Taxonomic status of Macaronesian Eucyclops agiloides azorensis (Arthropoda: Crustacea: Copepoda) revisited – morphology suggests a Palearctic origin

Macaronesia, with the exception of the Azores, is one of the few Palearctic provinces where basic taxonomic information on the freshwater copepods is still lacking. We redescribed Eucyclops azorensis, a cyclopid crustacean so far known only in the Azores, and report the occurrence of this species in Algeria and Madeira Island. Eucyclops azorensis was formerly considered to be a subspecies of E. agiloides (East Africa); therefore, the latter species is redescribed here as well based on type and nontype material. Morphological comparisons between E. azorensis, E. agiloides and other taxa (E. serrulatus and E. roseus), assumed to be closely related to our target species, support a closer relationship between E. azorensis and E. serrulatus (Palearctic) than between E. azorensis and E. agiloides (Afrotropical). The slight differences between E. azorensis and E. serrulatus in the surface ornamentation of the antennal coxobasis and intercoxal sclerites of legs 1 and 4 suggest a relatively young separation of these lineages. Eucyclops agiloides is morphologically close to E. roseus (temperate and subtropical Asia, southeastern Europe, East Africa). The numerous and clear-cut differences indicate a species rather than subspecieslevel differentiation between E. agiloides and E. roseus, unlike what was formerly proposed in the taxonomic literature.

Eucyclops Claus, 1893 (~ 120 species and subspecies) is one of the largest genera in the Cyclopidae Rafinesque, 1815, yet except for a few species (e.g., the parasitic E. bathanalicola Boxshall & Strong 2006 in Lake Tanganyika), the genus shows a surprisingly uniform limb segmentation and setation pattern. Diagnostic characters of the species and species groups (e.g., the serrulatus group and numerous subgenera currently defined by Alekseev 2019) overwhelmingly relate to the surface ornamentation of the body and appendages, and the body proportions (see Alekseev & Defaye 2011;Mercado-Salas et al. 2016). Mouthpart characters have received less attention in the systematics of Eucyclops (but see Ishida 2002;Chang 2009;Tang & Knott 2009). Morphometric features (e.g., relative length and robustness of the appendages of leg 5, and the length and width proportion of the caudal rami), emphasized especially in the older taxonomic works, show significant intraspecific variation in some species (e.g., in E. serrulatus − see Alekseev et al. 2006). Also, our poor understanding of the phylogenetic polarity of the morphological traits in the genus hinders a meaningful grouping of the species within Eucyclops (but see Gaponova & Hołyńska 2019 for outgroup comparisons of some morphological characters).
In a project on the microcrustaceans of Algeria, conducted by the laboratory of Dr Amarouayache in 2012−2016, two of us (SG & MA) found an Eucyclops in northeast Algeria (Tébessa and Souk-Ahras Provinces), which was identified as Eucyclops leschermoutouae Alekseev & Defaye 2004(Ghaouaci et al. 2017Ghaouaci 2018). The Algerian Eucyclops shared several surface-ornamentation characters (on the A2 coxobasis, P4 intercoxal sclerite and coxopodite setae) with E. leschermoutouae (Balearic Islands: Mallorca), yet differed, among other things, in the length proportion of the terminal caudal setae VI and III. Comparisons of the paratypes of E. azorensis (originally named as E. agiloides azorensis) and additional specimens collected by one of us (ŁS) in Madeira helped to clarify the taxonomic status of the Algerian Eucyclops, and also revealed an interesting Macaronesian−North African distribution pattern in this species. Eucyclops azorensis was considered by Defaye & Dussart (1991) as a subspecies of E. agiloides (G.O. Sars, 1909) originally described from East Africa. Therefore, we have also examined the types (from Lakes Victoria and Tanganyika) and additional specimens (from Lake Tanganyika) of E. agiloides, and redescribed the latter species as well. The morphology of E. agiloides remains poorly understood, although the species has been recorded in several sites in tropical Africa and Asia, and is currently associated with three other taxa (E. azorensis, E. roseus Ishida, 1997 andE. miracleae Alekseev, 2010) in subspecies relationships (Alekseev & Defaye 2011;Alekseev 2019).
A better understanding of the taxonomic relationships of E. azorensis and E. agiloides has implications for both systematics of the genus and future studies on the origin of the freshwater cyclopid fauna in Macaronesia.

