First report of an egg-predator nemertean worm in crabs from the south-eastern Pacific coast: Carcinonemertes camanchaco sp. nov

Nemertean worms belonging to the genus Carcinonemertes have been tied to the collapse of crab fisheries in the northeastern Pacific Ocean. A new species is described from egg masses of two commercial crabs, Cancer porteri and Romaleon setosum, inhabiting the central-north Chilean coast. This is the first species of Carcinonemertes described from the southeastern Pacific Ocean. Total body length of Carcinonemertes camanchaco sp. nov. ranged from 2.38 to 4.93 and from 4.29 to 8.92 mm, in males and females, respectively. Among others, traits that distinguish this new species from other previously described congeneric species include: presence of two gonad rows on each side of the intestine, a simple (not decorated) mucus sheath, and a relatively wide stylet basis. Maximum likelihood and Bayesian inference phylogenetic analyses distinguished this new species from all other species of Carcinonemertes with available cox1 sequences in GenBank. Prevalence and mean (± SD) intensity of C. camanchaco sp. nov. was 24% and 2.6 (± 2.07) worms per egg mass in C. porteri and 38.1% and 3.8 (± 2.4) worms per egg mass in R. setosum. The formal description of this new species represents the first step towards the understanding of this worm's impact on the health of crab fisheries in the southeastern Pacific Ocean.

Nemerteans belonging to the family Carcinonemertidae are voracious egg-predators that infect a variety of decapod crustaceans 1,2 . Some species are responsible for the collapse of crustacean fisheries in North America [3][4][5][6] given their negative effect on female reproductive performance 2,5,7,8 . Several species of commercial interest, such as the Caribbean spiny lobster Panulirus argus whose eggs are consumed by C. conanobrieni 9,10 , the Dungeness crab Metacarcinus magister infected by C. errans 5 , the Red king crab Paralithodes camtschaticus parasitized by C. regicides, and the sand crab Portunus pelagicus infected by C. mitsukurii 11 are used as hosts by carcinonemertid worms. One of the most studied nemertean worms is C. errans that during outbreaks causes considerable egg mortality in 50% or more of the brooding female crabs of Metacarcinus magister. Carcinonemertes errans has been tied to the collapse of this crab fishery in Central California early during the 1960s 1,5,12 .
The family Carcinonemertidae is comprised of two genera, Carcinonemertes with sixteen described species, and Ovicides with five described species 10 . Members of the genus Carcinonemertes have a single stylet and no accessory pouches while the genus Ovicides invariably exhibit accessory pouches and more than one stylet. All Carcinonemertes are gonochoric (separate sexes) while representatives of Ovicides can be either gonochoric or simultaneously hermaphroditic 13,14 . The life cycle of nemertean parasites is variable and appears to be adapted to the reproductive cycle and brooding pattern of their respective crustacean hosts 15 . At one extreme, some nemertean parasites exhibit a rather simple and short life cycle in which the worms feed, mature, and reproduce on a single ovigerous female host. In this case, host autoinfection is common (i.e., C. regicides in the Red king

Results
Prevalence of nemerteans per host species and locality. A total of 48 female crabs (non-ovigerous) of Cancer porteri from Iquique (20° S), 50 gravid females of Romaleon setosum from Coquimbo (30° S), and 21 gravid females of C. porteri from Valparaíso (33° S) were examined. Nemerteans were only detected in egg masses of gravid female crabs (Fig. 1); therefore, the prevalence of nemerteans in specimens from Iquique was 0%. The prevalence of nemerteans in C. porteri and R. setosum from Coquimbo and Valparaíso was 24.0% and 38.1%, respectively. We found several nemertean developmental stages (eggs, juveniles, and adult worms) infecting the egg masses of the infected crabs. Juvenile worms were found either roaming among the crab eggs or encapsulated on the surface of the crab's abdomen. Adult worms were found only in egg masses, covered by a mucus sheath (not ornamented). Long strands of worm eggs covered by translucent mucus were also observed among the eggs of 'berried' (= brooding) female crabs.
Morphometric analyses. The studied male worms had a mean ± SD body size of 3.64 ± 0.59 mm and the female worms had a mean body size of 5.84 ± 1.25 mm. A t-test demonstrated statistically significant differences in average body size between male and female worms (t = 4.42, d.f. = 19, P < 0.01), with the females being larger than the males. For both male and female worms, body length was not correlated with stylet length, basis length, and stylet-basis ratio (Spearman correlations; all P > 0.05).
