Parasites of Moroccan desert Coptodon guineensis (Pisces, Cichlidae): transition and resilience in a simplified hypersaline ecosystem

Sebkha Imlili (Atlantic Sahara) is a salt flat with over 160 permanent holes of hypersaline water generated in the Holocene and inhabited by euryhaline organisms that are considered to be relics of the past, including the cichlid fish Coptodon guineensis. We surveyed the fish parasites four times over one year, to i) identify the parasites, and ii) determine possible seasonality in infection patterns. Over 60% of the fish were infected by one to three helminths: an acanthocephalan in the intestine and two digenean metacercariae in the kidney, spleen, liver, muscle, and mesenteries. The acanthocephalan Acanthogyrus (Acanthosentis) cf. tilapiae was identified morphologically and molecularly; only one digenean (the heterophyid Pygidiopsis genata) could be identified molecularly. Both identified parasites were present throughout the sampling periods; the unidentified metacercariae were present only in summer and fall. Mean intensities, but not prevalence of infection by the acanthocephalan, reflected a biannual pattern of transmission. Infection accrued with fish size, possibly due to cannibalism. Because the water holes include only a few invertebrates, the intermediate hosts of these parasites can be inferred to be the gastropod Ecrobia ventrosa for the digeneans and either the copepod Cletocamtpus retrogressus or the ostracod Cyprideis torosa for the acanthocephalan. This ecosystem appears stable and provides a window into the past, as the acanthocephalan likely switched from freshwater tilapia to C. guineensis when the Sebkha formed. However, this is a vulnerable environment where the survival of these parasites depends on interactions maintained among only very few hosts.


Introduction
The Sebkha of Imlili is a unique Saharan wetland belonging to the Meso-Cenozoic basin of Tarfaya-Laâyoune-Dakhla located in the extreme south of Morocco, approximately 50 km south of Dakhla and about 15 km from the Atlantic Ocean [46]. Since 2018, the Sebkha has been classified as a wetland site of international importance under the Ramsar Convention as a significant waterfowl habitat. It is distinguished from other sebkhas by its physiognomy, its hydrological functioning, and its biological diversity [85]. The Sebkha Imlili is an endorheic depression, elongate (~13 km long by 2.5 km wide), and generally oriented NNE-SSW [84]. It is surrounded by a sandy dune formation, upon which fairly dense desert-type vegetation develops. In the northern part of the Sebkha, the sandy soil is brick red and covered, in the driest areas, with a fine whitish powder of crystallized salt. The Sebkha is unique because of the presence in its northern part of more than 160 holes (or pools) of permanent saline to hypersaline water, the sustainability of which is ensured by resurgences of the superficial groundwater table, itself fed by occasional flooding in the region [37,46]. These permanent pools vary in shape, diameter (1-10 m), volume (0.03-740 m 3 ), and depth (~0.1-6 m). Pool bottoms are sandy, and their edges have concretions of sand and salt. The salinity of the water in these pockets ranges from 39 to 71 [71]. The lithological nature of Sebkha Imlili and neighboring outcrops influence the chemistry of the waters in the pockets. In short, the area around the depression, as well as the bottom of the Sebkha, are drained by "chaâbas" (streambeds that only flow during intense rainy episodes generating floods) that carry evaporites and reddish saliferous silty deposits whose salt loaded contents influence the salinity of the waters in the pockets [46].
Sebkha Imlili belongs to the coastal Sahara in an area where the average temperature varies between 5°C in winter and 48°C in summer. Rainfall is typically between 50 and 60 mm per year as the result of brief, violent, and irregular storms. These low rainfall amounts, as well as fog and dew, are sufficient to maintain semi-desert type vegetation in the Sebkha [88]. The area exhibits a great richness in terms of biodiversity, with recent studies providing new data on its reptiles [69], mammals [83], birds [81,86], and flora [51]. On the other hand, the diversity of the aquatic fauna is relatively simple, with three species of crustaceans, three species of gastropod mollusks [40], and a single cichlid fish, Coptodon guineensis (Günther, 1862) [2,48,84]. Because these organisms are believed to have been trapped in the water pools when these were formed during the Holocene after the Green Sahara period, these aquatic animals are considered to be relics of the past [37].