Material and methods
Collection information on the specimens examined is provided in the species descriptions. Unless otherwise stated, all specimens were fully dissected. Drawings were made using a camera lucida attached to Olympus BX 50 compound microscope. Telescoping somites were measured separately and summed for total body length (Koźmiński 1936). For scanning electron microscopy, whole specimens were dehydrated in graded acetone followed by hexamethyl-disilazane (HMDS) drying. Surface structures were investigated using a Hitachi S 3400 N (Museum and Institute of Zoology PAS, Warsaw) scanning electron microscope.   Table 1. Prosomal and urosomal somites often adorned with shallow pits (Fig. 1B, D). Pediger 4 with fine hairs at posterolateral angle and small spinules on lateral margin (Figs 2A, 3A). Pediger 5 posterolaterally bearing long and robust hairs. Genital double-somite as long or slightly shorter than wide (Table 1). Single large copulatory pore oval shaped, located ventrally in anterior fifth of somite. Seminal receptacle as common in genus (cf. Fig. 8B), anterior and posterior parts short in length, posterior part wider than anterior. Anal operculum nearly straight, anal sinus with 1-1 longitudinal row of spinules. Posterior margin of anal somite with continuous row of spinules. Caudal rami 3.5−6.0 times as long as wide (Defaye & Dussart 1991, see also Table 1 for interpopulation variation), no hairs on medial margin. ʻSerraʼ (longitudinal row of spinules) on lateral margin extending from insertion of anterolateral (II) caudal seta to anterior ⅓−⅕ of rami, more anterior spinules shifted to ventral surface and can be unnoticed in dorsal view (Fig. 1C). Spinules present at insertion of posterolateral (III) caudal seta. Single pore present on ventral surface near lateral margin, slightly posterior to midpoint of ramus. Seta III with fine setules medially and short spinules laterally. Dorsal (VII) caudal seta shorter, while terminal accessory (VI) seta longer than posterolateral caudal seta (Table 1) Antenna. Composed of coxobasis and three-segmented endopodite, and bearing 3, 1, 9 and 7 setae, respectively. Exopodite seta reaching distinctly beyond enp3, and bearing long setules proximally and short setules more distally (Fig. 2C). Coxobasis (Figs 2C, 3B) caudally ornamented with: hair-like spinules on proximolateral margin; longitudinal double rows near lateral margin, spinules in distal row sometimes distinctly smaller than those in proximal row; few small spinules near insertion of exopodal seta; transverse row at height of insertion of medial setae; oblique row below insertion of medial setae; and one or more groups of tiny spinules more proximally in medial half of segment. Spinules absent near distal margin. Frontal surface of antennal coxobasis (Figs 2D, 3C) with longitudinal row of spinules (6−15) along lateral margin and one or two oblique rows of long spinules and small spinules near proximal margin. Long hair-like spinules absent on distal and mediodistal margin.

Morphological
Mouthparts. Labrum (cf. Fig. 8G). Distal margin with 9−12 teeth (some of those tiny), small spinules present on laterodistal lobes. Distal fringe hairs arranged in 1-1 group. Tiny spinules present on lateral margin at height of posterior end of epistoma in Madeiran and Algerian females − this character could not be verified in paratype specimens. Paragnaths with 4 (3+1) medial claws, as common in Cyclopidae. Mandible ( Fig. 2E) with palp bearing two long and one short setae. Near palp transverse row of long spinules and smaller spinules arranged in oval pattern present on anterior surface of coxal gnathobase. Maxillule setation as common in Cyclopidae. Maxillulary palp (Figs 1E, 2F) naked, with one proximal and three apical setae, lateral lobe bearing three setae. Armature of maxilla (cf. Fig. 9D Defaye & Dussart, 1991. Data in bold are from the original description of the species (Defaye & Dussart 1991). Numbers in subscript refer to the numbers of specimens measured.
Male (Fig. 1H) Only those features are mentioned, which differ from the corresponding character states in female.

3.
n/a n/a n/a n/a 4.  Table 3. Morphometric characters of the male in Eucyclops azorensis Defaye & Dussart, 1991. Data in bold are from the original description of the species: with the exception of the total body length, all indices were measured on the drawings provided by Defaye & Dussart (1991). n/a = non applicable.

Description
Female Habitus. Total body length 0.9−1.1 mm (for morphometric characters see Table 4). Pediger 4 and pediger 5 posterolaterally bearing long fine hairs and thick (spinule-like) hairs, respectively (Fig. 8A−B) -pilosity of pediger 4 could only be verified in females from Kirando, Lake Tanganyika. Genital double-somite (Fig. 8B) as long or slightly shorter than wide. Seminal receptacle as common in genus (Fig. 8B), anterior and posterior parts short in length, posterior part wider than anterior, single large copulatory pore in anterior fifth of somite. Anal operculum (Fig. 8D) nearly straight, anal sinus with longitudinal rows of hairs, posterior margin bearing continuous row of robust spinules. Caudal rami (Fig. 8D) 4.0−6.0 times as long as wide, no hairs on medial margin. Serra extending from insertion of anterolateral (II) caudal seta to anterior ⅕−⅙ of rami. Seta II inserted in posterior ¼−⅕ of caudal ramus. Spinules present next to insertion of posterolateral (III) caudal seta. Relative length of caudal setae in Table 4.
Antenna. Composed of coxobasis and three-segmented endopodite, and bearing 3, 1, 9 and 7 setae, respectively. Exopodite seta reaching slightly (paratype, Lake Victoria) or distinctly (Kirando, Lake Tanganyika) beyond enp3, proximal setules can be similar in length or distinctly longer than setules in distal section of seta (Figs 7B, 8F). Caudal surface ornamentation of antennal coxobasis (Figs 7B,8E) with same groups of spinules as those in E. azorensis in lateral half of segment, except for additional group of spinules next to distal margin (group marked with arrow in Fig. 7B). Medial half of segment with reduced ornamentation: few spinules sometimes present at height of insertion medial setae (Fig. 7B), or below setae (Fig. 8E). Frontal surface (Figs 7C, 8F) adorned with spinule-like hairs next to distal margin, hairs absent mediodistally, longitudinal / oblique row of spinules (8−13) along lateral margin, and oblique rows of large and small spinules near proximal margin.

