medusae. Estimating the timing of introduction of this species in a given region is therefore difficult if it is only carried out after jellyfish findings have been recorded.
A few individuals of a hydromedusa species were recently detected in a Moroccan reservoir. These specimens were initially identified (based on morphological characters) as the well-known alien widespread species Craspedacusta sowerbii. This record is the first confirmed finding of this species in North Africa. However, a molecular study of this material and its comparison to previously available information revealed a more complex scenario, with nomenclatural and biogeographic implications. As this hydromedusa species is often reported in lists of alien species, the correct specific identification of the different Craspedacusta lineages becomes an urgent challenge to correctly understand how many invasion events and species could be involved.
The present paper aims to stress the risks linked to the current trend of exponentially increasing numbers of morphology-based reports of invasive species. An integrative view, including both morphology and molecular information, should be applied as a rule for checking the current identity of these 'well known' species as there are several examples of cryptic species that are difficult or impossible to delimit due to overlapping morphological characters.
( fig. 1). It covers 3,740 ha and has a maximum depth of nearly 100 m. The total volume reaches 1,384 million m 3 . The reservoir was built between 1949 and 1953.
In December 2015 we surveyed two sites ( fig. 1), and found medusa stages attributable to the genus Craspedacusta at site S1 (6º 26' 13'' W; 32º 05' 33'' N). In total, four specimens were collected by scuba divers from the water-column between 1 and 5 m of depth. The water temperature, as indicated by the dive computer, was 20 ºC and the visibility was 3 m. Two specimens were fixed in absolute ethanol for the molecular study, while the other two were fixed in formalin 4 % for morphological observations.

Nomenclatural remarks
Despite the precise nomenclatural comments by Fritz et al. (2007: 54) about the discovery, first descriptions of this jellyfish species (see also Allman, 1880;Lankester, 1880aLankester, , 1880b, and ICZN decision (see also Allen, 1910;Stiles, 1910), several subsequent authors still reported the species with the specific epithet 'sowerbyi'. In the original description, Lankester (1880a: 148) used the spelling Craspedacusta sowerbii in honour of Mr. Sowerby, understanding the genitive singular of the complete latinization of Sowerby to Sowerbius. The use of the form 'sowerbyi' must be considered an erroneous spelling (Zarazaga, pers. comm.). In this case, Article 33.3.1 of the ICZN (1999) about the predominant use of erroneous spellings cannot be applied (indeed, according to Fritz et al. (2009) it is about 40 % of all references). Thus, in order to avoid the use of 'sowerbyi', all references to the species of Lankester in this paper will be made with the original spelling.

Morphological observations and measurements
Observation and photography of different parts of the medusae were performed with a camera (ToupCam™) attached to light microscopy (Olympus CX41). A Panasonic Lumix FZ28 camera was used for macroscopic photography. Measurements of bell and gametogenic tissues and tentacle length were performed using ImageJ 1.46r software (NIH, Bethesda, MD, USA).

Morphological remarks
The medusa was the only stage recovered ( fig. 2A). No polyps were found, and all specimens are female. The average bell/umbrella diameter is 20 ± 1 mm (19-21), flattened form. The mouth has four slightly folded lips overpassing the umbrella margin. Four gametogenic tissues pouch-like structures (7-11 mm length) are hanging from the radial canals. They are opaque in the basal fold-like part and translucent and voluminous in the apical part, giving a triangular shape ( fig. 2B). The tentacles have no organs of adhesion and are connected to the marginal end of the umbrella on the ring canal ( fig. 2E). Four long perradial tentacles (7-9 mm in length) emerging from the end of the four radial canals at the umbrella margin. About 60 medium tentacles (2.0-4.5 mm in length) arising from the pole of the bell were counted between the four long tentacles. Approximately 420 shorter tentacles (0.5-1.5 mm in length) extend around the bell edge. The three different sizes of tentacles are organized in a regular distribution along the umbrella edge. Along the tentacles, nematocysts are grouped in patches that

Phylogenetic analyses
Cox1 analyses ( fig. 3) placed the sequences obtained in this work for the two Moroccan specimens in a well-supported clade (Bootstarp [Bts.] 99, posterior probability [PP.] 0.99) with the German sequences and a Chinese (Sichuan province) sequence constituted, with an internal p-distances (German-Moroccan to Chinese sequence) of 0.3 %. This last German-Moroccan-Chinese clade is the sister group of a relatively poorly supported clade with two well defined groups, a Switzerland sequence and a well-supported clade (Bts. 100, PP. 1) including a conglomerate of Chilean-Italian-Indian-Grecian-Chinese sequences (all of which were also attributed to C. sowerbii), average uncorrected p-distance between these last two clades 17.5 %. Average uncorrected p-distance between German-Moroccan-Chinese clade and Switzerland sequence 13.6 %. Average uncorrected p-distance between German-Moroccan-Chinese clade and Chilean-Italian-Indian-Grecian-Chinese clade was 16.3 %. Phylogenetic hypotheses based in Cox1 suggest that there are at least three Craspedacusta species in Europe.   Lankester, 1880: A, superficie subumberlar. B, m, manubrio. C, gn, gametos;rc, canal radial;t, tentáculos;v, velo . According to our current ITS knowledge, four main lineages (species) can be detected, two of them (Clades I and III) including specimens identified as C. sowerbii. At least two species are present in Europe, while the known American sequences (Chile) and those from Central Europe and North Africa are definitively different lineages (Clades I and III, respectively).
