Distribution of Contracaecum (Nematoda: Anisakidae) Larvae in Freshwater Fish from the Northern Regions of South Africa

A total of 1 847 fishes (16 species) from 14 reservoirs in northern and north-eastern regions of South Africa were collected and examined for larval Contracaecum spp. between 2005 and 2013. This study, the first to examine several potential second intermediate hosts, found Clarias gariepinus, Coptodon rendalli, Cyprinus carpio, Hydrocynus vittatus, Labeobarbus marequensis, Marcusenius macrolepidotus, Micropterus salmoides, Oreochromis mossambicus and Schilbe intermedius infected with the third-stage larvae. Coptodon rendalli, Marcusenius macrolepidotus and Micropterus salmoides are new host records for South Africa. A generalised linear model identified locality as the main factor affecting parasite burden.


Fish examination
The standard length (SL; mm), total length (TL; mm), mass (g), sex and locality of each specimen were recorded. Fish were identified according to Skelton (2001). Fishes were sacrificed by severing their spinal cords, dissected, and their body cavities and gastrointestinal tracts examined. Nematodes were removed and placed into petri dishes containing a physiological saline solution for cleaning and their numbers recorded. The majority of parasites were fixed by placement in 70% ethanol heated to 70 °C and thereafter, when uncoiled, preserved in 70% ethanol. For morphological and identification purposes some specimens were temporarily mounted on glass slides in lactophenol. All specimens were identified to genus level based on morphological characters using keys by Bykhovskaya-Pavlovskaya (1964) and Moravec (1998). Identification was possible based on the presence of three anterior lips and the structure of the anterior alimentary tract comprising the visible ventriculus, ventricular appendix and an intestinal caecum. Infection parameters, i.e. prevalence (P), mean intensity (MI) and mean abundance (MA), were calculated in accordance with Bush et al. (1997).

Statistical analysis
Data used for this study were a composite of several unrelated surveys done from 2005 to 2013. Data pertaining to the five most infected fish, i.e. C. gariepinus, Coptodon rendalli (Boulenger, 1896), Micropterus salmoides (Lacepede, 1802), O. mossambicus and S. intermedius (see Supplementary Table S1), were subjected to analysis. Analysis of variables common to all sampling episodes such as season, fish gender, sampling locality and fish length were used to determine which variables affected parasite burden. Rows missing any data for variables tested were omitted from the analysis process. To evaluate which variable had an effect on parasite burden a generalised linear model was performed. All analyses were conducted in R 3.1.1 (R Development Core Team 2014).   Interpretation of p-values was in accordance with Wild and Seber (2000).

Results
A total of 1 847 specimens from 16 fish species, nine families and five orders were examined. Of these, 531 ( (Table 2). In addition, parasite burden in C. gariepinus, M. salmoides and S. intermedius was also affected at some level by all the other variables tested, while body length (p  0.001) was the only other variable to affect parasite burden in C. rendalli.

