Reports of Baetidae (Ephemeroptera) species from Tafna Basin, Algeria and biogeographic affinities revealed by DNA barcoding

Abstract Background The Mediterranean basin is known to be the cradle of many endemic species. Within mayflies (Insecta, Ephemeroptera), North African species belonging to the family Baetidae remain poorly known and, traditionally, affinities to European fauna were proposed. Recent studies, based on molecular reconstructions, showed closer relationships to Mediterranean islands fauna. New information Baetidae were sampled from North-West Algerian wadis (Tafna basin) and involved in COI barcoding reconstructions. Seven species were identified. The subgenus Rhodobaetis is represented by Baetis atlanticus known previously from Macaronesian islands, Europe and Morocco and the Maghrebian endemic Baetis sinespinosus. Specimens, previously identified as Cloeon cf. dipterum, correspond to Cloeon peregrinator and, until now, only reported from Macaronesia. Besides the confirmation of endemicity of some species, such as Procloen stagnicola and B. sinespinosus, our molecular study showed quite original results for relationships between European, insular and Algerian species. Baetis maurus stood out as a North African endemic sister clade to an Iberian clade. Furthermore, we found clear interspecific distances between Algerian and European clades for A. cf. sinaica and B. cf. pavidus, suggesting the presence of cryptic species in Algeria. However, additional studies are needed, as, for the moment, no clear morphological characters were found to separate the different clades and support them as valid species.

Recently, molecular reconstructions involving Baetidae were conducted for different projects, in particular for the origin of Macaronesian and Corsican mayflies fauna (Gattolliat et al. 2015, Gattolliat et al. 2018b, Rutschmann et al. 2014, Rutschmann et al. 2017). Despite being not directly focused on North Africa, they included specimens from Tunisia, Algeria and Morocco. These preliminary results for North Africa underlined important links between North African and Macaronesian faunas (in the case of Cloeon and Baetis). The discovery in Tunisia of a species of Leptophlebiidae, assumed as endemic to Sardinia (Zrelli et al. 2011), also confirmed possible connections between Italy and Maghreb, mainly during crucial geological events, such as the Messinian Salinity Crisis (Gattolliat et al. 2015).
North African species of Labiobaetis and Cheleocloeon Wuillot and Gillies, 1993 have most probably an Afrotropical origin as they are mainly diversified in this area. A dozenspecies of Cheleocloeon are described in Afrotropics, while the genus is only represented in the Palearctic by a single Maghrebian species (Cheleocloeon dimorphicum) and one in the Arabian Peninsula, Cheleocloeon soldani , Kluge 2016a. Despite also being present in Central Europe, Labiobaetis is mostly diversified in tropical areas, as proven by its high diversity in Afrotropics (Gattolliat 2001, Lugo-Ortiz andMcCafferty 1997) and South East Asia (Kaltenbach and Gattolliat 2018, Kaltenbach and Gattolliat 2019, Kaltenbach and Gattolliat 2020. While a part of the Afrotropical species present a very restricted distribution (Gattolliat 2001, Lugo-Ortiz andMcCafferty 1997), a recent molecular reconstruction proved that specimens from Comoro Islands, South Africa and Arabian Peninsula form a monophyletic clade corresponding to Labiobaetis glaucus (Agnew, 1961) (Gattolliat et al. 2018a).
The present study is the first molecular analysis for Algerian mayflies using the cytochrome oxidase subunit I (COI) region for species delimitation. The main aims are to clarify the status of the different species of Baetidae present in North West Algeria, especially for Central European species assumed to occur in Maghreb. We also want to clarify the species delimitation in some groups with potential cryptic species and significantdifficulties to identify, based on morphological characters only. Finally, we want to understand the affinities between Maghrebian and neighbouring fauna.