Given that parasites in general, particularly specialists with narrow host ranges, are an integral part and drivers of biodiversity [38], the objective of our study was i) to identify the parasites of C. guineensis in the Sebkha in order to understand their origin (marine or freshwater), and ii) to obtain baseline data regarding the population dynamics of the identified parasites in order to understand the role they play in ecosystem function for future studies.  (Fig. 1). Fish from the gueltas were photographed, their TL measured, and a piece of the pectoral fin removed and stored in 96% ethanol (EtOH) for subsequent molecular identification [62]. Specimens of O. aureus are kept at the Laboratory of Biodiversity, Ecology and Genome at the University in Rabat, Morocco, three specimens of C. guineensis from the Sebkha Imlili are deposited at the Scientific Institute of Rabat under the numbers MNHN ZD1 01 17-a, MNHN ZD1 01 17-b, and MNHN ZD1 01-c, four specimens are deposited at the Royal Museum for Central Africa (RMCA 2022.020.P.0001; RMCA 2022.020.P.0002; RMCA 2022.020.P.0003; RMCA 2022.020.P.0004) and three others at the Royal Belgian Institute of Natural Sciences (RBINS 952; RBINS 953; RBINS 954). Sequences of C. guineensis were deposited in GenBank from a previous study [62] (Sebkha Imlili: MG755500; MG755474 and Oued Aabar: MK955801). Some fish were dissected fresh in the field, and others were frozen prior to dissection. Collection of the parasites was carried out by examination of the intestine and the body cavity under a dissecting microscope, and squashes of kidney, mesentery, spleen, gonads and skeletal muscles of the fish under a compound microscope. Gills were also examined under a dissecting microscope for a subsample of fish (N = 132). When fish were dissected fresh, acanthocephalans were relaxed in bottled (drinking) water for about 15 min at ambient temperature prior to fixation. Some metacercariae were excysted and heat fixed by passing a flame under the slide prior to fixation. Fixatives were 5% Neutral Buffered Formalin (NBF) for scanning electron microscopy and voucher preparation, and 96% EtOH for molecular identification or light microscopy. Vouchers of parasites were deposited at the Museum National d'Histoire Naturelle, Paris, France under the numbers MNHN HEL1881 -HEL1888.

Light and scanning electron microscopy (SEM)
Acanthocephalans fixed in 96% EtOH were rinsed in water and examined in wet mounts to determine the sex ratio and stage of maturity of females. Measurements of morphological features were taken using a microscope (Leica DM 2500) equipped with a digital camera (Leica DMC 4500) and LAS version 4.12.0 software (all from Leica Microsystems GmbH, Wetzlar, Germany). A subset of specimens fixed in NBF was stained in acetocarmine or Meyer's hematoxylin, dehydrated in a series of EtOH, and mounted in Canada balsam or Kleermount. For SEM, acanthocephalans and excysted metacercariae fixed in NBF were dehydrated in an EtOH series and chemically dried overnight using hexamethyldisilazane (HMDS), coated with gold using a sputter coater (JEOL JFC-2300HR, Japan), and observed with a SEM JEOL JSM IT 100 (Japan) under 10 kV. Measurements are means (range; number of specimens observed) given in lm unless otherwise stated. Proboscises were measured from tip to anterior limit of neck; hook blades were measured from point to top anterior surface. Vouchers of Acanthogyrus (Acanthosentis) tilapiae Baylis, 1947 deposited at the Smithsonian Institution National Museum of Natural History, USA were examined, specifically USNM 1369857, 1383211-1383227, 1394959 [10].

Molecular study
DNA from parasites fixed in 96% EtOH was isolated using a DNeasy Blood and Tissue kit (Qiagen, Germantown, MA, USA), following the manufacturer's protocol except for decreasing the elution volume to 100 lL. Based on the results of the morphological analysis for the acanthocephalan, we chose three nuclear markers for molecular identification of parasites based on sequences available in GenBank ( Table 1): portions of the large (28S) and small (18S) subunit and the internal transcribed spacer (ITS) region of the ribosomal RNA (rRNA) gene. For  Table 1. Acanthogyrus (Acanthosentis) species and sequences used in this study, including host(s), locality, GenBank accession numbers, and sequence lengths. If only one sequence length is listed for multiple accession numbers, the sequences are the same number of base pairs; otherwise, ranges are reported. Accession numbers in bold are from this study. There are mitochondrial cytochrome c oxidase I (mtCOI) sequences in GenBank for Acanthogyrus cheni (KX108947) and A. kenyirensis (MN833316), but our efforts to amplify mtCOI for A. (A.) cf. tilapiae were unsuccessful. *As of this submission, the species names were not updated in GenBank. 28S = large subunit ribosomal RNA (rRNA) gene, 18S = small subunit rRNA gene, ITS = the internal transcribed spacer region of the rRNA gene.

Species
Host ( Table 2) differed as 1 lM of each primer, 2 mM MgCl 2 , and 1 lL of template was used; cycling followed García-Varela and Nadler [39] with an annealing temperature of 56°C. The ITS region rDNA PCR used the same reagents and concentrations as in the first PCR described above, and cycling was done as in Song et al. [93]. Amplification of partial digenean 28S and the second ITS region (ITS2) of the rRNA gene was done as in Hill-Spanik et al. [47]. Products were electrophoresed on 1% agarose gels stained with GelRed (Biotium, Fremont, CA, USA) and visualized under a UV light. Samples that did not produce a band, or produced a faint band, were subjected to another round of PCR, which was done as above except the template was the product from the first PCR (instead of genomic DNA). Products were cleaned using ExoSAP-IT (Affymetrix, Santa Clara, CA, USA) and sent to Eurofins MWG Operon LLC (Louisville, KY, USA) for direct, bi-directional sequencing. All PCR and sequencing primers are listed in Table 2. For the 18S rRNA gene region, we designed two internal sequencing primers using Primer-BLAST [102] in order to generate bidirectional sequence for this marker ( Table 2).
