The Iberian Peninsula presents an insular freshwater ichthyofauna as a consequence of its hydrological isolation since the formation of the unsurmountable barrier represented by the Pyrenees during the Oligocene and the opening of the Strait of Gibraltar in the Miocene 5.33 million years ago (Krijgsmann et al., 2018). This is especially noticeable for primary freshwater fish families, such as: Nemacheilidae, Cobitidae, Leuciscidae and Cyprinidae. That is, those fishes whose ancestors entered inland waters much earlier and cannot survive in seawater, being strictly confined to freshwater systems (Myers, 1938; Darlington, 1948).

The insularity of the Iberian freshwater ichthyofauna is manifested in its faunistic composition with few genera, due to the difficulty of colonizing the Peninsula by different lineages and by a high number of endemic species as a consequence of the several hydrological reorganizations that have occurred since the Oligocene within the Iberian Peninsula (Perea et al., 2016).

The genus Squalius Bonaparte, 1837 is one of the genera with a large number of endemic species in the Iberian Peninsula (Doadrio et al., 2011). This genus is currently represented in the Iberian Peninsula by two different lineages, one formed by ancient Mediterranean endemisms and the other by a lineage from Central and East Europe and the northern areas of the Mediterranean basin towards Minor Asia, described as Euroasiatic (Sanjur et al., 2003). Divergence between both Euroasiatic and Mediterranean lineages should have occurred in the Late Miocene, approximately 7 Ma (Sanjur et al., 2003). The Euroasiatic lineage is represented in Spain by an autochthonous single species, Squalius laietanus Doadrio, Kottelat & Sostoa, 2007, distributed next to the Pyrenees, in the northeast of the Iberian Peninsula and in the southwest of France.

The Mediterranean lineage is much more diverse in the Iberian Peninsula and some species have aroused the interest of evolutionary biologists in recent years. Currently, nine species are recognized, all of them endemic to the peninsula: Squalius alburnoides (Steindachner, 1866), Squalius aradensis (Coelho, Bogutskaya, Rodrigues & Collares-Pereira, 1998), Squalius carolitertii (Doadrio, 1988), Squalius castellanus Doadrio, Perea & Alonso, 2007, Squalius malacitanus Doadrio & Carmona, 2006, Squalius palaciosi (Doadrio, 1980), Squalius pyrenaicus (Günther, 1868), Squalius torgalensis (Coelho, Bogutskaya, Rodrigues & Collares-Pereira, 1998) and Squalius valentinus Doadrio & Carmona, 2006. One of these species, S. alburnoides is unique within the European leuciscid fauna due to its hybrid origin, with specimens of different ploidies and reproduction by hybridogenesis (Carmona et al., 1997; Cunha et al., 2004, 2011).Probably S. palaciosi, of which specimens with different ploidies have also been described, was a species of hybrid origin but seems to have been extirpated since the end of the 20^th century (Doadrio et al., 2011). The nomenclature of these two hybrid-origin species is controversial, as they result from ancient hybridizations between one species of the Squalius genus and species belonging to different lineages than Squalius. In this context, we follow the latest nomenclature revision of the group (Collares-Pereira & Coelho, 2010) to provide nomenclatural stability, but it requires further review.

A genetic study based on nuclear and mitochondrial gene sequences on the populations from southern drainages in Spain found two well different lineages within S. malacitanus (Perea et al., 2016). These two lineages, one inhabiting the Atlantic slope and the other the Mediterranean slope, were geographically separated with the reorganization of the Iberian southern drainages which occurred during the opening of the Gibraltar Strait 5.33 Ma (Krijgsmann et al., 2018). Only the population from the Miel River, belonging to the Atlantic lineage, drains to the Mediterranean Sea as a consequence of secondary contacts during the Quaternary (Perea et al., 2016). Divergence between these two lineages of S. malacitanus occurred in the Mio-Pliocene period around 4.9 Ma (2.1–8.5 Ma) (Perea et al., 2016). A posterior work extended the genetic study of the genus Squalius to all populations from the Iberian Peninsula and, by analyzing six nuclear genes, corroborated the existence of these two Atlantic and Mediterranean lineages within S. malacitanus (Perea et al., 2020, 2021). In addition, populations of S. pyrenaicus were not monophyletic based on nuclear markers and three independent lineages were recognized: a Northern clade distributed throughout the Tajo Drainage and ascribed to the nominal species S. pyrenaicus; a Sado clade inhabiting the Sado drainage in Portugal; and a Southern clade distributed in the remaining Iberian drainages (Perea et al. 2021). Genomic data also corroborated the existence of these three independent lineages (Mendes et al. 2021). Phylogenetic relationships among Squalius species, based on nuclear genes, have shown that S. pyrenaicus is not monophyletic and the Northern clade is a sister group of S. carolitertii and S. castellanus with respect to one group formed by the Southern clade, the Sado clade, S. valentinus and S. malacitanus (Perea et al. 2020, 2021). To determine whether the lineages of S. pyrenaicus from the southern Iberian Peninsula and Sado, and the two lineages of S. malacitanus (Atlantic and Mediterranean), could be identified as distinct species, Perea et al. (2020) conducted a Bayesian nuclear multilocus species delimitation analysis following the MSC framework (Yang & Rannala, 2010; Yang, 2015). They utilized 4871 base pairs obtained from six nuclear loci plus 1140 base pairs of one mitochondrial locus. The analysis yielded robust support, with a posterior probability of 1.00 for these lineages, endorsing the hypothesis that these lineages should indeed be considered as separate species.

