Analysis of variance (ANOVA) for sexual dimorphism showed significant differences between sexes (p<0.05) for variables associated with preorbital length, anal fin and body depth (Table 2). Orbital size was also different in males in relation to females but probably as a consequence of the large difference in eye size in the Ifni lake population (in comparison to the other populations). When we removed the Ifni Lake population from the analysis no significant differences were found between males and females. In general, all variables showing significant dimorphism were greater in females than in males (Table 3). Due to the presence of sexual dimorphism and because we collected few female samples (n=38), we removed females for subsequent morphological analyses.

On the other hand, most of the morphometric variables showed significant differences between populations in the two-way ANOVA analysis (Table 2). An analysis of body proportions based on non-parametric Kruskal-Wallis and Mann-Whitney post hoc comparisons were used, to find differences in body shape between populations that can remain masked if only linear untransformed measurements are considered.

According to these analyses, morphological differences in body shape were found in almost all morphometric measurements (Appendix 1). The population from Isli Lake showed the largest body size compared to the remaining populations. Otherwise, the maxilla in the Isli population was long but not significantly larger than in the other populations (Table 4). In Isli Lake trout, the size of the eye was smallest of all populations studied, but since mature individuals from this population exhibited the largest body size of all studied populations, the smaller size eye found in this population could reflect allometric growth (Table 4). The Isli Lake individuals showed the smallest pectoral fin in comparison to other populations. The population of Miaami Lagoon in the upstream Chbouka River also had a long preorbital distance, which is in agreement with a well-developed maxilla in length and height. Trout of Miaami Lagoon were shorter, which is revealed in body depth and caudal peduncle measurements. The Ifni Lake population, a trout population having small-sized mature individuals, presented the largest eye size in adult specimens (Table 4). The maxilla was very long and narrow and the caudal peduncle was shorter and deeper in the Ifni population than in other populations (Table 4). Dades, Moulouya, Ziz and Tensift populations were characterized by having shorter preorbital length and Dades and Ziz populations by shorter maxillary length. The height of maxilla was largest in the Dades population. Moulouya and Tensift populations showed long caudal peduncle in relation to the body height considered from the origin of the anal fin. The upper mandibular length was largest in the Moulouya and Tensift populations with respect to head length. The Tensift population had a smaller dorsal fin relative to the remaining populations.

The PCA divided the Salmo populations into three groups that correspond to those of Ifni Lake, those of Miaami Lagoon and a third large group containing the remaining Salmo populations (Fig. 2). The eigenvalues for the two first Principal Components, with the Burnaby-corrected matrix for removing size effect, mostly explained the variance (Table 5). The variables that contributed most to the ordination in the PCA were orbital and preorbital lengths, caudal peduncle lengths, size and height of the maxilla and length of the dorsal fin (Table 5).

With regard to meristic characters, we found no significant differences according to Kruskal-Wallis and Mann-Whitney non parametric analyses for the following traits: number of vertebrae, number of scales in lateral line, number of scales in transversal on adipose fin, number of rays in dorsal and anal fins, number of procurrent rays on the caudal fin and number of branchyostegal rays. Only two traits showed significant differences, dorsal and anal pterygophores (Table 6). The number of pterygophores in dorsal fin had its highest value in Salmo akairos and its lowest value in the Isli Lake population (Fig. 3). Likewise, the number of pterygophores in the anal fin presented its highest value in Salmo pallaryi from Sidi Ali Lake comparing with the remaining populations studied (Fig. 3). The species Salmo akairos (Ifni population) also showed significant differences in relation to the other populations in anal fin pterygophores.

The concatenated matrix including both mitochondrial genes rendered a total of 2088bp (MT-CYTB: 1140bp + CR: 947bp including gaps). Bayesian tree reconstruction revealed well-supported phylogenetic relationships among trout populations from Morocco and demonstrated the early divergence of the Dades population (Draa basin) from the others (Fig. 4). In fact, the genetic distances between Dades and the remaining populations, including the recognized species Salmo akairos (Ifni), were the highest recorded for both genes and for all comparisons (Table 7). The Dades populations also showed nine autapomorphies in the MT-CYTB, all of them transitions (Table 9). Genetic distances among the rest of trout populations from Morocco ranged from 0.2-0.6% for MT-CYTB and 0.2-0.5% for CR (Table 7). Autapomorphies in the MT-CYTB gene, although not so numerous as in the Dades population, were found in the Ifni, Isli, Lakhdar and Tensift populations (Table 8).