Complementary sequences were assembled, compared to their chromatograms, and edited accordingly using Sequencher version 5.4 (Gene Codes Corp., USA). Resulting sequences were compared to those in GenBank using BLASTN (Basic Local Alignment Search Tool [6]) and deposited in GenBank. All sequences from each marker in this study were then aligned with one another to examine any differences among specimens. The longest sequence for each respective marker was then aligned with sequences from GenBank (see Table 1 for acanthocephalans; see below for digeneans) except for the digenean ITS rDNA sequences for which the BLASTN queries resulted in an identical match to an existing GenBank sequence. Clus-talW was used to generate acanthocephalan 18S rDNA and digenean 28S rDNA sequence alignments in MEGA11 [95]. NGPhylogeny.fr webservice [61] was used to implement MAFFT [54] for acanthocephalan 28S and ITS rRNA gene sequence alignments. Multiple alignments of the 28S rDNA data were generated in order to include as many nucleotides as possible in p-distance calculations given the high variation in length of GenBank sequences. NGPhylogeny.fr webservice [61] was also used to implement Gblocks [27] for selection of conserved regions of the resulting acanthocephalan ITS rRNA gene sequence alignment for use in subsequent p-distance calculations. All alignments were trimmed to remove any gaps on the terminal ends, and p-distances were calculated in MEGA11 [95].

Population dynamics and statistics
Definitions (prevalence, intensity, etc.) follow Bush et al. [26]. Sex-ratio (male: female) was determined for each season by calculating the total number of female worms over total number of male worms. Descriptive statistics and regressions were calculated using Excel. Seasonal prevalences were compared via v 2 tests. Significance of regression results were determined using Spearman's rho (r s ; two-tailed). Seasonal Table 2. Primers used for PCR amplification and sequencing; primers used only for sequencing are indicated by an asterisk. 28S = a portion of the large subunit ribosomal RNA (rRNA) gene, 18S = a portion of the small subunit rRNA gene, ITS = the internal transcribed spacer region of the rRNA gene. For primer orientation, + = sense, À = antisense.

Parasite
Marker differences in intensity, the effect of sex on intensity, and the effect of fish size on abundance were determined using Kruskal-Wallis tests (https://www.socscistatistics.com/). To assess if transmission of acanthocephalans peaked at different seasons, we analyzed female worm trunk length (L) as a proxy for worm maturity. Because of unequal sample variances in these latter data, we used Welch's ANOVA (Minitab) and Welch's t-tests (Excel) to determine significant differences. Mean intensities and mean lengths are expressed as means AE standard error. Results were considered significant at P 0.05.     complete rows closer to one another on~anterior third of body (~level of lemnisci) and more spaced out in middle of body. Proboscis cylindrical, small, armed with 6 spirals of 3 hooks (18 total). No apical organ observed. Anterior hooks markedly separated and larger than middle and posterior hooks. Hook roots simple, shorter than blades. Proboscis receptacle single-walled with ganglion at its base. Lemnisci elongate, >3Â longer than receptacle, one slightly longer than the other (considered subequal). Genital opening terminal in both sexes.

Individuals
Males    Table 3. Measurements of main morphological characters of specimens of Acanthogyrus (Acanthosentis) tilapiae described by Baylis, 1947 [19] and redescribed by Troncy, 1970 [97] and Amin, 1978 [7], A. (A.) papilio described by Troncy and Vassiliades, 1974 [97], and specimens (all stages of maturity combined) from this study collected from Coptodon guineensis at Sebkha Imlili. n = number of specimens studied. na = no data available. Merged data indicate no precision for male or female. L = length. W = width. All measurements are in lm unless indicated otherwise.  ) tilapiae collected from an unknown host in the "Atlantic Ocean" (U53000) is very short (311 bp containing the D7 region; [28]) and was only 0.64% (or 2 bp) different from our sequences, while A. bilaspurensis (the only other named species with D7 region 28S rDNA sequencing data in GenBank) differed from our sequences and A. tilapiae by 2.6% (327-bp alignment). The other D7 Table 4. Morpho-anatomical metrical data for female specimens (immature with or without ovarian balls, and gravid) and male specimens (all stages of development combined) of Acanthogyrus (Acanthosentis) cf. tilapiae from Coptodon guineensis at Sebkha Imlili. Data are averages in lm, followed by ranges in parentheses and number of specimens studied in italics. L = length; W = width; na = data not available. region 28S rDNA GenBank sequences for Acanthogyrus spp. in GenBank are unpublished (OP476684-OP476687), and pdistances ranged from 6.3% to 11.1% based on the 327-bp alignment (and 6.7-11.0% based on a 1630-bp alignment). Partial 18S rRNA gene sequences (n = 6 from C. guineensis, n = 2 from O. aureus; 1005-1703 bp) were 100% similar to one another and were again most similar to A. (A.) bilaspurensis sequences (98.3% similarity to OM262113, OM262040 with 70% BLAST query coverage). ITS rRNA gene sequences (n = 5 from C. guineensis, n = 2 from O. aureus; 746-775 bp) were 99.9% similar to one another and 91% similar to sequence from A.