Sampling
We investigated twelve sampling sites, all located in the Tafna basin in North-West Algeria (Fig. 1)

COI gene amplifications
We performed DNA extraction using DNeasy Blood & Tissue kit (QIAGEN) and BioSprint 96 extraction robot (Qiagen) by soaking each specimen in buffer and proteinase K at 56°C for an overnight incubation. The mitochondrial DNA cytochrome oxidase c subunit I gene (COI) was amplified using the primers LCO1490 and HCO2198 (Folmer et al. 1994) with an initial denaturation temperature of 98°C for 30 sec followed by a total of 37 cycles with denaturation temperature of 98°C for 10 sec, an annealing temperature of 50°C for 30 sec and an extension at 72°C for 30 sec, final extension at 72°C for 2 min. We checked if the amplification was successful using agarose gel electrophoresis, then we purified PCR products and prepared bi-directional sequencing using the same primers LCO1490 and HCO2198.

COI gene trees
We corrected and edited forward and reverse sequencing reads using Bioedit, then we assembled each of the two complementary sequences using Codon Code Aligner (demo mode) and obtained sequence alignments (Suppl. materials 4,5,6,7,8,9) using Jalview 2.10.1 via Mafft alignment as in Vuataz et al. (2011). We aligned sequences of each taxon with analogue genus or species selected from Genbank database (GenBank 2019) or BoldSystem database (BOLD 2019). For Rhodobaetis, we added sequences corresponding to the known haplogroups of Baetis rhodani (Gattolliat et al. 2015, Gattolliat et al. 2018b, Lucentini et al. 2011, Williams et al. 2006

Species delimitations
Based on the molecular analysis, seven Baetidae species were recognised or confirmed in the Tafna basin sites.

Baetis cf. pavidus
We obtained two haplogroups of Baetis cf. pavidus (Fig. 3, Table 3): a highly-supported clade gathering Algerian specimens and two sequences from southern France (BP_Gp1) and a second clade (BP_Gp2) which is composed by Baetis pavidus from Italy. Both haplogroups are highly supported as sister clades. BP_Gp1 possesses a very low intraspecific distance (0.2%) and high interspecific distance with its sister clade BP_Gp2 (11%).

Baetis maurus
We delineated three B. maurus haplogroups (Suppl. material 2, Table 4). Our sequences (BM_Gp1) belong to a strongly-supported monophyletic haplogroup identified as Baetis maurus (Fig. 4). This haplogroup is highly supported as the sister clade of specimens identified also as B. maurus (BM_Gp2 and BM_Gp3), but coming from Spain; the distances between these clades, all identified as B. maurus, are of interspecific level (15-16%).

Acentrella cf. sinaica
The reconstruction divided Acentrella sequences into 6 haplogroups (Fig. 5, Table 5). The single sequence from Algeria was recovered as an independent clade (AC_Gp5) distant at least 19% from other clades. AC_Gp6 haplogroup is the sister clade of AC_Gp5; it contains A. sinaica sequences from France and Italy gathered with a very strong BS (100%).  Maximum Likelihood tree including a representative of Baetis maurus using TN93 (+G+I) model; only bootstrap supports (BS) higher than 70% are indicated on the corresponding branch. Table 5.

Cloeon peregrinator
We obtained 11 haplogroups from the reconstruction (Suppl. material 3,  Maximum Likelihood tree including a representative of Acentrella spp using the General Time Reversible model (+I); only bootstrap supports (BS) higher than 70% are indicated on the corresponding branch. Table 6.
The CO_Gp2 haplogroup, which includes four sequences from Algeria and sequences of C. pereginator from Madeira (type locality) and Gran Canaria (Fig. 6), is highly supported as a monophyletic haplogroup (100%). This haplogroup has a low intraspecific distance (0.2%) and high interspecific distance with all the other haplogroups of C. dipterum s.l., for instance with CO_Gp1 with which it has the least distance (8.9%). Consequently, the sequences from Algeria and Gran Canaria belong to C. peregrinator.

Discussion
The different trees we obtained allowed us to better understand the composition of Algerian Baetidae. Based on our analysis, we can link the Algerian lineages with their sister-groups, calculate the maximum and minimum distances and evaluate which lineages may represent putative species. Table 7.