Male
Unknown.
Java (Kiefer 1933), and Borneo (Alekseev et al. 2016). Part of these records may refer to other taxa (e.g., E. roseus), and the geographic distribution of the species is still poorly understood (Fig. 11 shows only the verified records). The same holds true for the habitat preferences of E. agiloides. The specimens examined here were collected in large lakes and a river mouth in Lake Tanganyika, yet precise information on the collection sites is missing. Defaye & Dussart (1991) described a Eucyclops from the Azores under the name E. agiloides azorensis. They noted that the new subspecies was morphologically identical with the cyclopid identified by Lindberg (1962) as Eucyclops serrulatus in the archipelago (São Miguel, Santa Maria, Pico, and Flores Islands). Defaye & Dussart (1991) argued for a close relationship between the Azorean Eucyclops and African E. agiloides, based on a few characters that showed similar states (short ʻSerraʼ of caudal rami; slender medial spine of leg 5; lack of aesthetasc on the 9 th antennulary segment; smooth hyaline membrane on 10 th to 12 th antennulary segment; and length and width proportion of terminal antennulary segment ~ 7) in these two taxa, but different in E. serrulatus. Later studies (Alekseev et al. 2006;Alekseev & Defaye 2011), however, revealed considerable variation in the morphology of the P5 medial spine in E. serrulatus, while variation in the extension of ʻSerraʼ of the caudal rami in E. azorensis is documented herein (see Results). The aesthetasc on the 9 th antennulary segment, albeit relatively small and sometimes not easy to notice, does occur in both E. azorensis and E. agiloides (Figs 2B,7A). ʻSmoothnessʼ of the hyaline membrane on the 10 th to 12 th antennulary segment is an elusive character, as it may depend on the microscope magnification applied. The length and width proportion of the terminal antennulary segment shows significant intraspecific variation and overlap between species (Table 5), although E. serrulatus indeed has a range with somewhat higher values in comparison to E. azorensis and E. agiloides.