Remarks
Five species of Acanthogyrus (Acanthosentis) species are known from Africa. Acanthogyrus (A.) maroccanus, is known from several barb fish species and is considered endemic in North Africa [33,68,70]. Specimens of this species have significantly larger anterior, middle, and posterior proboscis hooks (62 lm, 62 lm, 48 lm, respectively) and only 12-18 rows of spines on the body (see redescription [70]), and our 28S rRNA gene sequences differed from the A.  [99] was described from the mudskipper Periophthalmus papilio Bloch-Schneider (now P. barbus L.) based on the very small size of the specimens and anterior proboscis hooks that were markedly larger (35-40 lm) than the medial and posterior hooks (8-11 lm long). Body spines are minute and totally cover the trunk of the specimens. It is important to note that the very brief description of A. papilio is based on four specimens in poor condition according to the authors: two males (no maturity stage given) and two immature females, and that the sizes of our immature specimens as well as their proboscis hooks encompass the sizes provided in the description of A. papilio (Table 3); it is possible that the mudskipper could be an accidental host preventing full development of the worms. However, the types, although listed in the original manuscript as having been deposited at the Museum of Natural History in Paris, France, are non-existent and our efforts to obtain acanthocephalan specimens from this area were fruitless.  [19], has a broad distribution throughout continental Africa and Madagascar and is reported from over 10 species of tilapia (e.g., [7,10,66,68]) but also in the pufferfish Tetraodon fahaka (see [98] in [10]) and bagrid Labeo cylindricus [10]. Recent studies using SEM have brought up new morphological details [9,68], in particular regarding body spine distribution, size, and shape. Given our data and the very little information available on A. papilio compared to A. (A.) tilapiae, we conclude that our specimens most resemble A. (A.) tilapiae by their size and distribution and size of their proboscis hooks and body spines. Also, while the 28S rDNA sequence in Gen-Bank of A. (A.) tilapiae is only 311 bp and the origin of the specimen from which it was generated may be questionable (see Discussion), it was only 0.64% different from that of our specimens, while A. bilaspurensis differed from our sequences and A. tilapiae by 2.6%. Therefore, based on the information available at this time and because of a possible taxonomic issue with A. papilio, we identify the specimens we collected from individuals of C. guineensis and O. aureus as A. (A.) cf. tilapiae. Comparative measurements of the main morphological characters of A. tilapiae, A. papilio, and our specimens are in Table 3. We were successful at excysting metacercariae and obtaining sequences for only one of the two types, which was found both in C. guineensis at Sebkha Imlili and O. aureus at one of the gueltas. Heterophyidae Leiper, 1909. Based on two excysted specimens for SEM: body pyriform 289 (278-300) n = 2 Â 90 (89-91) n = 2 at ventral sucker level and 118 (118) n = 1 at widest (Fig. 4D). Posterior end wider than anterior end. Body covered with pectinate scales 2 (1.7-2.6) n = 11 (Fig. 4F, insert). Mouth subterminal; oral sucker unarmed 25 (24.8-26.3) n = 2 in diameter (Fig. 4E). Ventral sucker subequatorial, smaller than oral sucker, 16 (15-17) n = 2 in diameter. Given that this description is based on metacercariae, no attempt to compare with species descriptions was made and worms were identified molecularly. ITS rRNA gene sequences (n = 6, 353-369 bp long) were identical to one another and to that from a specimen of Pygidiopsis genata Looss, 1907 (AY245710) collected from cormorant Phalacrocorax carbo in Israel [36]. Our partial 28S rRNA gene sequences (n = 6, 862-877 bp) were identical to one another, and when aligned with 28S rRNA gene sequences from the only two species of Pygidiopsis in GenBank, P. summa Onji and Nishio, 1916 obtained from an experimental infection of an unknown host (AF181885) and P. macrostomum Travassos, 1928 from experimentally-infected hamster Mesocricetus auratus in Brazil (MF972527-MF972531, KT877409) and from greater bulldog bat Noctilio leporinus in Mexico (MW332629), p-distances were 9.5% (243-bp alignment) and 12.5% (816 bp-alignment), respectively.