Figure 7.
Maximum Likelihood tree including a representative of Procloeon spp using the General Time Reversible model (+G+I); only bootstrap supports (BS) higher than 70% are indicated on the corresponding branch.

Baetis (Rhodobaetis) sinespinosus Soldán and Thomas, 1983
Rhodobaetis is a subgenus of Baetis and corresponds to the concept of Baetis rhodani species-group (Jacob 2003, Müller-Liebenau 1969. It is widely distributed in all West Palearctic streams (Gattolliat et al. 2015). Three species of Rhodobaetis are reported from Algeria, two of them are endemic: Baetis chelif and Baetis sinespinosus (Bauernfeind and Soldán 2012, Thomas 1998. A representative of B. rhodani s.l. is also reported (Thomas 1998). As noticed by Sroka (2012), it is generally difficult to find morphological characters to support the molecular species delimitation within Rhodobaetis. In the present case, the absence of a single rudimentary scale on the tip of the maxillary palp and the presence of four rows of setae at the apex of the paraglossae indicate that the clade RB_Gp2 corresponds to B. sinespinosus. Baetis sinespinosus is a well-supported monophyletic clade and presents high interspecific distances with all the other European and Mediterranean species. According to our data and reports from literature, this species seems to be endemic to Algeria. However, its presence in nearby countries, such as Tunisia and Morocco, will not be surprising as several populations were provisionally identified as Baetis rhodani s.l. (Mabrouki et al. 2017, Thomas 1998, Zrelli et al. 2016) and the present study is based on material collected very close to the border of Morocco.

Procloeon stagnicola Soldán and Thomas, 1983
Our results showed a high interspecific distance between the Algerian clade and its European sister species Procloeon bifidum; thus, it confirms the validity of Procloeon stagnicola. This latter differs from P. bifidum especially by the flat and rounded bristles on the labrum margin; the pointed apex of the gills with an extremely reduced second lamella and also by the lateral margins of the abdominal segments which possess spines from segment V to IX. The species was originally described from Algeria (Soldán and Thomas 1983a) and was subsequently discovered in Tunisia (Boumaiza and Thomas 1995). The report of Procloeon bifidum from Morocco (El Alami et al. 2000) may be a misidentification and may also concern P. stagnicola. Identification of material collected by the last author (Jean-Luc Gattolliat) and stored in the MZL collection confirmed the presence of this species in Morocco (unpublished data). This species should be therefore considered as endemic to Maghreb.

Baetis maurus Kimmins, 1938
Baetis maurus is a representative of the Baetis alpinus species-group. The species is considered as an Atlanto-Mediterranean element (Bauernfeind and Soldán 2012). It was originally described from Morocco (Kimmins 1938), then reported from the Iberian Peninsula (Alba-Tercedor 1982, Müller-Liebenau 1974. Algerian haplotypes present high distances with presumably conspecific specimens from Spain (Murria et al. 2017), as well as with the other Euro-Mediterranean species belonging to the B. alpinus species-group. Our results tend to prove that B. maurus is, in fact, a Magrebian endemic species (originally described from Morocco) and that at least one sister undescribed species occurs in the Iberian Peninsula. These preliminary results must be confirmed by sequencing additional populations from Spain and Maghreb and by morphological evidence. With the presence of a second rudimentary row of denticles on claws, both Maghrebian and Iberian populations possess a unique morphological character amongst the genus Baetis (Thomas et al. 1983).

