Genetic relationship in goatfishes ( Mullidae : Perciformes ) of the Red Sea and the Mediterranean , with remarks on Suez Canal migrants *

The goatfish family (Mullidae) can be easily distinguished from all other percoids by their unique hyoid barbels, used in trophic foraging. This circumtropical family includes about 50 species belonging to six genera. Most goatfish inhabit inshore areas and are commercially important throughout their distribution. Due to their economic value, goatfish have been the subject of rather intense biological and taxonomic studies (e.g., Lee, 1974; Munro, 1976; Papaconstantinou et al., 1981; Sorden, 1983; Gosline, 1984; Wahbeh, 1992a,b; Golani, 1994). The taxonomy of Mullidae in the Red Sea has received considerable attention (Dor and Ben-Tuvia, 1984; Al-Absey, 1988; Ben-Tuvia and Kissil, 1988). Thirteen species have been reported from the Red Sea region, belonging to three genera: Mulloides Bleeker, 1849, Parupeneus Bleeker, 1863 and Upeneus Cuvier, 1829. Electrophoretic analysis can add a different dimension, which provides insight into the interrelationship of species and genera, to these taxonomic studies (Avise, 1974; Buth, 1984; Bolch et al., 1994). Regarding the Red Sea Mullidae, electrophoretic studies may be of particular interest, since two members of this family, the brownband goatfish Upeneus pori Ben-Tuvia and Golani, 1989 (previously known in this area as U. asymmetricus), and the goldband goatfish, Upeneus SCI. MAR., 63 (2): 129-135 SCIENTIA MARINA 1999


INTRODUCTION
The goatfish family (Mullidae) can be easily distinguished from all other percoids by their unique hyoid barbels, used in trophic foraging. This circumtropical family includes about 50 species belonging to six genera.
The taxonomy of Mullidae in the Red Sea has received considerable attention (Dor and Ben-Tuvia, 1984;Al-Absey, 1988;Ben-Tuvia and Kissil, 1988). Thirteen species have been reported from the Red Sea region, belonging to three genera: Mulloides Bleeker, 1849, Parupeneus Bleeker, 1863 and Upeneus Cuvier, 1829. Electrophoretic analysis can add a different dimension, which provides insight into the interrelationship of species and genera, to these taxonomic studies (Avise, 1974;Buth, 1984;Bolch et al., 1994). Regarding the Red Sea Mullidae, electrophoretic studies may be of particular interest, since two members of this family, the brownband goatfish Upeneus pori Ben-Tuvia and Golani, 1989 (previously known in this area as U. asymmetricus), and the goldband goatfish, Upeneus SCI. MAR.,63 (2): [129][130][131][132][133][134][135] moluccensis (Bleeker, 1855), have migrated through the Suez Canal, which was opened in 1869, into the Mediterranean Sea. Colonization exposed these populations to ecological pressures that may have differed from those exerted on their original populations. Therefore, it has been hypothesized that the colonizing population would undergo some changes (Parsons, 1984). Despite the different hydrological conditions and faunistic affinities of these two seas, each of these two species succeeded to establish a large permanent population in the Mediterranean. In their new region they encountered two indigenous mullids, the red mullet Mullus barbatus Linneaus, 1758, and the striped mullet, Mullus surmuletus Linneaus, 1758.
Another question presented by Lessepsian migration is the association between genetic variability and success in colonization. High genetic variability has been hypothesized as a preadaptation to a heterogeneous environment (Hedrick et al., 1976). Since successful colonization often requires the occupation of new and different habitats, it is expected that successful colonizer species have high genetic variability (Baker and Stebbins, 1965;Safriel and Ritte, 1980;Gray, 1986;Williamson and Fitter, 1996).
The purpose of this study is thus threefold: to determine the genetic relationship between Red Sea mullid species and their Mediterranean confamilials; to compare the genetic structure of Red Sea and Mediterranean populations of the colonizing species; and to compare the genetic structure of colonizing and non-colonizing species.