Unidentified metacercaria (Fig. 5) Metacercariae of a presumed single but unidentified species were found only in C. guineensis at Sebkha Imlili. Specimens showed ocelli, measured~70 lm diameter in fresh squashes, and were mainly encysted in the spleen, ovaries, kidney, and the intestinal mucosa of the fish where they were associated with intense granulocytic reaction (Figs. 5A-5B).
The acanthocephalan was present throughout the four sampling periods, and there was no seasonal pattern of transmission with respect to prevalence (v 2 = 7.035, df = 3; P = 0.071) (Fig. 6). Mean intensity was significantly lower in April when ovigerous females were in highest proportion compared to the rest of the year (Kruskal-Wallis: H = 25.16; df = 3, N = 221, P < 0.01; Figs. 7 and 8). Gravid females were found throughout the year and in highest proportion in December and July (Fig. 8). Analysis of female acanthocephalans indicated a significant overall effect of season relative to female worm length (Welch's ANOVA, P < 0.001; Fig. 9). Significantly longer (more mature) female worms were found during the summer (2820.5 AE 133.7) compared to the spring (2141.3 AE 89.4; P = 0.038) and in the winter (2460.2 AE 88.7) as compared to the fall (2013.4 AE 191.8; P < 0.001; Fig. 9). Sex-ratio was~1 male: 2 females at each collection time (male to female ratios were 125:246, 31:69, 95:175, 78:137 for winter, spring, summer, and fall, respectively). There was no effect of fish sex on intensity of infection (Kruskal-Wallis: H = 3.66, df = 1, N = 152, P = 0.06 for males vs. females, and H = 3.82, df = 2; n = 210, P = 0.148 with inclusion of a group for undetermined sex). Fish size had no effect on prevalence of infection, but abundance (Spearman's Rho: r s = 0.263, P < 0.001, Fig. 10) and intensity (Spearman's Rho: r s = 1, P < 0.001) of infection increased with fish total length.
There was a significant seasonal pattern of transmission as prevalence of metacercariae of P. genata was significantly lower in April (56.2%) and highest in October (94.3%) (v 2 = 10.93, df = 3; P = 0.012) (Fig. 6). The unidentified metacercariae were encountered only in July and October when, overall, 85.7% of the 60 fish examined were infected (v 2 = 180.95, df = 3; P < 0.0001). Fish size had no effect on prevalence of infection of either metacercariae.

Discussion
A large proportion of individuals of C. guineensis at Sebkha Imlili were co-infected throughout the year by adult eoacanthocephalans, Acanthogyrus (Acanthosentis) cf. tilapiae, in their intestines and metacercariae of heterophyid P. genata in various organs. Metacercariae of another unidentified digenean, encysted particularly in the spleen, were also present in July and October. Because the life cycle of this acanthocephalan is strictly aquatic, and because the fish have been isolated in the water holes since the Holocene [37], the occurrence of the acanthocephalan in definitive host C. guineensis indicates that the parasite infected this species when the current Sebkha was a lagoon, prior to marine regression and desertification [37]. Fish C. guineensis serve as second intermediate hosts for the digeneans, which use a gastropod as first intermediate host and terrestrial animal(s) as definitive host(s). The digeneans could be relics of the past but could also be more recently introduced via terrestrial vertebrates (definitive hosts) that use the Sebkha as a source for food or water (e.g., birds or mammals [83,86]). In both cases, infections of the fish at the Sebkha occur due to the capacity of invertebrate intermediate hosts and free-living stages of the parasites to survive the extremely high salinities of the pools.
Our knowledge of digeneans that infect the fish at Sebkha Imlili is very limited. Given the significant seasonal pattern of prevalence of both metacercariae, it is possible that their definitive hosts are migratory. The presence of the unidentified metacercariae for only part of the year could also indicate the possible removal of infected fish from the population due to severe pathogenicity associated with their very high abundance in the spleen [79]. For P. genata, while adults can infect mammals and be zoonotic [65], the most common definitive hosts are piscivorous birds (e.g., pelicans and cormorants [37]), and metacercariae have been reported in several fish species including the freshwater tilapia, C. zillii [35,52]. At the Sebkha, the predominant migratory birds (e.g., accipitrids and passerines) are not piscivorous and some have been reported to not survive in high salinity environments [81]. However, the cormorant Phalacrocorax carbo marrocanus and several ardeids (common hosts for other heterophyids [31,59]) do frequent the Sebkha at various times of the year [81,86] and are good candidate definitive hosts for P. genata. For both digeneans, the presence of metacercariae in the fish indicates that their life cycles certainly involve the only snail found in the water holes, the hydrobiid Ecrobia ventrosa (Montagu, 1803). This snail is broadly distributed in brackish areas and salt marshes along the Atlantic coast of Africa [42], although it is not known if it also inhabits the Oued Mrimima where we found individuals of O. aureus infected by metacercariae of P. genata as well. However, given that a known intermediate host for P. genata is the freshwater melanopsid, Melanopsis costata (Olivier, 1804) [35,36], it is more likely that the specificity of this digenean for its gastropod first intermediate host is as broad as for its second intermediate and definitive hosts.