Acentrella cf. sinaica Bogoescu, 1931
Acentrella sinaica is a South and Central European species (Bauernfeind and Soldán 2012). It was reported from Tunisia (Boumaiza and Thomas 1995) and then from Algeria (Mebarki et al. 2017). In his checklist of North African mayflies, Thomas (1998) considered the specific identification as possibly incorrect, referring maybe to Acentrella almohades Alba-Tercedor & El Alami 1999 described from Morocco (Alba-Tercedor and El Alami 1999). The important distances between our unique haplotype and those from France and Italy tend to confirm that Algerian specimens do not belong to A. sinaica. Unfortunately, no sequence is available for A. almohades. We refrain to attribute the specimens from Algeria to A. almohades, as important distinctive characters do not match between our specimens and the original description. Especially, the Algerian specimens possess long setae along the dorsal margin of the femora (similar to A. sinaica), while A. almohades present much shorter ones (Alba-Tercedor and El Alami 1999). There is, therefore, some probability that our specimens represent an undescribed species from North Africa. Additional sequences and close morphological studies are needed to confirm this hypothesis.

Baetis cf. pavidus Grandi, 1949
Baetis pavidus is a Western Mediterranean species and was originally described from Italy and then reported from the Maghreb (Bauernfeind and Soldán 2012, Jacob 2003. In Algeria and Tunisia, it is one of the most common and abundant species at low to middle elevation and is rather tolerant to pollution and low oxygenation (Benhadji et al. 2019, Boumaiza andThomas 1995). The Algerian haplotypes present low genetic distance with specimens from southern France, confirming the link for this species between North African and South European populations. The sequences from Sicily, Italy (BP_Gp2) were not assigned to Baetis pavidus with certainty by  and may represent an undescribed species close to B. pavidus or, alternatively, may be correctly associated with B. pavidus and the specimens from Algeria and from South of France represent a new species.

Cloeon peregrinator Gattolliat and Sartori, 2008
Cloeon peregrinator was first considered as an endemic species from Madeira , then was also found on the Canary Islands (Rutschmann et al. 2017). It belongs to the Cloeon dipterum species-group from which it can be separated by minute morphological characters . Algerian haplotypes present a low genetic distance with Madeiran specimens which prove their conspecificity. This discovery is rather surprising as C. peregrinator was first thought to be an insular endemic. Molecular studies showed mayflies are able to colonise islands, even for such taxa with presumably low dispersal capacity (Monaghan et al. 2005, Rutschmann et al. 2014. The origin of the species cannot yet be proven and two scenarios can be proposed: either a colonisation of Macaronesia from Continental Europe, then a speciation process on the islands and a subsequent colonisation of North Africa or, alternatively, a colonisation of North Africa by a European lineage followed by a speciation process and subsequently a colonisation of Macaronesia.

Baetis atlanticus Soldán and Godunko, 2006
Bauernfeind and Soldán (2012) (Soldán and Godunko 2006): larvae prefer coarse substrate composed of rocks, cobbles or pebbles in low, moderate to fast current velocities; they were less abundant and only present in relatively-preserved sites at higher altitude (CH0; CH1; KH1; SK1; IOM). In comparison, B. sinespinosus larvae have a much wider ecological range and are more pollution resistant; they are also highly abundant and present in all sampling sites with various substrates and velocities.

Concluding remarks
We summarise the state of the knowledge and the implication of the present study for the Baetidae fauna of North-West Algeria in Table 8. As we mostly found interspecific distances between Algerian and European lineages, our results generally highlighted and confirmed the high endemism of North African Baetidae. According to the present knowledge, the endemism may be restricted to Algeria or to the Maghreb, even for species that were supposed to present West Palearctic distribution (Western Europe and North Africa). The link with the Iberian Peninsula is less strong than expected, as no species included in the study is shared only between the two areas. From a genetic point of view, only one case of sister-species was found (Baetis maurus). Baetis cf. pavidus is the only species with a West Mediterranean distribution, as the same species occurs in Algeria and South of The next steps will be to sequence more specimens from different areas of Algeria and also from Morocco and Tunisia to confirm the monophyly of the different North African clades. The results, especially the validation of the new species hypotheses, need to be confirmed by integrative methods. Only morphological evidence and more mitochondrial or nuclear genes can validate the specific status of these clades. Our study may have implications outside of North Africa, as our results suggest that one or two lineages, previously supposed to belong to Baetis maurus, may represent new species in Spain, as well as the presumably non-conspecificity of the French and Italian lineages of Baetis pavidus.