In this paper we present a formal description for the two independent lineages inhabiting Spain: the Southern lineage of S. pyrenaicus and the Atlantic lineage of S. malacitanus.

Morphology

The morphometric and meristic study of the genus Squalius was based on the analysis of 58 specimens belonging to the Mediterranean lineage of S. malacitanus, 42 specimens from the type locality (Table 1). Ninety-eight specimens of the Squalius population previously assigned to the Atlantic lineage of S. malacitanus (Perea et al. 2020, 2021). Eighty-one specimens belonging to the Northern lineage of S. pyrenaicus. One hundred twenty-five specimens of the Squalius population previously assigned to the Southern lineage of S. pyrenaicus (Perea et al. 2020, 2021) (Table 1.).

Holotypes and paratype series of the two new species have been deposited in the Museo Nacional de Ciencias Naturales (MNCN-CSIC, Spain).

Twenty-four morphometric measurements (in mm) and eleven meristic variables were recorded from digital photographs using TpsDig ver. 1.4 (Rohlf, 2003). The following abbreviations were used for morphometric and meristic characteristics: TL, total length; SL, standard length; PrDD, predorsal distance; PrPD, prepectoral distance; PrVD, preventral distance; PrAD, preanal distance; APL, anal peduncle length; CPL, caudal peduncle length; HL, head length to opercular; PrOL, preorbital length; ED, eye diameter; PsOL, postorbital length; NL, head length to nape; HH, head high; PmxL, premaxilla length; PFL, pectoral fin length; VFL, ventral fin length; DFL, dorsal fin length; DHL, dorsal fin height; AFL, anal fin length; AHL, anal fin height; CFL, caudal fin length; BD, body depth; BLD, body least depth; LLS, lateral line scale rows; SRA, scale rows above lateral line; SRB, scale rows below lateral line; D, dorsal fin rays; A, anal fin rays; P, pectoral fin rays; V, ventral fin rays; C, caudal fin rays; RPT, right pharyngeal teeth; LPT, left pharyngeal teeth; Vr, Vertebrae. After constructing the measurement matrix, Burnaby’s method was used to correct for size effect. The Burnaby method removes the effects of a within population size-factor from between-group morphometric analyses through an orthogonal projection procedure (Burnaby, 1966). All analyses were conducted with the corrected matrix.Morphometric and meristic characters were analysed independently. To identify the variables that contributed most to the variation among populations, one principal component analyses (PCA) was performed using the covariance matrix for morphometric characters. Row-wise bootstrapping was carried out to 100.000 replicates and 95% bootstrap confidence intervals are given to the eigenvalues. Statistical analyses were carried out using Past ver. 4.12 (Hammer et al., 2001), we used the option scree plot to indicate the number of significant components.

Osteology

Osteological features were investigated through computer tomography (CT) scan and digital dissection using VGStudio MAX ver. 2.2 (Volume Graphics, http://www.volumegraphics.com) of the following localities:

Additionally dry skeletons preserved in the MNCN collections were studied from the following localities:

Institutional acronyms: MNCN_ICTIO, Ichthyological Collection, Museo Nacional de Ciencias Naturales (Spain).