Regarding the acanthocephalan, Acanthogyrus (A.) tilapiae is reported from numerous cichlids of various genera in several localities throughout Africa (see checklist [56] and review [10]). Included in this broad host distribution is the redbelly tilapia C. zillii and the blue tilapia O. aureus, the latter of which we also found infected in the freshwater Oued Mrimima north of Imlili in Morocco, thus extending the geographical distribution of this parasite in Africa. While such a broad host and geographical distribution may hide a complex of species, which would result Figure 10. Abundance of infection of Coptodon guineensis at Sebkha Imlili by acanthocephalan Acanthogyrus (Acanthosentis) cf. tilapiae according to fish total length. Infection occurred in fish as small as 20 mm. Larger fish were more often infected than smaller fish.  in the specificity for their definitive hosts being narrower than apparent, it is also possible that A. (A.) tilapiae is stenoxenous and owes its success to its flexibility for its definitive hosts (and possibly its intermediate hosts as well). It is, however, reportedly a freshwater parasite, as all known hosts to date inhabit lakes and rivers (e.g., [10,19,43,44,56]). Hence, its presence in hypersaline waters at Sebkha Imlili would have to be explained by it having switched to C. guineensis as definitive host from a freshwater tilapia (e.g., native C. zillii when the Sebkha was a lagoon fed by freshwater during the Holocene (Green Sahara period) [37]. In support of such a freshwater origin are that i) C. guineensis is known to occur in some freshwater rivers [2]; ii) the freshwater tilapia O. aureus and C. zillii (often reported as host for A. (A.) tilapiae elsewhere in Africa [15]) occur in Morocco in sympatry with C. guineensis [101]; and iii) the acanthocephalan we found in C. guineensis (both from Oued Aabar and Sebkha Imlili) as well as in O. aureus is the same species based on morphology and sequencing data. Such host shifting allows for host range expansion [14,17] with multiple examples of such parasite host switching [63,78]. Although parasite host switching might be easier among closely related hosts, it is not limited to closely related host species [25,45], and it is typically the consequence of environmental changes that increase encounter opportunities (e.g., captivity or introduction of new species in an ecosystem [67]). However, successful switching of parasites with complex life cycles from freshwater to marine hosts also requires that both the parasite and the host(s) have the physiological capacity to survive altered conditions [3]; this translates as both a lack of specificity of the parasites for their intermediate host(s) and the capacity of both the free-living stages (i.e., eggs and larvae) and the intermediate hosts to withstand changes in salinity. Nothing is known about the life cycle of A. (A.) tilapiae; however, eggs (the only acanthocephalan free-living stage) and cystacanths (being provided 'environmental stability' inside their hosts) are resting stages that allow the parasite to survive otherwise potentially unfavorable environmental conditions [55]. Significantly, the potential intermediate hosts of the acanthocephalan from this study are limited to the only two microcrustaceans present in this simplified ecosystem, the ostracod Cyprideis torosa (Jones, 1850) (see [48]) and the harpacticoid copepod Cletocamptus retrogressus Schmankevitsch, 1875 (identified by Marc Pagano, IRD, Université de Marseille, pers. comm.). Both species are common in various localities in Morocco [18,21,75,87] and these microcrustaceans are capable of withstanding substantial fluctuations in salinities, including survival in extreme salinities [48,72]. Although both ostracods and copepods are the typical intermediate hosts of eoacanthocephalans [90], the two species of Acanthosentis whose life cycles are known use copepods [50,92]. Of particular interest is the marine species, A. lizae Orecchia, Paggi and Radujkovic, 1988, reported, to our knowledge, as being the only acanthocephalan that uses a marine copepod in its life cycle, which also happens to be a harpacticoid copepod [50]. Hence, C. retrograssus is a very plausible intermediate host candidate for the acanthocephalan in the Sebkha. Consequently, while it is argued that it can be more difficult for parasites with complex life cycles to transition to new environments [5,16,20], at Sebkha Imlili all the hosts of this acanthocephalan are present in the water holes because the cycle is strictly aquatic, and they are all euryhaline capable, thus allowing the potential for the otherwise assumed freshwater A. (A.) tilapiae to have established itself successfully in this environment of extreme salinities.