Collection of material
Of the 13 known species of Red Sea Mullidae, 11 reach the Gulf of Elat (Ben-Tuvia and Kissil, 1988 Golani, 1989 were considered endemic to the Red Sea until the latter extended its distribution into the Mediterranean. Of the three species that were not sampled, two, namely Mulloides vanicolensis Valenciennes, 1831 and Parupeneus heptacanthus (Lacepède, 1801), are rare in the northern part of the Gulf of Elat and never appeared in our catches. The third species, Parupeneus cyclostomus (Lacepède, 1801), is associated with coral reefs in which collection is prohibited by law.  In the eastern Mediterranean, populations of the two migrants, Upeneus moluccensis and U. pori, were sampled, as well as the two indigenous goatfish Mullus barbatus and M. surmuletus which extend to the northeastern Atlantic.
Fish collection was conducted in the northern tip of the Red Sea (Gulf of Elat) and the eastern Mediterranean. Details of the collection are presented in Table 1. Fishes collected by beach seine and trawl were placed on dry ice immediately upon capture. Those captured by trammel nets and spear guns were placed on dry ice no later than 45 minutes after capture. Fish were delivered frozen to the Department of Genetics at the Hebrew University and stored at -40 o C.

Electrophoresis
Muscle and liver tissue were dissected from each individual and homogenized with equal volumes of 0.1 M Tris-HC1 pH 7.0: H 2 O: Glycerol (2:2:1) using a Sonifier Cell Disruptor, model W185 (Heat System-Ultrasonic, Inc.) (10,000 rpm). The homogenates were centrifuged at 4 o C for 30 min at 14,000 g. The supernatants were run on horizontal starch gels using 12% hydrolyzed starch (Sigma No. S-4501) at 4 o C.
The enzymes analyzed are listed in Table 2. Staining procedures followed Selander et al. (1971) and Harris and Hopkinson (1976). Enzyme nomenclature followed the International Union of Biochemistry recommendations. Allelic designations were determined according to their electrophoretic mobility, following Buth (1983).

Genetic analysis
The genetic distances between all possible pairs were calculated using Nei's (1978) index. The DIS-PAN computer program (Ota, 1993), using the neighbour-joining method (Saitou and Nei, 1987), was based on genetic distances and applied in the construction of an unrooted phylogenetic tree. This program was chosen because it allows compensation for possible bias due to a small number of individuals in sample size; a few species in this study were represented by less than 20 individuals. Allele frequencies of the different enzymes were used for the calculation of the levels of polymorphism (P  -) and heterozygosity (H -) for each species.
The genetic distances (Nei, 1978) between all species and populations are given in Table 4. The neighbour-joining phylogenetic tree constructed by the DISPAN program is given in Fig 1. The mean levels of polymorphism and heterozygosity are listed in Table 5.
For each of the two migrating species, Upeneus pori and U. moluccensis, the level of genetic differ-entiation between the Mediterranean and the Red Sea populations was estimated using both the twotailed Fisher's exact test for independence and Wright's F ST statistic. Since each polymorphic locus has only two alleles, and for each species we compare only two populations, we preferred to use the exact test over the more commonly used χ 2 test for independence. The significance of departure of F ST 132 D. GOLANI and U. RITTE  values from zero was calculated using the χ 2 test of Workman and Niswander (1970). The effective number of migrants per population per generation (N e m) can be obtained from the F ST statistic by the formula: 1-F ST N e m = --------- (Walpes, 1987).

4F ST
The results of the tests of the level of genetic differentiation in Upeneus pori and U. moluccensis are given in Table 6. For U. pori, the mean F ST over all six polymorphic loci was 0.1171. The effective number of migrants between populations is estimated as 1.89 individuals per generation. For U. moluccensis, the mean F ST over the two polymorphic loci was 0.0720. The effective number of migrants between populations is estimated as 3.22 individuals per generation.  (Nei, 1978) among Red Sea and Mediterranean goatfishes (Abbreviations as in Table 3 Table 3). Average heterozygosity for the Red Sea populations of the migrating species, for both U. pori and U. moluccensis, was 0.04283 (n=2), while the average heterozygosity for the non-migrating Red Sea species (e.g., Upeneus subvittatus, Mulloides flavolineatus, Parupeneus forsskali, P. rubescens and P. macronema) was 0.03212 (n=5). This difference is statistically not significant (t=0.8022 (5 d.f.); p>0.400).

