The occurrence, distribution and biology of invasive fish species in fresh and brackish water bodies of NE Morocco

The occurrence, distribution and biology of invasive fish species in fresh and brackish water bodies of NE Morocco. Monitoring the presence and expansion of alien species and upgrading their biological and ecological knowledge seems crucial to mitigate their possible impact on native communities. Within inland superficial waters, alien fish represent an important threat to the biodiversity and studies on their impact on native communities have increased around the world in the last years. However, little is known about their occurrence, biology and influences in North Africa in general, and more specifically in Morocco. In the present work we aimed to: 1) investigate the presence of any native Aphanius species, especially the Mediterranean killifish Aphanius fasciatus recorded from the lower basin of the Moulouya River (NE Morocco); 2) monitor the presence and expansion of two invasive species, the eastern mosquitofish Gambusia holbrooki and the mummichog Fundulus heteroclitus; and 3) contribute to the understanding of the ecological and abiotic affinities that govern the distribution of these alien fishes in North Africa. To achieve these goals, several field sampling campaigns were carried out between 2014 and 2018 across eastern Morocco, comprising the administrative Oriental Region and the Moulouya River Basin and covering an area of 119,268 km2. No native Aphanius species were found. The eastern mosquitofish has invaded the freshwater hydrosystems of the northern part of Morocco, including the study area, while the mummichog is currently limited to the brackish and salty wetlands of Lower Moulouya. Our results show that the known ranges of these two alien species have expanded. Data published through GBIF (Doi:10.15470/2qed9o)


Introduction
Biological invasions are one of the most important human impacts on a wide range of ecosystems and a major cause of global change (Ricciardi, 2006). The introduction and invasion of alien species are one of the main threats to biodiversity and ecosystem functioning worldwide, and represent the cause of huge ecological and economic costs around the world (Fausch and García-Berthou, 2013). Many invasive species have been implicated in species extinction, habitat degradation and ecosystem alteration (Cox, 1999;Pimentel et al., 2005;Coccia, 2015). Freshwater invasions have been less studied than terrestrial invasions, and the ecological and evolutionary consequences of most invasions remain unknown (Ricciardi and MacIsaac, 2011). Freshwaters are the most invaded and threatened ecosystem worldwide, with proportionally more invaders than terrestrial systems (Vitousek et al., 1997;Strayer, 2010).
Many members of the genus Aphanius are "victims" of biological invasion, such as the Mediterranean killifish, A. fasciatus (Valenciennes, 1821), a cyprinodotid fish whose native range includes coastal brackish-waters mainly in the coastal zone of the central and eastern Mediterranean (Bianco, 1995). Nevertheless, it can also be found in various inland water bodies, such as lakes, inland streams and even in some North African oases (Hrbek and Meyer, 2003;Güçlü et al., 2013). Aphanius fasciatus is also present in various Mediterranean islands, such as Sardinia, Corsica and Cyprus (Bianco et al., 1996;Kottelat and Freyhof, 2007;Englezou et al., 2018). The Mediterranean killifish was reported in Morocco for the first time at the wetland of the Moulouya River mouth by Melhaoui (1994), but the species has not been reported since this record.
Aphanius fasciatus has disappeared from many sites where it once existed, and its populations have declined dramatically. In many cases it has even reached local extinction due to problems such as pollution of continental and coastal waters, brackish-water habitat degradation, destruction and reduction of salt-works and, most importantly, introduction of exotic fishes (Bianco, 1995). Competition with the alien species Gambusia holbrooki Girard, 1859 has strongly reduced the presence of A. fasciatus in many parts of its range (Kessabi et al., 2009;Valdesalici et al., 2015).
Another alien fish which has been reported to cause range reductions in native toothcarps is the mummichog Fundulus heteroclitus (Linnaeus, 1766), a small teleost fish, naturally occurring in saltmarshes of the Atlantic coast of North America from the Gulf of St Lawrence, southwards to north-eastern Florida (Shute, 1980). It was found in Iberia for the first time between 1970 and 1973 (Bernardi et al., 1995). Since then, F. heteroclitus has been found in many parts of Spain and Portugal (Morim et al., 2019). We found it for the first time in Morocco and Northern Africa at the Moulouya River mouth This study aimed to evaluate the presence of A. fasciatus or any native Aphanius species in NE Morocco, in parallel with the two invasive species G. holbrooki and F. heteroclitus. Furthermore, our goal was to study the interaction between the invasive fish, and their occurrence, distribution and biology in the fresh and brackish water bodies of the study area.