The Isli population showed a pigmentation pattern without spots and parr marks, as also occurred in Salmo pallaryi. The Dades population showed numerous irregular parr marks, usually greater to or equal than 14, which usually are retained in adult specimens. In the Dades population, the body is covered by small and numerous red and black spots that sometimes extend to the caudal fin and opercular area. Adult specimens from the Moulouya, Ziz and Oum er Rbia basins usually retained large and scarce parr marks. Adults from the Tensift population did not retain parr marks and showed a dark coloration (Fig. 5).

Vomer teeth in Moroccan trout were arranged in zigzag with the exception of the Isli population, whose teeth were placed in one row within a narrow vomer. The vomer length was lesser than half of the parasphenoid length in all the studied populations except for the Ifni population. In this last population vomer length was greater than half of the parasphenoid length (Appendix 2).

For Dades individuals, the lateral processes of the ethmoid bone were extended, conferring a T shape to the bone, and the upper process of the premaxilla bone was small and laterally inclined. In comparison to other populations studied, where the upper maxillary process was greater than 47%, we found that the height of the upper maxillary process in the Ifni population was less than 45%, with respect to the length of the basal plate. Most individuals of the Isli population had the premaxilla strongly ossified and extended over the maxilla (Appendix 2).

The upper jaw in the Dades, Ziz and Oum er Rbia basins was shorter than in the other populations. In the Ifni and Isli populations upper jaw length usually exceeded the posterior border of the eye. The Ifni population had the lowest maxilla, while the Miaami and Moulouya populations had the deepest. The skull was narrower in the Isli population and the snout was more pointed in Isli and Ifni lakes, in comparison to the other populations (Appendix 2).

The anterior maxillary process in the Isli population was thinner than in the remaining Moroccan populations and the inflection of this process, in relation to the main axis of the maxilla, formed an angle greater than 135º, while in other trout populations from Morocco it was less than 135º (Appendix 2).

Our morphological and molecular analyses revealed that the population from Ifni Lake, recently described as a different species (Salmo akairos), had a high degree of distinct morphological features and moderate genetic differentiation in mitochondrial genes. The populations from the Oum er Rbia Basin showed scarce morphological and genetic differentiation relative to the Tensift Basin, where the species Salmo pellegrini was described (Delling & Doadrio, 2005; Geiger et al., 2014); therefore we provisionally assigned populations from Oum er Rbia basins to Salmo pellegrini. However, one of the populations of this latter basin (Miaami Lagoon) showed differential morphological traits, particularly in the length and height of the maxilla and premaxilla bones, but it was nested within the Oum er Rbia Basin mitochondrial clade and therefore this population should be also provisionally assigned to S. pellegrini. The populations of the Ziz and Moulouya basins showed moderate genetic differentiation for the MT-CYTB gene; genetic differences in allozyme markers had also been previously reported for the Ziz population (Lbadaoui et al., 2011). Nevertheless, no morphological differentiation was detected when compared to other populations from the Oum er Rbia Basin, and so in the absence of the inclusion of other Mediterranean populations in the morphological and genetic analyses, we named these two populations as Salmo sp. The populations from Isli Lake and the Dades Basin showed a high degree of morphological and genetic differentiation that justifies their consideration as distinct species. No available name for these populations exists at present, and therefore the two new species are described in this study.

Salmo multipunctata Doadrio, Perea & Yahyaoui, sp. nov.

urn:lsid:zoobank.org:act:413738BC-A868-413A-9371-76EBCDB0A132

Holotype: Figure 6, Table 9. MNCN 290753, 126.3 mm (SL). Female. Dades River, Draa Basin, Aït Aatou O'Moussa (31.871776, -5.738205), Morocco; 21/10/2014. Collected by Doadrio, I; Yahyaoui, A; Garzón, P; Perea, S.

DIAGNOSIS: Differs from other known species of the genus Salmo in the numerous irregular narrow parr marks, usually ≥14, in the lateral line. Upper row of parr marks ≥10. Lower row of parr marks variable, usually ≥5. Two conspicuous black spots placed at the right and left sides of the iris and two less visible spots at the upper and lower sides of the iris. Numerous black, small spots placed at the upper side of the body and several red ocellated and non ocellated spots placed in middle of the body. Black spots sometimes extended to the opercular region and the caudal fin. Ethmoid bone with a characteristic T shape. Premaxilla bone with small and inclined upper process. From 14 to 16 (Median=14) scales from base of adipose fin to lateral line. From 12 to 14 dorsal fin pterygophores (Median=13). Nine molecular autopomorphies in MT-CYTB gene in the nucleotide positions 60, 216,474, 504, 732, 849, 915, 990, 1090.