Unfortunately, there is no homogenous knowledge of the three genetic markers here examined along the entire distributional area where specimens attributed to C. sowerbii have been reported. Figure 6 shows the worldwide distribution of the main clades detected in the separate analyses of the three markers (see also fig. 3, 4, and 5 for comparison).

Morphological remarks
Caraspedacusta sowerbii has been recorded in several localities around the world. However, many identifications are not fully reliable since the records do not give detailed morphological characters (Moreno-Leon and Ortega-Rubio, 2009;Jakovčev-Todorović et al., 2010;Stefani et al., 2010;Gasith et al., 2011;Souza and Ladeira, 2011;Galarce et al., 2013;Gomes-Pereira and Dionísio, 2013;Fraire-Pacheco et al., 2017). Moreover, many hydromedusae species have several similar morphological characters especially within the genus Craspedacusta (Jankowski, 2001), and only a few records gave more detailed descriptions of specific morphological characters (Kramp, 1950;Jankowski, 2001;Lewis et al., 2012). Indeed, up to eleven Craspedacusta species have been described, mostly recorded from China only (Jankowski, 2001). However, according to Bouillon et al. (2006) and , many species may not be valid and are likely to be just morphological variations of the same species. Jankowski (2001) studied all the species recorded within Craspedacusta in detail and found that only three should be considered valid (C. sowerbii, C. iseanum Oka and Hara, 1922 and C. sinensis Gaw and Kung, 1939), and two are uncertain (C. sichuanensis He and Kou, 1984 andC. ziguiensis He andXu, 1985); the rest seem to be synonyms of C. sowerbii, keeping in mind that two other species were synonymised [the marine species C. vovasi Naumov and Stepanjants, 1971 and the brackish water one C. marginata Modeer, 1791 (see Hummelinck, 1938). In the present paper, the morphological characteristics of our specimens coincide with the typical characters of the medusae belonging to the genus Craspedacusta (Russell, 1953;Bouillon and Boero, 2000;Bouillon et al., 2004Bouillon et al., , 2006. They have, apart from a well-developed marginal nematocysts ring, four simple radial canals from which pouch-like gametogenic tissues are hanging, and centripetal vesicles embedded in the The 16S analyses ( fig. 4) indicate that the analyzed Moroccan individuals merge well among other Craspedacusta sequences. There is very little previous 16S information on Craspedacusta sowerbii, just a sequence from Lake Huato (USA), another from Uruguay, two sequences from Switzerland, and a sequence of unknown locality (KY077294). Moroccan sequences form a well-supported clade (Bts. 99, PP. 1) with the sequence KY077294 and the two Switzerland sequences. Uncorrected p-distance between Morocco-unknown locality and Switzerland sequences is 0.2 %. This last European clade (having in mind the unknown origin of one of the sequences) is the sister group of the clade formed by both American sequences (USA + Uruguay), which are identical. All these mentioned sequences, all identified as C. sowerbii, are related to another two Craspedacusta species (C. sinensis and C. ziguiensis) in a relatively well-supported clade (Bts. 82, PP. 0.95).