Study area
Morocco is currently divided into 12 administrative regions, including the Oriental Region ( fig. 1), which occupies almost all the eastern side of the country and covers an area of 88,681 km 2 (see Mabrouki et al., 2018 for details). The Oriental Region includes the wilaya of Oujda (Oujda-Angad prefecture) and the provinces of Berkane, Driouch, Figuig (Bensaad et al., 2017;Mabrouki et al., 2017).

Surveys
Field investigations (often in the framework of various hydrobiological studies) have been carried out since 2014 at 45 stations throughout the Moulouya River basin, including its main affluents: Oued Anzegmir (side of the High Atlas), Oued Melloulou (Middle Atlas slope) and Oued Za (High Plateau), and at about 60 stations spread throughout eastern Morocco from the northern regions of Nador and Saïdia, to Figuig in the southeast and Talessint and Bouanane in the southwest ( fig. 1). The stations that showed the occurrence of the mentioned fish species were re-examined between 2017 and 2018. Quantitative sampling of fish fauna was carried out using nets, searching in the most suitable places for the studied species. Sampling lasted an average of one hour over an area of 10 m 2 at each station, sufficient time  to trap and catch virtually all the fish in each area. Invasive species caught were preserved in formalin solution, while native fish (especially fry) were returned to the water. Permission to perform these studied was granted by the authorities.

Abiotic factors
For this study, ten environmental parameters were selected: sulfate (SO 4 2-), biological oxygen demand after 5 days (BOD 5 ), phosphate (PO 4 3-) and nitrate (N-NO 3 ) were measured in the laboratory. Conductivity, salinity, pH, dissolved oxygen, temperature and the mean depth were measured in situ using a multiparametric measuring device (WTW, Multi-Line P4). Two replicas of water samples from each station were taken in 500 ml polyethylene bottles. The water samples were preserved with 2 ml of concentrated hydrochloric acid (pH = 2). According to standards ISO 5667-6 (1990), ISO 5667-2 (1991) and ISO 5667-3 (1994), water samples were transported in a cooler at a low temperature (± 4 ºC) to stop the metabolic activities of organisms in the water.
The current velocity, well known for its selective action on habitat and species distribution (Mabrouki et al., 2019a), was quantified by its mean value at three different locations of the same station. In the absence of a hydrometric reel to measure current velocity it was estimated using a stopwatch at various points of the watercourse by measuring the time it takes for a floating object to cross a given path. The average speed (converted to cm -1 ) was semi-quantitatively estimated, followed by a transformation into three modalities: 1, very low to no current < 5 cm -1 ; 2, low current 5 < 2 < 25 cm -1 ; and 3, average current 25 < 3 < 50 cm -1 .

Statistical processing
Statistical analyses were carried out using software R package version 3.3.1. (R Core Team, 2019). In the modelling of counting processes, here the abundance of a species, two kinds of models are commonly implemented, namely the Poisson model and the negative binomial model (Hilbe, 2011). Frequently, counting data is characterized by overdispersion, whereby the sample variance is greater than its average. In the case where the anomalous dispersion is proved, the Poisson regression is no longer suitable for modelling this distribution, and the negative binomial regression model, allowing more flexibility in the dispersion, should be used. The relevance of Poisson regression or negative binomial models was evaluated by the Pearson Residue Test (Plackett, 1983).