DESCRIPTION: LL 104-112 (x=108.5 Median=108). RSA 14-16 (x=14.3; Median=14). AP 9-11 (x=10.3, Median=10). AR iv 7-8 (x=7.8, Median=8). DP 12-14 (x=12.9, Median=13). DR v (iv) 9-10 (x=9.3, Median=9). Bq 9-10 (x=9.5, Median=9.5). C 19. Cpr 10-13 (x=11.5, Median=11). Ve 53-55 (x=54.1, Median=54). It is a small-sized species that rarely reaches 250 mm standard length. The body is slightly more elongated in comparison to other trout populations, with maximum body depth, at the level of the anal origin, ranging from 16.1-18.7% SL in females and 15.5-18.5% SL in males. The head is large in relation to the body, in comparison to all trout populations except those of the Ifni and Miaami populations, with head length 22.6-24.9% SL in females and 23.7-25.8% SL in males. The anterior part of the head shows a flat snout. The preorbital distance is the shortest of the populations studied and the proportion with respect to the head length is 18.6-20.7% HL in females and 18.1-21.3% HL in males. The fourth and fifth infraorbitals are narrow and sparsely extend over the hyomandibular bone (Fig. 7). The eye size in Salmo multipunctata and Salmo akairos is the largest in relation to other populations studied, and in Salmo multipunctata the horizontal diameter of the eye is 29.4-31.8% HL in females and 28.5-33.5% HL in males.

The supramaxilla is large and slender; its length is more than half of the maxilla. The maxilla is short and the proportion, with respect to the head length, is 46.3-53.2% SL in females and 48.9-52.3% SL in males. The upper jaw is deep as in the populations of the Ziz and Oum er Rbia river basins, except for the Miaami population that has the significant deepest upper jaw. The maximum height of the upper jaw, in proportion to its length, is 33-38.4% UJL in females and 31.8-38.7% UJL in males. The caudal peduncle (CPL) is slightly deeper than in other populations, except that of Ifni Lake (S. akairos), and the proportion of the body at the least depth, with respect to caudal peduncle length, is 23.6-26.9% CPL in females and 24-28% CPL in males. The teeth in vomer, premaxilla and glossohyal are very robust, as those of the Ziz and Moulouya populations.

The coloration is slightly yellow, lighter in the ventral area, and brownish in the dorsal region. The fins are slightly yellow with orange marginal borders of caudal and adipose fins. Dorsal, pelvic and anal fins in adults have white anterior edges. The dorsal fin presents a black line below the white mark. The dorsal fin has small black and red spots. Body with numerous small spots, which are black and slightly ocellated in the dorsal region and red spots (not ocellated) in the middle of the body. The iris has two black, conspicuous lateral spots and two less conspicuous above and below the iris. Parr marks are narrow, irregular and numerous, placed in the middle of body, sometimes divided into two marks. The species has a characteristic opercular black spot, and another spot just above it (usually ocellated). There are usually numerous black spots in the opercular area and the caudal fin. There are eight or nine black spots on the dorsal body, very conspicuous in juveniles, which tend to disappear in adults.

ETYMOLOGY: The species name “multipunctata” refers to its diagnostic pigmentation pattern, which is constituted by numerous spots and marks along the body.

DISTRIBUTION: This new species is endemic to the Draa Basin in southern Morocco, inhabiting the headwaters of the Dades and M'Goum rivers. In the Draa basin the species usually co-occurs with the cyprinid fish Luciobarbus lepineyi (Pellegrin, 1939).

COMMON NAME: We propose the English common name “Draa trout” for this species.

HABITAT AND BIOLOGY: Similar to other riverine trout of Morocco, this species inhabits small and shallow streams with clear and cold waters. No information about spawning season is available.

CONSERVATION: The habitat of this species suffers catastrophic seasonal floods that influence dramatically the demography of the population. The continuous deforestation of the Atlas Mountains and the disappearance of the riparian vegetation, increase the adverse effect of the overflows over the small Atlas rivers. Overfishing is also a problem for a small population such as the one from the Dades River. For this reason, we suggest that this species should be included in the IUCN category of Endangered (EN).

GENETICS: Uncorrected genetic distances between Salmo multipunctata and other studied populations of trout for the MT-CYTB gene ranged from 0.9 to 1.4%. Salmo multipunctata is phylogenetically basal to S. trutta from Europe and other Salmo populations from Morocco assigned to S. akairos, S. pellegrini and S. macrostigma.

Salmo viridis Doadrio, Perea & Yahyaoui, sp. nov.

urn:lsid:zoobank.org:act:A6209574-45BE-4A3E-BCB6-DC21AFDC502C

Holotype: Figure 8, Table 10. MNCN 280413, 254.1 mm (SL), Male. Isli Lake, Imilchil (32.217526, -5.548031), Morocco; 30/10/2009. Collected by Doadrio, I; Yahyaoui, A; Garzón, P; Cuerva, I.