DISCUSSION
The electrophoretic analysis of the Red Sea and Mediterranean goatfish generally supports the conclusions of the morphomeristic-based systematic studies of this family (Al-Absey, 1988;Ben-Tuvia and Kissil, 1988).
The only notable exception is U. pori, which clusters to its other congeneric species at a level which has been found in this study to distinguish between genera. This finding supports Kühlmorgen-Hille (1974), who suggested to split Upeneus into two subgenera; Upeneus (Upeneus) for species with five to seven interdorsal fin scales and 12 rows on the upper part of the peduncle (including U. moluccensis and U. subvittatus), and Upeneus (Penon) for species with four interdorsal fin scales and ten scale rows on the peduncle (including U. pori).
The interrelationship between genera found in this study agrees with that of Shaklee et al. (1982), who found that Upeneus clusters first with Mulloidichthys (= Mulloides) and later with Parupeneus.
The morphological similarity between Mullus barbatus and M. surmuletus has led some authors (Day, 1880-84;Lozano Ray, 1952) to treat them as different subspecies. Despite the results of our electrophoretic analysis which also found them to be very close, we refrain from drawing a categorical conclusion, especially in light of Basaglia and Callegarini (1988) and Mamuris et al. (1998), who also used electrophoresis and clearly distinguished these two fish as belonging to two separate species.
It has been speculated that the Red Sea fishes, found today in the Mediterranean, inhabited it prior to the opening of the Suez Canal. Unfortunately, before that event, little was known on the ichthyofauna of the eastern Mediterranean. However, due to their Atlantic origin, almost all indigenous fishes from the eastern Mediterranean are also known from the western Mediterranean. The occurrence of Red Sea fishes with Indo-Pacific distribution exclusively in the eastern Mediterranean supports the interpretation of their recent arrival by migration through the Suez Canal. Early comparative studies of Lessepsian goatfishes in the Red Sea and the Mediterranean revealed some meristic changes, which were attributed to the adjustment of the spawning season to a more fluctuating temperature regime in the Mediterranean (Golani, 1990). The present electrophoretic study revealed a low level of genetic distance between the Red Sea and Mediterranean populations of the two colonizing species, U. pori (D = 0.081) and U. moluccensis (D = 0.004). These values are lower than those calculated in Mullidae (Stepien et al., 1994) in particular, and among conspecific populations in general (Avise, 1974;Buth, 1984). Thus, it can be concluded that the short time elapsed since the establishment of the Mediterranean populations has not been sufficient to establish discernible genetic differences. Only a single allele (PGM-1-C) detected in the Mediterranean population of U. pori was not found in the Red Sea (Table 3). Because of the short time since the separation, it is reasonable to assume that this allele may also occur, perhaps rarely, in the Red Sea. The low level of genetic distance and the calculated effective number of migrants between the colonizing and the source populations (1.89 for U. pori and 3.22 for U. moluccensis) indicate that the establishment of the Mediterranean populations probably involved relatively large propagules and/or continuous gene flow from the Red Sea.
Our results concerning the question, whether high genetic variability contributes to success in colonization, are negative. This conclusion was reached by many other authors (Ritte and Pashtan, 1982;Ehrlich, 1986;Holdgate, 1986;Gray, 1986;Lavee and Ritte, 1994;Williamson and Fitter, 1996). The distinction between closely-related colonizing and non-colonizing species seems rather to be associated with ecological differences (Ayal and Safriel, 1983;Golani, 1993). crew of F/V Ophir for help in the collection of fish from the Mediterranean and Ms. B. Appelbaum for editorial assistance. This study was partly supported by the Lerner-Gray Fund for Marine Research, American Museum of Natural History, New York.