Discussion
Understanding of the taxonomic relationships of the Azorean Eucyclops has also been impeded by the relatively poor knowledge of the morphology of E. agiloides sensu stricto. Sars (1909) originally described the species from East Africa (Lakes Victoria and Tanganyika), and since then E. agiloides has been reported from several regions in Africa, (sub)tropical Asia and southeast Europe (for more details see ʻGeographic distributionʼ of the species in the Results). Alekseev & Defaye (2011) provided a short diagnosis and a few drawings of E. agiloides, based on specimen(s?) from Lake Malawi, mistakenly considered as the type locality. The lateral hair ornamentation of P4 exopodite (exp1−exp3 are pilose) and the presence of hairs in the full width of the free margin of P4 intercoxal sclerite (see Alekseev & Defaye 2011: fig. 7b), however, might indicate that at least the fourth leg illustrated there belongs to E. roseus rather than to E. agiloides sensu Sars (cf. Figs 7E−F, 10E−F, and Table 5). In that same paper Alekseev and Defaye changed the taxonomic rank of E. roseus (subtropical and temperate Asia, East Africa and southeast Europe) and E. miracleae (Spain) from species to subspecies (see Alekseev & Defaye 2011: 63). This decision has been confirmed in a current publication of Alekseev (2019), in which he listed four subspecies of E. agiloides (E. a. agiloides, E. a. roseus, E. a. miracleae and E. a. azorenis).
In Table 5 we compare 14 characters in the females of four taxa (E. serrulatus, E. azorensis, E. agiloides and E. roseus) to infer the taxonomic affinity of E. azorensis. Two character states (see character 4 in E. azorensis and character 6 in E. agiloides) occur in one species only. Characters 13 and 14 are intraspecifically variable and largely overlap between species; therefore, they are not informative of the morphological relationships of these taxa. Character 2 (presence / absence of spinules at height of the medial setae on the caudal surface of the antennal coxobasis) is an ambiguous character: the spinules are present (without variation) in E. serrulatus and E. azorensis, while the character is polymorphic in E. agiloides and E. roseus. Of the nine informative characters, six (1, 3, 7, 8, 10, 11) support a closer relationship between E. azorensis and E. serrulatus, while three (5,9,12)  0-1; 0-0 3 ; 0-0; 0-0 0-1; 0-0; 0-0; 0-0 0-1; 0-1; 0-1; 0-1 1 / 0 4 -1; 1-1; 1-1; 1-1 12. P4 intercoxal sclerite free margin, hairs in middle section present absent / sparse absent present / absent 13. A1 terminal segment, length / width 5.8−7.5 4.3−6.3 5.0−7.1 4.9−6.9 14. P4 coxopodite seta, lateral margin, setulation prox. abs. / discont. prox. abs. / discont. / cont. 5 prox. abs. / cont. prox. abs. / cont. Table 5. Comparison of the diagnostic morphological characters in the female of Eucyclops azorensis Defaye & Dussart, 1991 and the putative close relatives. Character states were verified in all dissected specimens listed in ʻMaterial examinedʼ and ʻComparative materialʼ (see the species descriptions). 1 = spinules are present in one female, Ribeira de São João, Madeira; 2 = P1 intercoxal sclerite is likely naked on the caudal surface (verified in frontal view) in two females from Poland; 3 = P2 exp2 with lateral hairs in Alekseev & Defaye 2011 (fig. 5b); 4 = no hairs on P1 exp1 in one female, Anga River, Siberia; 5 = continuous setulation in one female from Ribeiro de Frio, Madeira; abs. = absent; cont. = continuous; discont. = discontinuous (with ʻgapʼ); prox. = proximally. '0' and '1' (character 11) code 'hairs absent' and 'hairs present', respectively. E. serrulatus. Eucyclops azorensis differs from E. serrulatus in a few features only: distal hairs are absent on the frontal surface of the antennal coxobasis (vs hairs present in E. serrulatus); P1 intercoxal sclerite is frontally naked or bearing just a few tiny spinules (vs spinules present in E. serrulatus); and hairs are absent / sparse in the middle section of the free margin of P4 intercoxal sclerite (vs whole width pilose in E. serrulatus). Nonetheless, a sister relationships between E. serrulatus (Palearctic, as far as Yenisey River in Siberia − Alekseev & Defaye 2011), and E. azorensis (Azores, Madeira and North Africa) could only be confirmed after a thorough comparison of the Western Mediterranean forms, such as E. hadjebensis (Kiefer, 1926) from North Africa, E. albuferensis Alekseev 2008, E. romaniensis Alekseev 2010 and E. miracleae from continental Spain, and E. leschermoutouae from the Balearic Islands (Mallorca). With the exception of E. leschermoutouae all the other species mentioned above, unlike E. azorensis, possess hair-like spinules next to distal margin on the frontal surface of the antennal coxobasis (Alekseev & Defaye 2004;Alekseev 2008Alekseev , 2010. Eucyclops leschermoutouae differs from E. azorensis in the length proportion of caudal setae VI and III (seta VI / seta III < 1 in E. leschermoutouae, while seta VI / seta III > 1 in E. azorenis), and the relative length of the lateral terminal spine of P4 enp3 (nearly as long as P4 enp3 in E. leschermoutouae, vs distinctly shorter in E. azorensis) (comparisons were based on Alekseev & Defaye 2004). Applying the subgeneric classification currently proposed by Alekseev (2019), Eucyclops azorensis would belong to the subgenus Speratocyclops Alekseev, 2019, while the morphologically close E. serrulatus and all the Western Mediterranean species mentioned above, with the exception of E. leschermoutouae, would be members of the subgenus Eucyclops s. str. We hesitate to accept such a grouping. The subgenera Speratocyclops and Eucyclops s. str. are distinguished from each other by a single character in fact (character 4 in Table 5). A natural classification should reflect the evolutionary relationships; therefore, it would be desirable to conduct a comprehensive analysis of the phylogenetic relationships, or at least to discuss the distribution of numerous characters within and outside the genus (outgroup comparisons), before proposing a subgeneric classification of Eucyclops.
Concerning the taxonomic position of E. roseus, the morphological differences between this taxon and E. agiloides in the surface ornamentation of the antennal coxobasis, mandible, maxillulary palp, P1 intercoxal sclerite and P1−P4 exopodites (Table 5: characters 5−7, 9 and 11), as well as the sympatric occurrence of these forms in Lake Victoria (Sars 1909;Ishida 1998), indicate a species, rather than subspecies, level differentiation between these taxa. Nevertheless, to corroborate this hypothesis more information on the intraspecific variation within E. agiloides in tropical Africa is definitely needed.