DIAGNOSIS: Differs from other known species of the genus Salmo from Morocco (except Salmo pallaryi) by the absence of spots or marks on the body. The species Salmo viridis has a lesser number of anal pterygophores (AP 10-11; Median=10) than Salmo pallaryi (AP 12-14; Median=13) and shorter orbital length. Horizontal orbital length with respect to the head length is 16-23% HL in Salmo viridis vs 25-31% HL in Salmo pallaryi. The following combination of characters are also diagnostic to Salmo viridis: Vomer teeth in one row, premaxilla extended laterally, supramaxilla short and deep, dorsal fin pterygophores from 11 to 13 (Median=12). One molecular autapomorphy in the MT-CYTB gene in the nucleotide position 508.

DESCRIPTION: LL 105-116 (x=109.4; Median=108). RSA 15-18 (x=16.9; Median=17). AP 10-11 (x=10.3, Median=10). AR iv 7-8 (x=7.6, Median=8). DP 11-13 (x=11.7, Median=12). DR v(vi) 9-10 (x=9.7, Median=10). Bq 8-11 (x=9.2; Median=9). C 19. Ve 53-55 (x=54.4, Median=54). It is a large-sized species that reaches 500 mm of standard length. The caudal peduncle is more elongated, in comparison to other trout populations, and the proportion of body to least depth, with respect to caudal peduncle length, is 25.3-26% CPL in females and 23.5-27.1% CPL in males. In comparison to all trout populations the head is small, relative to the body, with head length 21.7-24.4% SL in females and 21.4-24.8% SL in males. The skull is narrow with a pointed snout. The preorbital distance is the largest, compared to the populations studied, and the proportion, with respect to head length, is 26-31% HL in females and 21.6-34% HL in males. The fourth and fifth infraorbitals are wide and extended over the hyomandibular bone (Fig. 9).

The supramaxilla is short and deep. The length of the maxilla is variable in this species and usually reaches the edge of the posterior eye orbit (Fig. 10). The length of the maxilla, in proportion and with respect to head length is 44.4-56.3% SL in females and 46.5-52.8% SL in males. The upper jaw is narrower than in other populations except that of Ifni Lake, which belongs to S. akairos. The maximum height of the upper jaw, in proportion to its length, is 28.4-33.1% UJL in females and 24.6-32.2% UJL in males. The eye orbit is small, in proportion to head length, the horizontal diameter being 17.5-19.4% HL in females and 15.6-22.8% HL in males and the vertical diameter being 16.4-19.8% HL in females and 13.8-20% HL in males (Fig. 10). The vomer teeth are placed in one row and are small and thin.

The coloration is uniformly grayish with some greenish iridizations, paler in the ventral region. The pectoral fins are slightly yellow and caudal, adipose and anal fins are greenish. No spot or marks are found, only some traces in caudal fin and dorsal skull.

ETYMOLOGY: The species name “viridis” refers to its diagnostic greenish coloration in some parts of body and fins.

DISTRIBUTION: This new species is endemic to Isli Lake in the Atlas Mountains of Morocco. It is the only fish species inhabiting this lake.

COMMON NAME: We propose the English common name “green trout” for this species.

HABITAT AND BIOLOGY: This species inhabits the Isli Lake (Fig. 11). Isli is situated on a high plateau, 2270 meters above sea level. It is a large and deep (≈2 km of diameter, 92 m deep) oligotrophic lake. The geological origin of this Lake is under debate. Some authors point to its origin as a consequence of a meteorite impact that occurred approximately 40 000 years ago (Ibhi et al., 2013; Nachit et al., 2013), while others consider a tectonic origin during the Pleistocene period, similar to the one proposed for other Moroccan lakes (Chaabout et al., 2013). The presence of a natural population of Salmo viridis in Isli Lake is in agreement with the latter hypothesis. The spawning of this species takes place at the end of October and November in the sandy beaches of the lake.

CONSERVATION: The habitat of this species is threatened due to eutrophication, related to intensive overpasture. Overfishing is also a problem due to the low protection of the lake. No juveniles have been observed in the last years and several malformations in fins and skull were observed in this population. For this reason, we suggest that this species should be included in the IUCN category of Endangered (EN).

GENETICS: Uncorrected genetic distances between Salmo viridis and other studied populations of trout for the MT-CYTB gene ranged from 0.2 to 1%. The divergence degree is similar to that of Salmo akairos in comparison to other populations inhabiting Atlas lakes.