The PCAs performed separately for female and male worms infecting the studied crabs in the SEP and those from other four closely related species of Carcinonemertes, characterized by having two gonad rows on each side of the intestine, indicated that > 95% of the variability in body proportions among the studied specimens and species was explained by the first three PCA axes (Table 1). EW/BL, DW/BL, BW/BL, and SL/SBL/BL (factor loads > 0.95; Table 1) were the most important measures that distinguished both male and female nemertean worms collected during this study from specimens belonging to the other four similar, previously described species (Fig. 2a, b). The DFA, using the orthogonal components of the PCA, revealed significant differences among female (Wilks' Lambda = 0.01; F (12, 21) = 7.62; p < 0.01) and male (Wilks' Lambda = 0.001; F (12, 21) = 24.96; p < 0.01) nemertean specimens of the four species. Additionally, the DA showed that 93% of the female specimens and 100% of the male specimens were correctly assigned to their respective nemertean species. , demonstrated that the specimens of Carcinonemertes infesting R. setosum (three sequences) and C. porteri (five sequences) belong to the same species. Furthermore, both ML and BI phylogenetic reconstructions indicated that the studied specimens of Carcinonemertes form a clade sister to C. errans, a worm that infects Metacarcinus magister in the coast of California (Fig. 3). The intra-specific genetic distances calculated for the 8 sequences of this study varied between  Table 1. Factor loadings of the PCA using morphometric data obtained from females and males of Carcinonemertes camanchaco sp. nov., based on a correlation matrix. Measurements: BL = total body length, BW = body width, EL = eye length, EW = eye width, MCW = middle chamber width, SL = stylet length, SBL = stylet basis length, PPCL = posterior proboscis chamber length.  Table 2). The phylogenetic trees also indicated that the genus Carcinonemertes is paraphyletic, given the inclusion of Ovicides in the same well-supported phylogenetic clade containing all species of Carcinonemertes (Fig. 3).
Morphological characterization of specimens from the SEP. We described morphological characteristics including body color and body measurements of female and male nemertean worms. Comparisons of morphology, morphometry, host species and geographical distributions of Carcinonemertes camanchaco sp. nov.    Prevalence and mean intensity (± SD) 24.0% and 2.6 (± 2.07) in Cancer porteri (Valparaiso) and 38.1% and 3.8 (± 2.4) in Romaleon setosum (Coquimbo).

Discussion
This is the first study that has tested for the presence/absence of nemertean egg-predators in commercial crabs from the SEP coast. We did find nemertean worms infecting two species of crabs from the SEP and our morphometric and molecular analyses (based on the mitochondrial cox1 gene) indicated that these nemerteans parasitizing eggs of Romaleon setosum and Cancer porteri belong to the same species of Carcinonemertes, here named C. camanchaco sp. nov. The same analyses also indicate that this newly discovered entity is genetically different from all other known species of Carcinonemertes with information available in GenBank. We note that the cox1 is not useful to distinguish among species in some clades within the Nemertea (Sundberg et al., 2016). An integrative taxonomic approach, including the (classical) comparison of morphological traits, multivariate morphometric analyses, and molecular data (cox1 mitochondrial gene) demonstrated that the nemertean inhabiting two sympatric crab species is a new species. The nemertean species recorded in our study is characterized by the presence of two ovaries on each side of the intestine and the production of a simple mucus sheath (not ornamented). The combination of these morphological characters plus genetic characters clearly distinguished our specimens from previously described species in the same genus. The genus Carcinonemertes now comprises 17 valid species. However, genetic information (i.e., cox1) exist for only five species (plus the new species), which were used here to construct a phylogenetic tree. This tree showed that C. conanobrieni, parasitizing Caribbean spiny lobsters, is sister to a clade comprised of congeneric species infecting brachyuran crabs (Fig. 3). Only two nemertean species have been reported before from crabs belonging to the family Cancridae: C. errans described from Metacarcinus magister 33 and C. camanchaco sp. nov. from Cancer porteri and Romaleon setosum (this study). The phylogenetic tree indicated that C. camanchaco sp. nov. and C. errans are sister species (with 3.6% of genetic distance). These nemertean species parasitizing two phylogenetically close host species, belonging to the family Cancridae, implies that the genus Carcinonemertes could be coevolving together with different species used as hosts 6,14 .