The clade grouping all Craspedacusta species is the sister group of Limnocnida tanganjicae, with high support (Bst. 90, PP. 0.99). Mean uncorrected p-distance between American populations attributed to C. sowerbii (USA and Uruguay) and the Moroccan-Switzerland-unknown-origin sequences is 4.3 % ± 0.1 (range 4.2-4.5 %). The genetic distance between the other two Craspedacusta species (C. ziguiensis and C. sinensis) is 6.2 %, and the distance between the Moroccan specimens and the latter species is 6.8 % and 8.3 %, respectively. In general, all Olindiidae genera are well supported in this 16S phylogenetic hypothesis. Inter-genera genetic distances (uncorrected p-distances) in 16S seem to be around 13-30 %. According to 16S' knowledge, a single species occurs in Morocco and Switzerland that is different from that present in America (USA + Uruguay).
The ITS phylogenetic analyses ( fig. 5) benefits from a higher number of sequences; thus, the analyses is here focused on the genus Craspedacusta instead of the whole available olindiid taxa. Craspedacusta sequences are mainly obtained from central Europe and China, although recent sequences from Chile and Sicily (Italy) have been published. Three main clades can be detected, all of them including Chinese specimens. Clade I (Bts. 99, PP. 1), includes Chinese sequences attributed to C. kiatingi and C. sichuanensis, as well as all German sequences attributed to C. sowerbii and the sequences obtained in this study from Moroccan specimens. The sister group of Clade I is composed of a single sequence of C. ziguiensis from China [support between both sister groups (Bts. 99, PP. 1)]. On the other hand, Clade II (Bts. 99, PP. 1), includes Chinese sequences attributed to C. sinensis and C. brevinema, while Clade III (Bts. 90, PP. 0.56) includes Chinese sequences attributed to. C. sowerbii and C. xinyangensis, as well as sequences from Italy and Chile.
Sequences from Clade I (where the Moroccan specimens are included) have genetic distances (uncorrected p-distances) between 0.0 and 0.9 % (mean and SD 0.1 % ± 0.2), while this Clade I is 3.8 % ± 0.1 (range 3.6-4.2 %) distant from its sister group (C. ziguiensis). Uncorrected p-distances between Clade I and Clade II are 19.9 %± 0.4 (range 18.9-20.6 %),   velum as internal closed ecto-endodermal statocysts. Because these common similar characters within the Craspedacusta species lead to confusion and doubts when identifying a specimen and ascribing it to a determined species, the application of more specific characters is needed. The here-observed specimens identified as C. sowerbii have four prominent large perradial tentacles, which are clearly shorter in C. sinensis (Kramp, 1950;Jankowski, 2001). This latter species, also found in China, is very similar to C. sowerbii. According to Kramp (1950), it differs from C. sowerbii also in having a markedly irregular distribution of the different tentacle sizes (which are evenly distributed in our medusae), as well as a characteristic nematocyst distribution on tentacles. In our specimens, transverse belts of clustered groups of two to 10 nematocysts cover the tentacles, while in C. sinensis, nematocysts are located at the end of elongated cylindrical papillae that are not arranged in transverse rings on tentacles (Kramp, 1950;Jankowski, 2001). Moreover, in active swimming specimens, C. sinensis is easily recognizable by its remarkable changes in the umbrella diameter. This species actually varies from 0.48 cm at systole (contracted bell-shaped umbrella) to 1.8 cm during diastole (maximum dilated flattered umbrella) (Kramp, 1950). On the other hand, the observed extended tubular statocysts of different lengths and embedded in the velum of the here-studied specimens confirm our identification and discard the possibility of ascribing our medusae to C. iseanum. This species, found in Japan, is also very similar to C. sowerbii. According to Uchida (1955), the statocysts in C. iseanum are oval-shaped. Moreover, adult specimens in this species vary from five to 18 mm of umbrella diameter and have up to    (Lewis et al., 2012), while C. sowerbii adult specimens, like in the here-observed medusae, can reach up to 25 mm of umbrella diameter and have more than 400 tentacles (Russell, 1953;Jankowski, 2001). Concerning the nematocysts, C. iseanum have scattered, and not clustered, nematocysts on the tentacles. Nevertheless, all these specific characters may be trustworthy only when dealing with the identification of adult living specimens or at least adult well-preserved ones.