Genetic analyses

Genetic analyses of Iberian Squalius have been the focus of previous studies to resolve their phylogenetic relationships and biogeography with mitochondrial and nuclear genes (Sanjur et al., 2003; Perea et al., 2016, 2020, 2021). For this reason, the Iberian populations of the genus Squalius are well studied group from the phylogenetic point of view. In this study we reanalyse a data set of 268 sequences for the mitochondrial cytochrome b gene (MT-CYB, 924 pb) obtained from previously published sequences (see Appendix 1 for GenBank accession numbers). Two different phylogenetic analyses were performed using Bayesian inference (BI), implemented in MrBayes ver. 3.2 (Ronquist et al., 2012), and Maximum Likelihood, carried out in the IQ-tree online web server from the Vienna University (http://iqtree.cibiv.univie.ac.at/; Trifinopoulus et al., 2016). ModelFinder, implemented in the previous IQ-Tree web server (Kalyaanamoorthy et al., 2017) and the Bayesian Information Criterion (Schwarz, 1978) were used to estimate the evolutionary model that best fitted the data. The selected evolutionary model was TNF+F+G4. The Bayesian analysis was performed with two simultaneous independent runs each with four Markov chain Monte Carlo (MCMC), which were run for 5 × 10⁷ generations. The first 25% of generations were removed as burn-in. Posterior probability (pp) values were used to assess the reliability of the phylogenetic hypothesis. The accuracy of the Maximum Likelihood phylogeny was evaluated with the UltraFast Bootstrap method (1000 replicates) (Minh et al., 2013). For nuclear phylogeny, we derived a concatenated tree topology from the phylogenetic tree presented by Perea et al. (2020), based on the analysis of six nuclear genes and 4871 bp including gaps (Bayesian Inference and Maximum Likelihood analysis; GenBank Accession Numbers: MT-CYB: MT008486–MT008603; RAG1: MT008604–MT008704, S7: MT00855–MT008805, EFA1α: MT008910-MT009018, EGR2b: MT051740–MT051843, RHO: MT008806–MT008909, ACTB: MT051635–MT051739). We also calculated the uncorrected p-distances and verify the presence of autapomorphies among Squalius populations studied for the MT-CYB gene usingMega X (Kumar et al., 2018) to sequences download from the GeneBank data base (Appendix 1).

The principal component analysis to all populations and species divided the studied species of Squalius into four groups corresponding to populations of: a) S. pyrenaicus from the Northern Lineage (Tajo Drainage); b) S. pyrenaicus from the Southern Lineage (Guadalquivir, Guadalhorce and Vélez Drainages); c) S. malacitanus from the Mediterranean Lineage (Guadalmina and Guadiaro Drainages); and d) S. malacitanus from Atlantic Lineage (Jara drainage). All populations of Squalius pyrenaicus and all populations of Squalius malacitanus were considered in different groups without overlapping and the same occurred with the Northern and Southern lineages of S. pyrenaicus (Fig. 1). On the contrary, Atlantic and Mediterranean lineages of S. malacitanus showed a wide overlapping (Fig. 1). This overlapping between the Mediterranean and Atlantic lineages of S. malacitanus was similar to that found between populations from different rivers within the same species. A certain arrangement can be observed between the populations from the Almonte and Jerte Rivers, within the Northern Lineage, which live in rivers with different typologies. The Almonte River is a Mediterranean-like river influenced by severe water stress during the summer, with specimens of S. pyrenaicus surviving in disconnected pools.On the contrary, the Jerte is a mountain river with a permanent flow throughout the year. It can also be observed in the Southern Lineage a certain arrangement separating the populations from Vélez and Guadalquivir drainages.

The eigenvalues of the two first principal components, with the Burnaby-corrected matrix, explained most of the variance (Table 2).

The highest values for eigenvectors, and consequently the variables that contributed most to the ordination in the PCA, were the measurements related with the different proportions of the head (Table 2) as occur in other Iberian species of the genus Squalius (Doadrio, 1988; Doadrio & Carmona, 2006; Doadrio et al., 2007a, 2007b). The length and height of the caudal peduncle was not as decisive in the ordination of the populations as in other morphological studies of Squalius species (Doadrio & Carmona, 2006).

To clarify the variables that in each species contribute most to the ordination of the PCA we conducted a separated PCA for the populations of each species. Populations of S. pyrenaicus were divided in Northern and Southern lineages (Appendix 2) and the eigenvalues of the two first principal components, with the Burnaby-corrected matrix, explained most of the variance (Appendix 3).

The eigenvectors and consequently the variables that most contributed to the ordination of the PCA were preorbital, postorbital and premaxilla length (Appendix 3). This was a result of the different position of the eye on the head between the specimens of both lineages. In the southern lineage, the eye was displaced toward the snout resulting in a shorter preorbital lengths and a longer postorbital length. The PCA also shown a small differentiation between the populations from the Tajo drainage, as was explained previously. This was not so evident in the Southern lineage.