While we conclude that A. (A.) tilapiae was the reasonable identification of our specimens given our current state of knowledge, the alternate identification as A. papilio would generate a very different scenario, which must be discussed herein because of its biological relevance. Coptodon guineensis is a coastal cichlid on the African west coast that ranges from Angola to the north of Morocco [4,80,82,94]. Because A. papilio was described from mudskippers from mangroves in Senegal [99] where C. guineensis also lives [57], and because this genus of acanthocephalan is common in tilapia species in Africa, it is not unreasonable to also consider that C. guineensis could have been infected when the Sebkha was a lagoon [1,37]. It is also plausible for such a marine parasite to have persisted over a long period of time in an environment where its intermediate hosts would already be present, as all organisms associated with the Sebkha are coastal or estuarine and known from other areas in Morocco [48]. To add to the argument, there is no host detail for the sole 28S rDNA sequence of A. (A.) tilapiae available in GenBank [28] other than the notation that it came "from the Atlantic Ocean and [was] provided by Dr. Mattei" (p. 579-580), who was a researcher in Dakar, Senegal (https://en.wikipedia.org/wiki/Xavier_Mattei#Career). Hence, it is likely that this specimen, identified as A. (A.) tilapiae [13], was from the Dakar areathe area where A. papilio was described (Joal-Fadiouth, Senegal [99]) and where C. guineensis also occurs [57]. In short, it is likely that the short GenBank sequence could as well have been from specimens of A. papilio. The 99.9-100% similarities among sequences (using all three markers) that we generated from acanthocephalan specimens taken from C. guineensis and O. aureus would imply that A. papilio is found in both marine and freshwater systems. This conundrum can be reconciled if we question the validity of A. papilio, as its description is based only on immature specimens and overlaps somewhat with that of A. (A.) tilapiae. We should also consider that the acanthocephalan species could be an altogether different and cryptic species because it has been isolated for a long period of time, and given that the Coptodon from the Sebkha Imlili is possibly an "incipient" species of tilapia [2]. Given what we know at this stage however, it is most reasonable to identify this acanthocephalan as A. (A.) cf. tilapiae. Future studies will need to investigate the validity of A. papilio as well as the genetics of A. (A.) tilapiae to disentangle the conundrum and identify a putative complex of species.
Unraveling the population dynamics of this acanthocephalan can help us understand how the Sebkha ecosystem functions and how it has persisted for such a long period of time. Both the abundance and intensity of infection by the acanthocephalan increased with total fish length. This is not unusual and is typically explained for trophically transmitted parasites by an increase or a shift in the diet of their hosts, sometimes amplified with the occurrence of paratenic hosts [29,76] and/or an extended longevity of the adult worms [55]. This latter explanation, however, is unlikely for fish acanthocephalans that rarely live longer than a few months [55], which seems the case for the acanthocephalan we found. The simplest explanation is that larger individuals of C. guineensis graze and ingest more organisms at the bottom of the water holes (including copepods and ostracods [62]) than smaller fish; Significantly, larger fish also cannibalize smaller ones (pers. obs.). Snails and shrimp are the only organisms that could, in theory, fulfill the role of paratenic hosts for this parasite in this particular habitat, as they too graze at the bottom of the water holes. Snails and shrimp do not appear to be part of the fish diet, however, and no cystacanth was found in the~40 shrimp we examined, although it must be noted that these examinations were sporadic. Furthermore, acanthocephalans typically use vertebrates as paratenic hosts [76], with the rare exception of some Neoechinorhynchus species that use invertebrates [58,60]. Hence, the occurrence of paratenic hosts in this acanthocephalan life cycle is not expected but cannot be totally discounted for lack of thorough investigation. In contrast, there is evidence of cannibalism by fish larger than 60 mm, which indicates the possibility for eupostcyclic transmission of this acanthocephalan [22,76], a phenomenon that, at least in part, could explain the higher intensities we found in larger fish [23,77].
The distribution of adult acanthocephalans was aggregated, and the mean intensity of infection was~5 worms per fish, which is quite low for hosts whose environmental conditions can be compared to that of captive animals. The non-occurrence of super infection (maximum intensity = 21 worms in only one fish out of the 322 examined) indicates the likelihood of a negative feedback mechanism acting as a regulatory factor to limit infrapopulations, which can explain that this host/parasite system has persisted for a very long time. Given the constant occurrence of immature and gravid females and the likelihood of the continuous presence of infective cystacanths in the pools' microcrustaceans throughout the year (development of cystacanths in those is~15-30 days [90]), along with low mean acanthocephalan intensities beg the question regarding which factor(s) operate in this system to regulate infection; however, at this stage of knowledge of this fish/parasite system, we can only speculate in this regard. For instance, several experimental studies of fish acanthocephalans demonstrated density-dependent survival of cystacanths associated with limited resources, such as space [100], which could be the case herein as the worms seem to occupy a short post-gastric portion of the intestine and thus may have a constricted "zone of viability" [55]. A somewhat unique character of the acanthocephalan population in fish at the Sebkha is that gravid female worms, which occurred throughout the year, were in significantly high proportion twice a year (December and July), which could indicate the occurrence of two marked reproductive periods per year. Overall, the acanthocephalan population in C. guineensis appears to display a dynamic equilibrium between recruitment and loss of parasites, possibly because of a continuous turn-over of infrapopulations in the fish coupled