Phylogenetic analyses
Despite the abundant literature reporting the occurrence of Craspedacusta sowerbii around the world (see Dumont, 1994;Didžiulis and Żurek, 2013 for additional references), available molecular information is relatively scarce. Part of this information is published as a representation of the genus (or family) for general phylogenetic papers about different cnidarian taxa (e.g. Collins, 2002;Kayal et al., 2015;Grange Fig. 5. Molecular analysis by the ML method. Relationship of Craspedacusta species using Obelia dichotoma as outgroup; the analysis is based on ITS. The tree is drawn to scale, with branch lengths measured in the number of substitutions per site. Symbols of groupings/clades correspond to of the ITS map in figure 6.  al., 2017), and there are few specific papers on the phylogeny, molecular systematics, diversity and distribution of the genus Craspedacusta (Zou et al., 2012;Fritz et al., 2009;Zhang et al., 2009). A number of sequences can be obtained in databases such as GenBank, ca. 55 of them for the ITS region (for seven putative species), ca. 50 sequences are for the Cox1 fragment (all of them attributed to C. sowerbii), and seven are for the 16S (ascribed to three different species).
As previously commented, recent morphological (Jankowski 2001) or molecular (Fritz et al., 2009;Zhang et al., 2009) contributions drastically reduced the number of species to three (or maybe five). On one hand, for Jankowki (2001), only C. sowerbii, C. iseanum, and C. sinensis (and perhaps C. sichuanensis and C. ziguiensis) could be considered valid. On the other hand, the two simultaneous contributions by Fritz et al. (2009) andZhang et al. (2009) pointed out the existence of different lineages (in the former) and species (in the later), thus stressing the lack of consensus on the real diversity and systematics of the genus. Fritz et al. (2009) considered that the morphology of their German and Austrian samples agrees with C. sowerbii, and hence the Chinese ITS sequences (identical to their European material) attributed to C. kiatingi should be considered as C. sowerbii var. kiatingi (Gaw and Kung, 1939;Kramp, 1950). Fritz et al., (2009) identified three main clusters within their dataset: 'sinensis' [for C. sinensis, and C. brevinema (considered by these authors as a variety of the former)], 'sowerbyi' [sic, for Chinese sequences of C. sowerbii and C. xianyangensis (considered by these authors as a variety of the former)], and "kiatingi" (for the German and Austrian sequences of C. sowerbii, and the Chinese sequences of C. sichuaensis and C. kiatingi (considered by these authors as a variety of their European C. sowerbii)], remaining as doubtful the status of C. ziguiensis. In short, for Fritz et al. (2009) the "data support the assumption that there are three valid species, with the possibility of C. ziguiensis being a fourth one, and several, morphological quite different sub-species or variations of the freshwater jellyfish C. sowerbii". Although the identification of C. sinensis and C. ziguiensis as different species seems to be clearly stated, the assignable different specific name to be used for the two other Clades ('kiatingi' and 'sowerbii') is not so clearly defined in this last paper. Zhang et al. (2009) analysed eight putative Craspedacusta species using the nuclear marker ITS. Obviously, the trees obtained by these authors show similar conclusions, as both research groups shared a similar set of sequences: C. xinyangensis should be the synonym of C. sowerbii, C. sichuanensis the synonym of C. kiatingi and C. brevinema the synonym of C. sinensis, while the taxonomic status of C. ziguiensis is still uncertain. However, the main difference between the two contributions is the implications of those Austrian and German sequences, defining a clade C. kiatingi-C. sowerbii. The Chinese authors were probably unaware at that moment that a number of European sequences could be attributed to their C. kiatingi.
Recently, Schifani et al. (2018) and Fuentes et al. (2019) obtained sequences from specimens identified as C. sowerbii from Sicily (Italy) and Chile, respectively. In their Cox1 analyses, Schifani et al. (2018) identified a Sicilian-Grecian-Chinese clade as C. sowerbii, a second German-Chinese clade as C. kiatingi, and a third clade of unknown origin as Craspedacusta sp. (actually, this last sequence was deposited by Dr. P. Schuchert from a polyp stage specimen collected in Ringwiler Weier (Canton Zurich, Switzerland) (see also features part of this sequence in GenBank Accession number MF000493). The ITS analyses of these authors also identify a Sicilian-Chinese C. sowerbii clade and a German-Chinese C. kiatingi clade (the origin of the sequence FJ423632 is indicated to be German Donau (Danube) in GenBank.