Regarding Squalius malacitanus, the first two PCs arranged the specimens in Atlantic and Mediterranean lineages (Appendix 4). However, the variance was spread among most of the PCs. The four first PCA were significant and eigenvalues to the first two PCs only explained 38.99 % of the variance (Appendix 5). For these reasons, we interpret that the PCA was not successful and our morphometric variables do not adequately separate the two lineages of S. malacitanus.

The number of scales in the lateral line is very similar in all Mediterranean species of the genus Squalius, although some differences were observed in the studied populations (Fig. 2).

The number of scales in the lateral line was smaller in populations of the Atlantic lineage of S. malacitanus ( = 38, 36-40, n=98) and in the Southern populations of S. pyrenaicus ( = 39, 37-41, n=125). Populations of the Northern Lineage of S. pyrenaicus ( = 41, 39-43, n=81) and of Mediterranean Lineage of S. malacitanus ( = 40, 39-43, n=58) had highest number of scales in the lateral line.

The number of vertebras was also smaller in the Atlantic Lineage of S. malacitanus ( = 37, 36-38, n=20). The values greater were to the Northern Lineage of S. pyrenaicus ( = 39, 39-41, n=20) and intermediate values were to the Mediterranean Lineage of S. malacitanus ( = 38, 37-39, n=20) and Southern Lineage of S. pyrenaicus ( = 38, 37-39, n=20) (Appendix 6).

In the case of S. malacitanus these differences cannot be explained by differences in environmental variables since both lineages inhabit adjacent rivers with identical typology.

Osteology features

Infraorbital bones were large in all populations of Squalius except in the Mediterranean population of S. malacitanus, which had narrower infraorbital bones. This was more conspicuous on 2^nd and 3^rd infraorbitals (Fig. 3). All the examined adult specimens from the Southern population of the S. pyrenaicus had exceptionally wide infraorbital bones (Fig. 3).

The skull of the Southern population of S. pyrenaicus was wider than in other populations of the genus and with a wide ethmoids bone (Appendix 7).

The maxilla of the Northern populations of S. pyrenaicus was very robust, with a posterior process stronger than in any other studied population (Appendix 8). The anterior process of the maxilla was more pointed in S. pyrenaicus populations, both Northern and Southern lineages, than in S. malacitanus, as was previously described (Doadrio & Carmona, 2006). The coronoid process was variable depending on the size of the specimen, but always was more perpendicular to the skull axis in S. pyrenaicus (northern and southern populations), whereas in S. malacitanus (Atlantic and Mediterranean populations) it was generally more inclined towards the back of the skull.

Basioccipital shape was also distinguishable between S. pyrenaicus and S. malacitanus. All populations of S. pyrenaicus had a clear triangular shape while populations of S. malacitanus had a more rounded and laterally expanded pharyngeal plate short (Appendix 9).

The pharyngeal teeth in the populations of S. pyrenaicus had strong and very conspicuous denticulations. On the contrary, in the populations of S. malacitanus the denticulations were less evident (Appendix 10). The pharyngeal teeth were thinner in the Mediterranean populations of S. malacitanus than in remaining populations, which was very evident in the two small pharyngeal teeth of the second row (Appendix 9). Considering specimens of the same size, the Mediterranean populations of S. malacitanus had the lower branch of the pharyngeal bone longer and less robust.

The cleithrum shape was very variable with the size of the individuals but usually the Southern population of S. pyrenaicus had a posterior lamina more expanding, as was previously described (Doadrio & Carmona, 2006) (Appendix 11).

Genetics

The most divergent species based on the MT-CYB gene were the Portuguese Squalius aradensis and S. torgalensis, sister to the remaining Iberian Squalius species. Previous studies on the phylogenetic relationships of the Iberian Squalius showed three highly divergent nuclear and mitochondrial clades in the species S. pyrenaicus, as is showed in the phylogenetic tree of the Fig. 4A. These three mitochondrial lineages are constituted by Northern populations (Tagus Drainage), Southern populations (Guadiana and Guadalquivir Drainage) and Sado Drainage, and they were the sister group of Squalius valentinus. Nevertheless, the nuclear monophyly of these three lineages was not recovered (Fig. 4B): indeed, Northern populations of S. pyrenaicus were clustered in a polytomy together with S. carolitertii and S. castellanus, and they were not closely related with the other two S. pyrenaicus clades (Southern and Sado). The species Squalius malacitanus also exhibited highly mitochondrial and nuclear divergent clades (Fig. 4A and 4B), which encompassed the Atlantic and the Mediterranean populations separately. The species Squalius carolitertii was closely related with Squalius castellanus. The mitochondrial and nuclear phylogenetic relationshipsinferred in this study were in concordance with previous studies of the genus (Doadrio & Carmona, 2003, 2006; Sanjur et al., 2003; Almada & Sousa-Santos, 2010; Perea et al., 2020, 2021; Mendes et al. 2021).