Results
Occurrence, distribution and habitat types of the studied species Despite our repeated sampling and the large number of prospected stations, we were unable to find A. fasciatus (or any Aphanius species). In contrast, its probable range was totally occupied by two exotic fish species F. heteroclitus and G. holbrooki. The species F. heteroclitus was found in six of 45 sampling sites while G. holbrooki appeared in 35.
In Morocco, the distribution of F. heteroclitus seems to be limited to the Oriental region so far, from the wetlands of Ain Chabbak to the mouth of the Moulouya River ( fig. 2). Its habitat consists of coastal swamps and large sections of the Moulouya to its confluence with the Mediterranean, at an altitudinal ranging between 0 and 307 m above sea level. In shallow habitats, where predators are less abundant, such as in the Ain Chebbak wetland, F. heteroclitus can be found swimming freely in pelagic and superficial parts along with G. holbrooki. On the other hand, in the main watercourse of the lower Moulouya, where the depth is greater and predatory fishes are more abundant, F. heteroclitus was only caught among the aquatic vegetation, always together with G. holbrooki, except near the river mouth in extreme lower Moulouya (stations S37, S78 and S39), where it was captured alone.
With regard to G. holbrooki, the species appears to be widely distributed through Morocco, including the Moulouya River Basin ( fig. 2). It is highly confined to brackish waters near the coast (as in Nador lagoon and the Moulouya SIBE), and it is omnipresent along Za watercourses, Sebra and the complex Zeghzel-Cherraa rivers. It can go up to Melloulou and the Middle Moulouya during the summer period with the weakening current. The species also appears to be well established in other areas of the oriental region of Morocco, from Nador and Al Hoceima north and to Figuig south-east and the Talessint region south-west. The eastern mosquitofish seems able to cope remarkably well in a wide range of habitats, even those that are highly degraded as long as there is water. Its habitats range from large permanent streams of average and low altitudes, to dams, natural and managed springs, and small temporary tributaries (provided that they are connected to a permanent source or underground water). It can be also found in human constructions such as irrigation canals and cement tanks for water retention.
The physicochemical variables of water Both G. holbrooki and F. heteroclitus species are well known for their tolerance to large variations in abiotic parameters (Kneib, 1986;Lipcius and Subrahmanyam, 1986;Pyke, 2008). However, in the study area, G. holbrooki appears to excel F. heteroclitus in term of oscillation intervals of most of the studied parameters (see appendix). G. holbrooki showed great plasticity in terms of salinity; was recorded in low mineralized waters with a minimum of 0.587 g/l (587.67 ppm), and also in the brackish waters of the coastal system with a maximum of 1,837 g/l (1,839.09 ppm). The greater plasticity of the species was demonstrated mainly by

Fundulus heteroclitus Gambusia holbrooki
Aphanius fasciatus (according to Melhaoui, 1994) the large intervals of the organic pollution indicators, showing that it can tolerate BOD 5 concentrations between 1.49 and 20.3 mg/l, nitrate between 1.25 and 67 mg/l, sulfate between 33 and 403 mg/l phosphate between 0.016 and 2.82 mg/l and finally pH between 6.65 and 8.3. This same observation was noted for dissolved oxygen; the species seems to withstand great fluctuations of this vital element and it can be found not only in very well oxygenated waters 11.3 mg/l but also in weakly oxygenated waters with a minimum of 2.01 mg/l. Fundulus heteroclitus is instead more demanding in terms of the physicochemical quality of the water and showed weak ranges of variation for most of the studied parameters (see appendix). However, it can survive in a hypersaline waters compared to G. holbrooki, with a maximum of 2,757 g/l (2,760.15 ppm) recorded at the Moulouya mouth (S39).
Concerning temperature, it is difficult to accurately evaluate variation for these two species in the study area because we did not perform a monthly follow-up in all the stations because of the violent floods recorded during the cold period. However, water temperature is linked to local conditions (climate, duration of sunshine, flow and altitude) and to seasonality in the Moulouya River Basin Taybi, 2016). During the study period, the maximum temperature did not exceed 26 ºC for either species. Given the lack of sampling during winter, our study does not report the lowest temperature at which the fish remain alive or active.

Modelling results
Poisson regression and negative binomial regression showed the same explanatory variables of the abundance frequency, with similar effects. However, the Poisson regression model showed an abnormal over-dispersion, and its statistical relevance was rejected by the Pearson Residue Test. The negative binomial regression model was better adapted to these data, and its statistical relevance was verified.