Estimating the prevalence of nemertean egg-predators is exceedingly relevant considering the impact that Carcinonemertes and Ovicides have on the reproductive performance and population dynamics of their host species 34,35 . Notably, studies reporting the prevalence of nemerteans are rare. Santos et al. 13 found that the prevalence of C. divae in female and male crabs of Libinia spinosa was 87% and 43.2%, respectively. In turn, C. errans prevalence in Metacarcinus magister is remarkably high; up to 97% of the gravid females can be infected by this nemertean during one year/ fishing season. Importantly, the intensity of C. errans on infected Metacarcinus magister can also be very high 5 . In the 70's, C. errans became an epidemic, resulting in the direct mortality of an average of 55% of the eggs produced by females of M. magister. The above suggests that C. errans might probably  www.nature.com/scientificreports/ be one of the most numerically significant predators on these crabs and was responsible for the collapse of this fishery in Central California 5,12,33 . The prevalence of C. camanchaco sp. nov. was lower than that reported in the aforementioned species (38% in R. setosum and 24% in C. porteri). Given the previously demonstrated impact of nemerteans on the fecundity of their respective host species, and the considerable and steady decreases in landings reported for R. setosum and C. porteri along the coast of Chile in the last four years, we argue that additional studies are needed to evaluate the effect of C. camanchaco sp. nov. on the fertility and population dynamics of these crabs. The prevalence of several parasite species can vary across short-and long-time scales and can be associated to environmental variability and/or host availability 11 . Shields and Kuris 3 showed that prevalence of C. epialti on the shore crab Hemigrapsus oregonensis varied between years, finding that during non-outbreak periods prevalence was 48%, while during the outbreak periods this prevalence increased to 97%. Campbell et al. 24 showed that prevalence of C. australiensis was 67% in females of the spiny lobster Panulirus cygnus. Most recently, Baeza et al. 35 determined that the prevalence of C. conanobrieni in the lobster Panulirus argus was 7.4%, and Simpson et al. 10 showed that prevalence increased to 93.9% a year later (107 of 114 gravid females). This information suggests that nemertean infection levels can change quickly between consecutive years. Therefore, we argue in favor of additional studies to determine short-and long-term spatial and temporal variability in the prevalence and intensity of C. camanchaco sp. nov. in the different species infected, as well as the effect of this parasite on the reproductive performance and overall health of its crab hosts. Carrying out temporal monitoring of the newly discovered nemertean species across host species is needed to manage these heavily fished species towards the goal of sustainability in Chile.

Methods
Sampling and parasite collection. Females of Cancer porteri and Romaleon setosum were collected by divers from three localities, Caleta Chanavayita, Iquique (20°42′10″S, 70°11′15″W), Puerto Aldea, Coquimbo (30°17´37´´S, 71°36´36´´W), and Caleta Portales, Valparaíso (33°1′52.74″S, 71°35′25.18″W). We collected a total of 48 crabs belonging to Cancer porteri from Iquique, 50 crabs belonging to Romaleon setosum from Coquimbo, and 21 specimens of C. porteri from Valparaíso. The samples of C. porteri collected in Iquique and Valparaíso were transported to the laboratory at the Universidad de Antofagasta, and alive specimens of R. setosum were transported to the laboratory at the Universidad Católica del Norte, Chile. Collection of these specimens was authorized by the Fishery and Aquaculture Sub-Secretary of the government of Chile (R. Ex N° E-2020-481).
In the laboratory, each crab (either alive or frozen [− 10 °C]) was examined for the presence or absence of nemertean worms. All eight pleopods were removed from each studied crab, and all embryos were gently stripped away from the pleopods of each gravid female using fine forceps and placed into Petri dishes with filtered seawater (10 µm). The egg mass of each gravid female, as well as the abdomen, pleopods, gills, and branchial chambers from all analyzed crabs were inspected under a stereomicroscope (Leica M80, Leica Microsystems, Wetzlar, Germany) to determinate the presence or absence of nemertean worms. Nemerteans collected from gravid female crabs were placed in Petri dishes with filtered seawater up to the point in time when measurements and photographs were taken, and morphological characters were studied in detail. Nemerteans were first relaxed in a 1:1 solution of 1 M MgCl 2 (prepared with distilled water) and seawater for 1-5 min. The specimens were fixed in formaldehyde at 10% for morphological examination and 99% EtOH solution for genetic analyses.