Uncorrected-p genetic distances based on MT-CYB between all Iberian Squalius species ranged from 1.6%, between Northern and Southern populations of S. pyrenaicus and 12.1%, between S. valentinus and S. torgalensis (Appendix 12). Uncorrected-p genetic distances of Southern populations of S. pyrenaicus relative to the remaining species were 2.7% with S. valentinus, 6, and 6.4% with S. carolitertii and S. castellanus, 7.6 and 8.6% with Mediterranean and Atlantic populations of S. malacitanus, and finally 10.6 and 11.6% with S. aradensis and S. torgalensis. In turn, uncorrected-p distances of Atlantic populations of S. malacitanus relative to the other analyzed species were 4.2% with Mediterranean populations of S. malacitanus, from 7.9 to 8.7% with the clade formed by S. valentinus and the three divergent lineages of S. pyrenaicus, and, finally, 10.9 and 11.4% with S. aradensis and S. torgalensis.

Atlantic population of S. malacitanus had two autapomorphies in the mitochondrial MT-CYB gene, none of them were transversions (Appendix 13).

Description of the Squalius populations

The high degree of morphological and genetic differentiation of Squalius malacitanus populations endemic to the Atlantic drainages and to the Miel drainage in the Mediterranean slope, and of the Squalius pyrenaicus populations from Southern Iberian drainages justifies the consideration of these population as distinct species. No available names for these populations exist, and therefore, these are described as new species in the present study.

Squalius gaditanus Doadrio & Perea sp. nov.

urn:lsid:zoobank.org:act:5FB352FD-3789-4707-B74F-8C924579ED0D

Figs. 5–6, Table 3

Holotype: MNCN_ICTIO 296955 89.3 mm SL, 104.2 mm TL; Vega River, Jara Drainage, Tarifa, Cádiz, Spain, 36.028230, -5.610120, 7 m.a.s.l., Leg. P. Garzón-Heydt, T. Nester, A. López Solano and I. Doadrio, 13.V.2022.

Paratypes: MNCN_ICTIO 296956-70, 15 specimens, Vega River, Jara Drainage, Tarifa, Cádiz, Spain, 36.028230, -5.610120, 7 m.a.s.l., Leg. P. Garzón-Heydt, T. Nester, A. López Solano and I. Doadrio, 17.V.2022. MNCN_ICTIO 296971-99 29 specimens, Vega River, Jara Drainage, Tarifa, Cádiz, Spain, 36.028230, -5.610120, 7 m.a.s.l., Leg. P. Garzón-Heydt, T. Nester, A. López Solano and I. Doadrio, 30.IV.2022. MNCN_ICTIO 297000-23, 24 specimens, Jara River, Jara Drainage, Tarifa, Cádiz, Spain, 36.103309, -5.632100, 8 m.a.s.l., Leg. P. Garzón-Heydt and I. Doadrio, 29.X.2022.

Additional material

Barbate Drainage: MNCN_ICTIO 196716-20, 5 specimens, Almodovar River, Facinas, Cádiz, Spain, 36.175707, -5.718908, 130 m.a.s.l., Leg. B. Elvira, 28.X.1986. MNCN_ICTIO 197661, 197671-74, 25348, 34272, 7 specimens, Barbate River, Casas Viejas-Benalup, Cádiz. Spain. 36.333889. -5.791512. 112 m.a.s.l.. Leg. P. Garzón-Heydt and I. Doadrio. 25.X.1978. MNCN_ICTIO 297024-34, 11 specimens, Celemín River, Casas Viejas-Benalup, Cádiz, Spain, 36.299330, -5.781360, 112 m.a.s.l., Leg. P. Garzón-Heydt and I. Doadrio. MNCN_ICTIO 297035-65, 31 specimens, Celemín River, Casas Viejas-Benalup, Cádiz, Spain, 36.304878, -5.720132, 25 m.a.s.l., Leg. J. L. González, S. Perea, P. Garzón-Heydt and I. Doadrio, 25.VI.2010. MNCN_ICTIO 196546-196547, 2 specimens, Rocinejo River, Alcalá de los Gazules, Cádiz, Spain, 36.456306, -5.660076, 165 m.a.s.l., Leg. P. Garzón-Heydt and I. Doadrio, 21.IX.1979.