It is somehow ironic that the type locality of this considered-to-be invader alien species was a water-lily tank in the Botanical Garden in Regent's Park, London (Lankester, 1880a), while the origin of the species (and the diversity hot-spot of the genus Craspedacusta) is currently supposed to occur in the Yangtze River valley in China (e.g. Didžiulis and Żurek, 2013). Many changes have occurred since the late 19 th century in the type locality, which finally disappeared in 1932 (C. Magdalena, pers. comm.). Anyway, the species was subsequently reported also from Southern England (Broom Water, Teddington), only a few kilometres away from the type locality (Green, 1998). Thus, it is plausible that future sequences obtained from a specimen collected in the London area could be considered a topotype (or a neotype could be established, since the original type material seems not to have been deposited in any institution), and could help to soundly and univocally define C. sowerbii. The nomenclatural consequences of the current absence of molecular data from the area of the type locality are in fact very important.
For the moment, according to the ITS phylogenetic hypothesis, mainland Europe and Morocco share the same haplotype, which is also shared with specimens identified as C. kiaitingi and C. sichuanensis from China, the diversity hot-spot of this hydromedusa genus. Most Chinese specimens identified as C. sowerbii are in a well separated clade (see fig. 5). It is possible to speculate on two scenarios: 1) the South England sequences are identical (or similar) to those from mainland Europe and Morocco (Clade I); and 2) the South England sequences are different from those from mainland Europe and Morocco, but identical (or similar) to those identified as C. sowerbii by Zhang et al. (2009) and that from Sicily (Clade III). The direct consequence of the first scenario would be that C. kiatingi and C. sichuanensis sequences from Zhang et al. (2009) must instead be attributed to C. sowerbii, while another available name should be selected for those sequences identified as C. sowerbii by Zhang et al. (2009), the Clade 'sowerbyi' of Fritz et al. (2009), including the Sicilian and Chilean specimens (Clade III in this paper). Conversely, the consequences of the second scenario suggest that at least two species of Craspedacusta occur in Europe and North Africa, C. sowerbii in Southern England,   fig. 3, 4 y 5).

Cox1
16S ITS and C. kiatingi [or another available name according to the ICZN (1999)] in the rest of this area, as well as in China. For the moment we have no ITS information from the specimens collected in Switzerland, Greece, or India.
The Cox1 phylogenetic hypothesis clearly shows that Europe experienced the invasion of at least three Craspedacusta species, one of them in central Europe (Germany) and Morocco, the second one is currently present in Greece and Sicily, while the third one is found in Switzerland (see fig. 3 and 6).The two first invasions are connected (or simply share similar sequence) with their respective Chinese populations. This was already detected by Karaouzas et al. (2015), suggesting that the phylogeny of the genus is in need of further investigations, as genetic distances between the C. sowerbii clades are around 15 %. In the available Cox1 information for olindiid species, all genera except Craspedacusta are represented by a single species or haplotype, making difficult to discuss about the expected range of genetic distances at species level. In our phylogenetic hypothesis uncorrected pdistances between olindiid genera varies between 15 and 26 %. As in the previous discussion, the name to be used for each Craspedacusta species will depend on the knowledge of a (still unknown) Cox1 sequence from a putative Southern England population, possibly after the establishment of a neotype.
Information based on our 16S phylogenetic hypothesis about Craspedacusta species delimitation is scarce, but it is well defined that a single species that can currently be identified in America (Lake Huato, USA and Uruguay) is different from the one present in North Africa (Morocco), Switzerland, and an unknown locality (sequence KY077294, see Grange et al., 2017). The same problem already discussed in assigning the name of C. sowerbii to one or another clade is present here. For this marker it is possible to discuss about the relative genetic distances (uncorrected p-distances) that are recognized between species of another olindiid genus, the genus Olindias (see Bouillon et al., 2004Bouillon et al., : 206, 2006. Genetic distances between the three species of Olindias, from which 16S sequences are available, vary from 5.5-5.7 % (O. phosphorica to O. sambaquiensis) to 9.8-10.0 % (O. formosus to O. sambaquiensis) (see also Collins et al., 2005Collins et al., , 2008. The genetic distance observed between North African-Switzerland and American sequences identified as C. sowerbii is 4.2-4.5 %, between North African-Switzerland and C. ziguiensis and C. sinensis it is 7.8-8.9 % and 6.8-7.4 %, respectively; and between the latter two species it is 6.2 %. This suggests that American specimens attributed to C. sowerbii should perhaps be considered a different species from the specimens analysed here from North Africa, as well as from those from Switzerland.