with the constant availability of infective cystacanths. The lower intensity observed in April may be an instance where we captured a shift in generations, as we also saw more immature females in that same collection period. Typically, factors linked to acanthocephalan cycles (e.g., length of prepatent and patent periods) are associated with host movements and environmental conditions (including seasonality). None of these factors apply to the acanthocephalan in the Sebkha, however, as there seems to be no particular period of unfavorable conditions in the water holes: environmental conditions are fairly stable throughout the year and water temperatures do not fluctuate as in higher latitudes, so no host appears to "overwinter" per se, and none of the hosts migrate. In this regard, the water holes display similar conditions to captivity. While there appears to be a cyclic pattern in female worm maturation, there is no obvious pattern of seasonal transmission as prevalence of infection was similar throughout the year, the male/female sex ratio was constant, and gravid and immature and mature females were present at all periods of collection, albeit in different proportions. This type of pattern is not unique (e.g., [8,74]) but is atypical for acanthocephalans, which most often display a yearly cycle closely associated with seasonal periodicity or biotic factors such as a qualitative or quantitative change in host diet [55,76]. Hence, this particular host/acanthocephalan system may project the dynamic modifications that can occur in some other parasite cycles, which should be taken into consideration with respect to global climate change [24]. Lastly, while mathematical models tend to oversimplify complex systems, the ecosystem at Sebkha Imlili is relatively very simple in terms of biodiversity and, as such, constitutes a natural laboratory allowing the testing, at least to some extent, of such models. The acanthocephalan population in C. guineensis in the Sebkha appears to be stable and fits one of the model patterns suggested by Dobson and Keymer [32]: long term consistency where hosts are present throughout the year and where there is a relatively constant intermediate host density.
In conclusion, fish at the Sebkha Imlili were found to act as definitive hosts for one acanthocephalan and second intermediate hosts for two digeneans. The acanthocephalan population appears to be stable, which indicates the occurrence of regulatory factors given that they live under conditions comparable to hosts held in captivity under otherwise ideal conditions for super infection to occur. Nevertheless, there also appears to be at least two non-exclusive periods of peak maturation for the worms as measured by both total female length and by the proportion of gravid worms found (winter and summer), indicating that there may be some times of the year that are more favorable to transmission, maturation and development than others. While our data do not allow us to state if these relative spikes are the only two periods of high worm maturation because we did not sample continuously throughout the year, they do indicate that there are clearly times of the year when there are relatively more female worms that are smaller and immature (herein spring and fall) and other times in which more females reach maturity (herein winter and summer). This suggests at least some level of cycling in the recruitment and/or development of the worms. Drivers of such a maturation cycle, whether intrinsic or extrinsic, are as yet unknown and further year-round sampling would be required to determine these factors.
The acanthocephalan we found was identified as A. (A.) cf. tilapiae, which appears to have infected C. guineensis via host switching from a freshwater tilapia during the Holocene and transitioned to salt water. However, we also entertain the idea that this parasite could be A. papilio, which, assuming the latter is a valid species, would have been carried over, with all its hosts (intermediate and definitive) during the Holocene and whose life cycle persisted thanks to the euryhaline character of all involved hosts and their capacity to withstand extreme salinities. Although this is the most parsimonious explanation of the persistence of this acanthocephalan at the Sebkha, the current status of information regarding this species of acanthocephalan is, in our opinion, not reliable and a thorough genetic study of the African Acanthogyrus (Acanthosentis) species, in particular A. (A.) tilapiae, is called for to determine its origin and to determine if this is a species complex encompassing a much broader distribution than reported thus far (i.e., encompassing marine hosts). Given the simplified ecosystem of the Sebkha, the intermediate host of this acanthocephalan must be one of the two microcrustaceans (harpacticoid copepod or ostracod) that inhabit the water holes and are known prey items of the definitive host. The harpacticoid is the most plausible candidate given that other cycles known for Acanthosentis species also incorporate a copepod, including a marine harpacticoid. While such a two-host cycle is rigid, and consequently vulnerable, several factors appear to have allowed this parasite to escape these constraints, including the likelihood of post-cyclic transmission and the fact that both putative intermediate hosts are euryhaline and able to live at very high salinities. Thus, this acanthocephalan illustrates a case of ecological fitting and resilience in an extreme environment [14,49]. Regarding the digenean P. genata, we infer that the first intermediate host must be the hydrobiid snail E. ventrosa, because it is the sole gastropod found in the water holes, and that the definitive hosts are likely not omnipresent at the Sebkha given the seasonal pattern of infection. Thus, this digenean, in contrast to the acanthocephalan, could be a recent parasite acquisition for the fish in the Sebkha. It is, nevertheless, as for A. (A.) tilapiae, reportedly a freshwater parasite that owes its survival in the water holes to its low specificity for intermediate hosts with the capacity of living in hypersaline waters. While the persistence of all three parasites in the Sebkha shows the resilience of life in extreme conditions and can project the fate of some parasites in times of climate change, they are nevertheless highly vulnerable as their life cycles would be interrupted if their invertebrate hosts were to disappear.