Jara Drainage: MNCN_ICTIO 196203-07, 5 specimens, Jara River, Tarifa, Cádiz, Spain, 36.058995, -5.637340, 7 m.a.s.l., Leg. L. Domínguez-Nevado and B. Elvira, 15.V.1981. MNCN_ICTIO 208205 – 08, 4 specimens, Jara River, Tarifa, Cádiz, Spain, 36.057973, -5.637296, 7 m.a.s.l., Leg. P. Garzón-Heydt and I. Doadrio, 30.IV.2000. MNCN_ICTIO 243727-55, 29 specimens, Jara River, Tarifa, Cádiz, Spain, 36.076393, -5.633670, 7 m.a.s.l., Leg. I. Doadrio, 11.V.2002. MNCN_ICTIO 272136-157, 22 specimens, Jara River, Tarifa, Cádiz, Spain, 36.075378, -5.632994, 7 m.a.s.l., Leg. B. Prieto, J. L. González and I. Doadrio, 27.V.2009. MNCN_ICTIO 297066-69, 4 specimens, Jara River, Tarifa, Cádiz, Spain, 36.078611, -5.632257, 53 m.a.s.l., Leg. J. L. González, S. Perea, P. Garzón-Heydt and I. Doadrio, 25.VI.2010. MNCN_ICTIO 243780-86, 7 specimens, Vega River, Tarifa, Cádiz, Spain, 36.028230, -5.610120, 7 m.a.s.l., Leg. I. Doadrio, 1.VI.2001. MNCN_ICTIO 297070-99, 30 specimens, Vega River, Tarifa, Cádiz, Spain, 36.028230, -5.610120, 5 m.a.s.l., Leg. J. L. González, S. Perea, P. Garzón-Heydt and I. Doadrio, 25.VI.2010. MNCN_ICTIO 297100-01, 2 specimens, Vega River, Tarifa, Cádiz, Spain, 36.028230, -5.610120, 7 m.a.s.l., Leg. P. Garzón-Heydt, T. Nester, A. López-Solano and I. Doadrio, 13.V.2022. MNCN_ICTIO 297102-17, 16 specimens, Vega River, Tarifa, Cádiz, Spain, 36.103309, -5.632100, 8 m.a.s.l., Leg. P. Garzón-Heydt. and I. Doadrio, 29.X.2022.

Miel Drainage: MNCN_ICTIO 286853-286868, 16 specimens, Miel River, Algeciras, Cádiz, Spain, 36.116310, -5.485037, 20 m.a.s.l., Leg. P. Garzón-Heydt and I. Doadrio, 10.V.2002. MNCN_ICTIO 272271-272323, 52 specimens, Miel River, Algeciras, Cádiz, Spain, 36.116951, -5.483589, 20 m.a.s.l., Leg. M. Casal, S. Perea and I. Doadrio, 27.V.2009. MNCN_ICTIO 297118-21, 4 specimens, Miel River, Algeciras, Cádiz, Spain, 36.118285, -5.481130, 58 m.a.s.l., Leg. J. L. González, S. Perea, P. Garzón-Heydt and I. Doadrio, 26.VI.2010.

Diagnosis. Squalius gaditanus sp. nov. is a member of the Mediterranean clade of the Iberian species of the genus Squalius (Sanjur et al., 2003; Perea et al., 2020). Squalius gaditanus sp. nov. can be differentiated from all other known species of Squalius from the Iberian peninsula according to the following set of characters: 36-40 (x̄=38; = 38; n = 98) pored scales on the lateral line; 6-7 (x̄=6.7; = 7; n=98) scales above the lateral line; 2-3 (x̄=2.8; = 3; n=98) scales below the lateral line; 36-38 (x̄=37; = 37; n=10) number of vertebrae. Second infraorbital bone narrower than the third in adults. Maxilla with reduced pointed anterior process. Dentary short with inclined coronoid process. Posterior process of the maxilla long and thin. The lower branch of the pharyngeal bone is short and robust. Pharyngeal plate of basioccipital rounded. Squalius gaditanus sp. nov. is distinguishable from S. malacitanus, the morphological and phylogenetically most related species by lesser number of pored scales on the lateral line x̄=38 36-40 vs x̄=41, 39-43; lesser number of vertebrae 36-38 (x̄=37)vs 37-39 (x̄=38); second infraorbital bone wide vs narrow; lower branch of the pharyngeal bone short and robust vs long and thin and pharyngeal teeth robust vs thin. Genetic distances from the other species of Squalius inferred from the mitochondrial MT-CYB gene were: 4.2% with respect to S. malacitanus; 8.5% with respect to S. pyrenaicus of Northern population; 8.6% with respect to S. pyrenaicus of Southern population; 8.6% with respect to S. valentinus; 8.7% with respect to S. castellanus; about 7.9% with respect to S. carolitertii; 11.4% with respect to S. torgalensis and 10.9% with respect to S. aradensis. The new species has two autapomorphies none of them transversions in the MT-CYB gene (positions 714 and 870; Appendix 9).