Discussion
Growth in international trade and concurrent increases in transport capacity have accelerated the rate of introduction of alien species worldwide, with freshwater ecosystems, and their native fish communities, being particularly susceptible (Sala et al., 2000;Kolar and Lodge, 2002;Macdonald and Tonkin, 2008). The most important pathways of freshwater invasions are aquaculture, pet/aquarium trade and stocking activities, shipping, and inland canals, without forgetting that some species can be introduced intentionally (Coccia, 2015;Nunes et al., 2015). Invasive species are a major threat to global biodiversity, causing the decline and extinction of native freshwater fish species throughout the world (Strauss et al., 2006;García-Berthou et al., 2005;Srean, 2015). The piscivorous 'Nile perch' Lates niloticus (Linnaeus, 1758) is a classical example, as its introduction in the 1950s in Lake Victoria in East Africa led to the extinction of over 200 endemic fishes (Kitchell et al., 1997).
During our survey, we did not find A. fasciatus in Morocco in spite of our exhaustive sampling. This could be due to several reasons, such as pollution and destruction of the habitat on one hand, and the impact of alien fish on the other, as has occurred in many areas of its range (IUCN, 2006). We should take into account that the record of A. fasciatus from Morocco could be erroneous due to misidentification and the fact that scientific collections from Morocco contain no samples. Although healthy populations still exist in some parts of the Maghreb, occupying the same type of habitats in Algeria and Tunisia (Kraiem, 1983;Boumendjel et al., 2015), the palaeogeographic history of Morocco differs. Until 5.3 million years ago (Krijgsman et al., 1999a(Krijgsman et al., , 1999b, together with Spain, the north of Morocco formed the Betico-Rifian massif. As Spain and Morocco shared fauna not present in the other countries, in Morocco populations it would be more likely to find species related to Spanish aphaniids, such as A. baeticus Doadrio, Carmona and Fernández-Delgado, 2002 or A. iberus than A. fasciatus. On the other hand, the present distribution of A. iberus, A. Taybi et al. baeticus, and even A. saourensis Blanco, Hrbek and Doadrio, 2006 occurs along the coastline that existed in the Miocene before the separation of the Betico-Rifian Massif at the Miocene-Pliocene boundary 5.3 million years ago (Doadrio, 1994, Krijgsman et al., 1999a, 1999b. It is therefore more likely to find Aphanius in inland Morocco than in coastal rivers.
In consensus with these observations, the previous record of A. fasciatus from Morocco would be the misidentification of a different species of the genus. Nevertheless, the idea that alien fishes have driven the Moroccan population of a native Aphanius species to extinction is also possible, especially in eastern Morocco, where G. holbrooki and F. heteroclitus have invaded all suitable hydrosystems, from the fresh continental waters of rivers and springs, to brackish and salty waters of coastal lagoons and wetlands.
The nature of G. holbrooki with its high reproductive potential, fast maturation rate, flexible behavior and broad environmental tolerances have contributed to its success as an invader, and the species is considered to pose a serious threat not only to native fishes worldwide but also to amphibians, macro-invertebrates, zooplankton and phytoplankton communities (Courtney and Meffe 1989;Howe et al., 1997;Caiola and de Sostoa, 2005;Macdonald and Tonkin, 2008;Rowe et al., 2008;Mabrouki et al., 2019b). Usually, the species of the genus Aphanius can take refuge in estuarine habitats to escape competition with G. holbrookii due to its strong salt-tolerant character like all cyprinodontids, as has previously been reported (Nordlie and Mirandi, 1996;Doadrio, 2002;Alcaraz et al., 2008;Valdesalici et al., 2015). However, in our case these refuges are occupied by another alien fish, which is also a cyprinodontiform specialized in mesohalyne and hyperhalyne waters and which is possibly responsible for the population regression of A. baeticus in the Iberian Peninsula (Oliva-Paterna et al., 2006).