Discussion
Contracaecum spp. are widely distributed globally with their larvae recorded in various fish species from different continents (Aloo 1999;Wharton et al. 1999;Farahnak et al. 2002;Lymbery et al. 2002;Martins et al. 2005;Barson et al. 2008; Al-Zubaidy 2009; see Supplementary  Table S3 for South African records). In this study Contracaecum spp. were recorded in nine of the 16 fish species and seven of nine families examined. In the life cycle of Contracaecum spp., the first intermediate host is a copepod and the second a fish. After an infected copepod is ingested by a fish, the nematode migrates along the alimentary canal and makes its way into the host's body cavity by burrowing through the intestinal wall where it attains a specific length while encapsulated within a transparent sheath. Further development occurs only if the fish host is consumed by a final host and maturation and reproduction occurs in the host's gut (Moravec 1998). Hence the probability of a fish being infected with Contracaecum spp. will largely depend on the feeding behaviour and presence of all hosts necessary to complete the life cycle of this parasite (Boomker 1994a(Boomker , 1994bBergmann and Motta 2004).
In this study a high prevalence of Contracaecum larvae in C. gariepinus and S. intermedius was inferred to be due to both species being opportunistic and omnivorous feeders (Skelton 2001). Conversely, infection levels in C. carpio, M. salmoides, O. mossambicus and C. rendalli were far lower. For example, a single parasite was found in the 28 C. carpio examined. Findings in this study for C. carpio and O. mossambicus are in agreement with those reported by Boomker (1994b) and Barson et al. (2008), and the likelihood of these and M. salmoides, and C. rendalli ingesting the first intermediate host may be coincidental. For example, O. mossambicus may ingest infected copepods incidentally as they both feed on phytoplankton, whereas C. rendalli may consume copepods that shelter among their macrophyte food. Of interest is that none of the 116 S. zambezensis specimens examined were infected, even though this species shares a similar feeding preference to those of C. carpio, C. gariepinus and S. intermedius (Skelton 2001).
Although female C. gariepinus were larger than males, significantly higher levels of infection occurred in males. Differences can be attributed to males having a MA of 71 as opposed to 37 for females, despite both genders sharing a parasite prevalence of 78%. Although female specimens were larger, growth in male C. gariepinus exceeds that of female fish after three years (Skelton 2001). Higher growth rates in fish equate to a higher food demand and hence higher ingestion rates, which in turn can increase the likelihood of ingesting the first intermediate host.
By contrast, a prevalence of 38% and a MA of 2.5 were reported for female M. salmoides as opposed to 19% and 0.7, respectively, for males. Here the higher food requirement of females increase the prospect of ingesting infected hosts. In addition, males are restricted to the proximity of breeding nests in order to safe-guard eggs and newly hatched larvae during periods when copepod numbers are expected to swell, thereby greatly reducing the likelihood of consuming infected copepods. Similarly, S. intermedius females had a higher prevalence of 80% and a MA of 39 compared to 60% and 20, respectively, for males. Given that S. intermedius are opportunistic predators that feed from the mid-depth and surface waters (Skelton 2001), a higher food demand by females will increase the possibility of ingesting infected free-swimming copepods.
Despite Mashego and Saayman (1981) and Barson (2004) not finding an effect of host gender on burden of Contracaecum spp., studies on M. salmoides by Aloo (1999) and on Liza abu (Heckel, 1843) by Al-Zubaidy (2009) indicated a higher prevalence in females. Findings in this study with regard to M. salmoides concur with those of Aloo (1999). However, higher infection levels in females is thought by Thomas (2002) to be attributed to testosterone immunosuppression with a fluctuation of hormone levels occurring during the breeding season (Desjardins et al. 2005). Hence variations of parasite burden in male and female fish during pre-and post-spawning should be investigated.
Hydrocynus vittatus  Data analysis revealed that differences in parasite infection levels (P, MI and MA) between localities were significant for the five most infected species. This is similar to previous studies (Supplementary Table S3) where infection levels for a given species varied between localities. This variation may be due to several factors such as the presence and density of the intermediate and definitive hosts (Hockey et al. 2005 Statistical analyses revealed seasonal variations of parasite burden to be significant in C. gariepinus, M. salmoides and S. intermedius, but not in the cichlids examined. Seasonal variation in parasite burden can be attributed to a number of factors such as fish activity, feeding behaviour and feeding rate, all of which depend on water temperature. As temperatures increase, so do the metabolic rates of fish and the demand for food which, in Species  Locality a  AP  FBD  HR  KPT  LD  MO  ND  NL  PB  P1  P2  P3  TS  TZ  Micropterus  salmoides c   n  1  2  3  7  -1  2 1  ------  turn, leads to an increase in food intake and the probability of ingesting infected sources (Rohlenová et al. 2011). For example, Barson (2004) suggested that low prevalence in C. gariepinus occurred in winter due to a reduction in feeding activity and because Contracaecum eggs optimally hatch at temperatures 21 °C. According to Mashego (1989) and Smit and Luus-Powell (2012), parasite infection by Contracaecum spp. in hosts occupying the northern region of South Africa show no seasonal trends as climatic fluctuation in the region are slight and because the definitive hosts, such as White-breasted cormorant Phalacrocorax lucidus (Lichtenstein, 1823) and African Darter Anhinga rufa (Daudin, 1802), are present all year round (Hockey et al. 2005). Conversely, Amin (1985) and Scholz (1986) reported seasonal variation in the infection of helminth species in fish inhabiting temperate waters. Findings in this study concur with those of Mashego (1989), Szalai and Dick (1990) and Smit and Luus-Powell (2012) with regard to the cichlids tested, but for C. rendalli, C. gariepinus and S. intermedius, results are in agreement with those of Amin (1985) and Scholz (1986). Boomker (1994a) and Bergmann and Motta (2004) reported that nematode burden increases via ontogenic changes with infection being absent in juveniles but present in adults. Reasons for a greater presence of worms in larger individuals are attributed to a larger body cavity and/or the feeding preference of the second intermediate host (Aloo 1999;Al-Zubaidy 2009). These assumptions are supported by previous studies done on Barbus spp. (Mashego 1989), M. salmoides (Szalai and Dick 1990) and S. intermedius (Smit and Luus-Powell 2012). Failure to obtain a wide size range of O. mossambicus to establish any ontogenic shift in parasite burden can be attributed to the sampling methods used. In conclusion, it is suggested that future surveys adopt a protocol whereby similar numbers of different species be sampled between localities within a reasonable period and that all potential variables be recorded in order to better interpret factors affecting parasite burden.