Description. D III (II) 8; A III (II) 8; P I 14; V I 8; C 17; LLS 38 (36-40); SRA 6-7; SRB= 2-3; RPT 5.2; LPT 5.2; Vr = 37 (36-38). Morphometric and meristic characters of the type material are given in Table 3; measurements used in the morphometric study appear in Appendix 11. A medium-sized species that rarely reaches 130 mm of standard length. The head is short with the mouth terminal and SL/HL is 3.6-4.3 (x̄=4). The head length is similar to the height maxima of the body and BD/HL is 0.9-1.2 (x̄=1). The preorbital distance is short and HL/PrOL is 3.8-4.8 (x̄=4.3). The caudal peduncle is high and CPL/BLD is 3-3.8 (x̄=3,4). The minimum body depth is 2-2.5 (x̄=2.3) times lesser than the maximum body depth. The ventral fins are inserted approximately at the same level of the origin of the dorsal fin and PrDD/PrVD is 1-1.1 (x̄=1.1). Fins are short. Without or very small nuptial tubercles in males.

Pigmentation pattern. The body is silver with the dorsal portion dark grey, which is clearly visible in all specimens. The scales have one big black spot on the base and a series of small black spots on the distal border. The basis of pectoral fins is brown or orange. Fins rays dark grey.

Etymology. The species name gaditanus is derived from the Phoenician name of the current Cádiz province where the species is distributed.

Distribution. This new species is endemic to three small drainages of southern Spain: Jara, Barbate that drain on the Atlantic slope and Miel on the Mediterranean slope around of Gibraltar Strait. Probably S. gaditanus was widely distributed in the ancient Janda lagoon but this was dried up in the 20^th century (Perea et al., 2016) (Fig. 6).

Common name. Cachuelo gaditano.

Remarks. The species typically inhabits rivers with a Mediterranean typology conditioned by severe water stress during the summer, with specimens of S. gaditanus surviving in disconnected pools. During the autumn these rivers can have large discharges that considerably increase the flow of the river, sometimes causing disasters in human infrastructures. The drying up of the Janda Lagoon (Finlayson et al., 1997), with its 50 km², eliminated an important refuge for S. gaditanus in the face of the great discharges of autumn and the summer droughts. Reservoirs have drastically transformed the habitat of S. gaditanus in the Barbate drainage and have been a source of introduction and proliferation of invasive species. In this basin, the species is distributed almost exclusively in the headwaters of the rivers. The species should be considered Critically Endangered according to the IUCN red list criteria due to the extent of its occurrence being less than 100 km² and to the fragmentation of its populations.

Squalius tartessicus sp. nov.

urn:lsid:zoobank.org:act:E78C1F26-5FAC-4976-8148-CC89B9597E1D

Figs. 7–8, Table 4

Holotype: MNCN_ICTIO 272254 112.2 mm SL. 128 mm TL; Ciudadeja River, Guadalquivir Drainage, Las Navas de la Concepción, Sevilla, Spain, 37.917956, -5.480368, 434 m.a.s.l., Leg. P. Garzón-Heydt, J. L González, B. Prieto and E. Herrero, 05.V.2007.

Paratypes: MNCN_ICTIO 272255-272271, 17 specimens Ciudadeja River, Guadalquivir Drainage, Las Navas de la Concepción, Sevilla, Spain, 37.917956, -5.480368, 434 m.a.s.l., Leg. P. Garzón-Heydt. J. L. González, B. Prieto and E. Herrero, 05.V.2007. MNCN_ICTIO 272656-272701, 46 specimens Cala River, Guadalquivir Drainage, Santa Olalla de Cala, Huelva, Spain, 37.959673, -6.222448, 518 m.a.s.l., Leg. P. Garzón-Heydt & I. Doadrio, 19.IV.2009.

Additional material. See Appendix 14.