Final remarks
The described scenario could be much more complicated when considering that in the type locality of C. sowerbii, the aquatic plants of the water-lily tank in Regent's Park (i.e. the potential dispersal vector of this hydromedusa species) were imported from Brazil, and not from China (C. Magdalena, pers. comm.). At present, it is difficult to know when the dispersion of Craspedacusta species from Easter Asia began, and most of the proposed vectors are in part speculative. Perhaps there was a combination of initial introduction by trade of aquatic plants and a subsequent natural dispersion by aquatic animals (e.g. birds, insects). For this reason, to solve this unstable nomenclatural and biodiversity problem, it is highly desirable to start with an important (or at least representative) field and molecular sampling programme in Southern England, in order to decide which haplotype (or set of related haplotypes) could be considered the true Craspedacusta sowerbii.
At this moment, for the present contribution, the most parsimonious solution would be the existence of a single species (clade) in England and the relatively close Central Europe (also shared by the Moroccan examined specimens). If this is correct, this clade should retain the specific epithet sowerbii, and then, as in the same group of sequences there are some attributed to C. kiatingi by Zhang et al. (2009), these sequences and individuals should also be assigned to C. sowerbii. According to these considerations, all materials included in the 'sowerbyi' clade of Fritz et al. (2009) and the clade that included all sequences attributed to C. sowerbii by Zhang et al. (2009) (including the Chilean and Sicilian sequences) should be assigned to a different species, which should be selected among the available names after a complete bibliographical and morphological review. It has also become clear that according to Cox1, at least three Craspedacusta species are present in Europe (see fig. 6, Cox1): one in central Europe (and Morocco), one in Greece and Sicily (for the moment ITS sequences from the Greek specimens are not available), and one in Switzerland (for the moment ITS information is not available). Unfortunately, no geographical information is currently available for a 16S sequence (KY077294, see Grange et al., 2017) and a Cox1 sequence (LN901194, see Kayal et al., 2015). In the future, knowing the origin of these and other additional sequences would provide important information on invasion events concerning this intriguing hydromedusa species.
The possible introduction vectors of Craspedacusta sowerbii in the recorded new sites generally remain unidentified. Accordingly, several hypotheses about the possible introduction paths have been mentioned in different reports and works on this species (Dumont, 1994;Angradi, 1998;Karaouzas et al., 2015). They can mainly be resumed in two possible vectors: 1) vectors facilitated by human activities, and 2) natural vectors. With regard to the former, the most likely dispersal vector may be the transfer of the species polyp stage, or the result of a resistance structure in aquaria or exhibition tanks, or an association with commercial ornamental aquatic plants or animals (Oscoz et al., 2010;Gasith et al., 2011;Gomes-Pereira and Dionisio, 2013;Minchin et al., 2016). The minuscule and hardly recognizable resting forms of the species make its unintentional human-mediated dispersal likely. Some authors reported observations of C. sowerbii (medusa or polyp stage) coinciding with the introduction of stocked fish or aquatic plants (Parent, 1982). As for the second possible vector, desiccated podocysts attached to body parts of aquatic invertebrates and vertebrates (including birds) could have allowed this species to colonize near freshwater reservoirs (Jankowski, 2001;. The cysts are able to survive for about 40 years while being completely desiccated (Bouillon and Boero, 2000;Bouillon et al., 2006;Lewis et al., 2012). These resting bodies may accidentally be transferred to new sites on bird's feet or plumage. Then, in favorable conditions, cysts turn into medusae and podocysts become polyps that can lead to more budding. This makes the aerial passive dispersal a possible introduction path for C. sowerbii (Parent 1982;Dumont 1994;Oscoz et al., 2010;Didžiulis and Żurek 2013;Failla-Siquier et al., 2017).
The characteristic drought-resistant forms of the species suggest that the natural aerial dispersal vector by migrating birds (see Reynolds et al., 2015;Green, 2016) may be an important factor in the introduction of alien species such as C. sowerbii into Bin El Ouidan, although in this reservoir, in order to limit the proliferation of algae as well as to enhance the biodiversity within the reservoir ecosystem, many exotic species including fish and aquatic plants (e.g. Oncorhynchus mykiss Walbaum, 1792; Barbus barbus Linnaeus, 1758; Micropterus dolomieu Lacepède, 1802; Sander lucioperca Linnaeus, 1758; Hypophthalmichthys molitrix Valenciennes, 1844; Ctenopharyngodon idella Valenciennes, 1844 and Cyprinus carpio Linnaeus, 1758) started to be introduced in the reservoir one year after its construction (Rabii Souilem, pers. comm.).