Regarding the ecological affinities of the two invasive species, the results of the negative binomial regression of mosquitofish abundance as a function of the environmental variables selected for this study (table 1) showed the existence of a highly significant static relationship between Gambusia abundance and salinity, and water velocity (p-value < 0.001), and a very significant relation with depth (p-value between [0.001, 0.01]). This correlation is negative for the three variables, i.e., if the salinity of the environment decreases by 1.4, the abundance of G. holbrooki will increase by one. G. holbrooki has extensive tolerance for several environmental factors as reported in other works (Pyke, 2008), but according to our results it seems to prefer shallow waters with average to low salinity and slow current velocity, in agreement with findings in previous studies (Stearns and Sage, 1980;Brown-Peterson and Peterson, 1990). Its peak of colonization will thus be carried out, most probably, in shallow lentic water bodies with low or moderate salinity. Brown-Peterson and Peterson (1990) reported that females from freshwater were in better somatic condition and matured later, since ovarian development was significantly more advanced in oligohalyne waters. The same model was used for F. heteroclitus; a positive correlation was detected between this species and salinity (table 2), and the relevance of this regression model was verified by the Pearson residue test; the model is well adapted to the data.
To conclude, both invasive species generally occur in water that is shallow, stagnant or slow moving, with dense aquatic and riparian vegetation. In water bodies that are moderately deep, they are generally found along the shallow edges, especially those that are well vegetated. However, F. heteroclitus seems to be specialized in brackish saline waters of coastal hydrosystems, while G. holbrooki seems specialized in freshwater and slightly brackish water, showing scarce selectivity in terms of habitat, as was observed in its native area (Moyle and Nichols, 1973). Indeed, this species can occur in environments that range from almost completely undisturbed swamps, lakes, and streams to highly disturbed water bodies including urban drains with various water quality problems such as pollution and dense growth of aquatic and surface weeds. G. holbrooki is often extremely abundant in disturbed habitats near urban areas (Arthington and Lloyd, 1989;Pyke, 2008). This could be explained by its high tolerance to fluctuations in abiotic factors (Odum and Caldwell, 1955;Keup and Bayliss, 1964;Cherry et al., 1976;Fastelli et al., 2012), this trait being one of the main causes behind its expansion success and outcompeting F. heteroclitus in the  Research on invasive species is an urgent research topic for the conservation of biodiversity (Dukes and Mooney, 2004;Rice and Sax, 2005;Taybi et al., 2020a). Studies on the impact of invasive fish on native fauna have flourished around the world in recent years. However, little is known about their occurrence, biology and influence in North Africa in general or in Morocco in particular. Since G. holbrooki was first introduced in Morocco in 1929 to fight against waterborne diseases, it has invaded all the northern Moroccan hydrosystems (Azeroual, 2003;Clavero et al., 2014). Our surveys in eastern Morocco have significantly increased its known range. Future assessments in other parts of the country could expand the species' distribution area, especially in the Western Rif and the High and the Anti-Atlas valley, and even in the Sahara. On the other hand, F. heteroclitus seems to be limited to the wetland of Moulouya mouth for the time being, but thanks to its 'salt-tolerant' character, this newcomer could soon invade habitats of the same type. Indeed, considering the habitat destruction and deterioration of the quality of surface waters currently occurring in Morocco Taybi et al., 2016;Bensaad et al., 2017;Mabrouki et al., 2017;Taybi et al., 2020b), in addition to the climate change effects and intensification of saline stress (Moss et al., 2009), both alien fishes would be in a better position to invade North African ecosystems. It is also of note that freshwater ichtyofauna of Morocco is still little known and new species are being recorded (Ford et al., 2020). Several native fish species could co-occur with the two invaders in eastern Morocco and in the Moulouya River basin, such as the North African shad Alosa algeriensis Regan, 1916 and many endemic barbels (i.e. Luciobarbus guercifensis Doadrio, Casal-López and Yahyaoui, 2016;L. lanigarensis Brahimi, Libois, Henrard andFreyhof, 2018 andL. yahyaouii Doadrio, Casal-Lopez andPerea, 2016). Knowledge of the distribution of both native and non-native freshwater fishes in Morocco therefore merits urgent research. Taybi et al. Appendix. Abundance and value of abiotic parameters measured at the sampling sites where G. holbrroki and F. heteroclitus were found: Gam, abundance of Gambusia; Fun, abundance of Fundulus; Depth, in cm; Temp, water temperature in ºC; Sali, salinity in g/l; Dis_O, dissolved oxygen in mg/l; NO 3 , nitrate in mg/l; SO, sulfate in mg/l; PO, phosphate in mg/l; BOD 5 in mg/l; Vel, current velocity in cm/s.