Diagnosis. Squalius tartessicus sp. nov. is a member of the Mediterranean clade of the genus Squalius (Sanjur et al., 2003; Perea et al., 2020). Squalius tartessicus sp. nov. can be differentiated from all other known species of Squalius from Iberian peninsula according to the following set of characters: 37-41 (x̄=38.8, = 39, n=125) pored scales on the lateral line; 6-7 (x̄=7, = 7, n=125) scales above the lateral line; 2-3 (x̄=2.9; = 3; n=125) scales below the lateral line; 37-39 (x̄=38; = 38; n=20) number of vertebrae. Infraorbital bones unusually wide in adults. Maxilla with discernable pointed anterior process. Dentary short, not inclined. Posterior process of the maxilla long and thin. The lower branch of the pharyngeal bone is short and robust. Pharyngeal plate of basioccipital triangular in shape. Posterior lamina of cleithrum expanding posteriorly. Squalius tartessicus sp. nov. is distinguishable from S. pyrenaicus, by lesser number of pored scales on the lateral line 37-41 (x̄=38.8) vs 39-43 (x̄=40); short preorbital length vs long preorbital length; mouth subterminal vs terminal mouth; in adults specimens 2^nd infraorbital bone as wide as 3^rdvs 2^nd infraorbital narrower than 3^rd; ethmoid bone wide vs narrow; in adults lamina of cleithrum expanding posteriorly vs scarcely expanding posteriorly.

Genetic distances from the other species of Squalius, inferred from the mitochondrial MT-CYB gene sequences, were: 7.6% with respect to S. malacitanus; 1.6% with respect to S. pyrenaicus of Northern population; 8.6% with respect to S. gaditanus; 2.7% with respect to S. valentinus; 6.4% with respect to S. castellanus; 6% with respect to S. carolitertii; 11.6% with respect to S. torgalensis and 10,6% with respect to S. aradensis.

Description. D III 8; A III 8; P I 114-15; V I 8; C 17; LLS 39 (37-41); SRA 6-7; SRB 2-3; RPT 5.2 LPT 5.2; Vr= 38 (37-39). Morphometric and meristic characters of the type material are presented in Table 4; measurements used in the morphometric study are listed in Appendix 12. Squalius tartessicus sp. nov. is a medium-sized species that rarely reaches 200 mm of standard length. The head is short with the mouth terminal and SL/HL is 3.8-4.7 (x̄=4.2). The head length is shorter than the height maxima of the body and BD/HL is 1-1.4 (x̄=1.2) and head length to nape; is similar to the head high and NH/NL is 0.9-1.2 (x̄=1). The preorbital distance is usually shorter than in other Squalius species but was very variable in the different populations and HL/PrOL is 3.4-5.2 (x̄=4.1). Maxilla is shorter than in S. malacitanus and the length is slightly greater than eye diameter and MxL/ED is 1-1.3 (x̄=1.1). The caudal peduncle is high and CPL/BLD is 2.9-3.7 (x̄=3,3). The minimum body depth is deeper than in other Squalius species and is 2-2.7 (x̄=2) times lesser than the maximum body depth. The ventral fins are inserted approximately at the same level of the origin of the dorsal fin and PrDD/PrVD is 1-1.1 (x̄=1.1). With small nuptial tubercles in males distributedthroughout the body. Summary of diagnostic traits of S. malacitanus, S. gaditanus, S. pyrenaicus and S. tartessicus is presented in Table 5.

Pigmentation pattern. The body is silver to brownish but without the characteristic dorsal portion dark grey of S. malacitanus and S. gaditanus. The scales have one big black spot on the base and a series of small black spots on the distal border. The basis of pectoral fins is brown or orange. With small nuptial tubercles in males.

Etymology. The species name tartessicus is derived from Tartessos, a culture that for about 400 years (8^th-5^th centuries BC) was present in the southwest of the Iberian Peninsula where S. tartessicus sp. nov. is currently distributed.

Distribution. This new species is distributed throughout the Atlantic drainages of the southern Iberian Peninsula from Almargem in Portugal to Guadalete in Spain, including the main drainages of Guadiana and Guadalquivir. In the Mediterranean slope the species is distributed in Guadalhorce Vélez, Guadalfeo and Segura drainages (Fig. 6).

Common name. Cachuelo meridional.

Remarks. The species lives in very different types of habitats from mountain rivers with a permanent flow throughout the year to Mediterranean-like rivers conditioned by severe water stress during the summer, where surviving specimens are found in disconnected pools. This species prefers deep pools within rivers and it forms a well-studied hybrid complex in many of the basins with S. alburnoides (Cunha et al., 2004). The species should be considered Vulnerable according to the IUCN red list criteria due to the decreasing of its distribution area and population number mainly by increasing of reservoirs, presence of invasive species and pollution by agriculture.