Morphometric study and morphological characterization. Different body parts were measured in all adult worms (in mm, precision = 0.01 mm) using a digital camera (Mshot MD90) connected to a stereomicroscope (Olympus SZX7, Tokyo, Japan) and the programs MShot Image Analysis System 1.5.2 and ImageJ 36 . We measured total body length (BL), body width (BW), ocelli length (OL), ocelli width (OW), diaphragm width (DW), stylet length (SL), and stylet basis length (SBL) of each studied specimen. The different body measurements of the studied specimens were compared to those from all other previously described species of Carcinonemertes (n = 16).
Principal component analyses (PCAs) were also performed separately for both male and female worms, using body part measurements standardized by BL. For these analyses, only previously described nemertean species with taxonomic characteristics most similar to the specimens collected during this study were used; only species that exhibit two rows of gonads (i.e., Carcinonemertes coei, C. errans, C. regicides, C. wickhami). The body proportions used were BW/BL, OL/BL, OW/BL, DW/BL, SL/BL, SL/SBL, SBL/BL and SL/SBL/BL. The first two axes of the PCA were also used to evaluate the importance of each measured ratio on the morphometric variability observed among different species of Carcinonemertes. Finally, orthogonal variables (principal components), which preserve the original information of the database, were used to perform a Discriminant Function Analysis (DFA) and to assign specimens to the different worm species 37,38 . All statistical analyses were conducted using the software Statistica 10.0 (Statsoft, Tulsa, USA).

DNA isolation, PCR amplification and sequencing.
A total of eight juvenile worms were dissected with forceps and DNA was isolated from each one of them separately using a protocol modified from Miller et al. 39 that included treatment of each specimen with sodium dodecyl sulphate and digestion with Proteinase K. Proteins were removed by precipitation with NaCl, and the DNA was finally precipitated with ethanol. The resultant DNA was eluted in nuclease-free water and quantified using a spectrophotometer until a concentration of 30 to 50 ng/μL was reached.
The mitochondrial gene cytochrome c oxidase 1 (cox1) was amplified using universal primers described in Folmer et al. 40  www.nature.com/scientificreports/ deoxynucleotide triphosphate (dNTP; 10 mM), 1.4 ul 10 pM each primer, and 3 μl template DNA. PCR amplifications were performed in a Boeco Ecogermany M-240R thermal cycler using the following optimal cycling conditions: 95 °C (5 min), followed by 40 cycles at 95 °C (45 s), 50 °C (45 s) and 72 °C (1 min), and a final extension step at 72 °C (10 min). PCR products were sent to Macrogen (Seoul, Korea; http:// www. macro gen. com) for purification and sequencing of both forward and reverse DNA strands. Sequences were edited and contigs were assembled using ProSeq 2.9 beta 41 . All haplotype sequences obtained during this study were deposited in GenBank under the following accession numbers MW596479-MW596486.
Phylogenetic analyses. Four other species of Carcinonemertes were used as ingroup terminals for molecular comparisons with our new species, and two species of Nipponnemertes were included in the analysis as outgroup terminals. All sequences were first aligned using the software Clustal W 42 . The aligned dataset was analyzed with the software JModelTest2 43 which compares different models of DNA substitution in a hierarchical hypothesis-testing framework to select a base substitution model that best fits the data for each gene. The best model found by JModelTest2, selected with the corrected Akaike information criterion 44 , was GTR + I. The model parameters were as follows: assumed nucleotide frequencies A = 0.1988, C = 0.1266, G = 0.2261, and T = 0.4485; substitution rate matrix with A-C substitution = 0.0025, A-G = 11.4049, A-T = 2.8337, C-G = 0.1128, C-T = 7.4120, G-T = 1.000, and p-inv distribution with shape parameter 0.6420. Next, the best model was implemented in MrBayes 3.2.7a 45 for Bayesian Inference analysis (BI) and in IQ-TREE 46 for Maximum Likelihood analysis (ML). All phylogenetic analyses were conducted in the CIPRES Science Gateway V. 3.3 platform (http:// www. phylo. org/) 47 .
For the BI analysis, unique random starting trees were used in the Metropolis-coupled MCMC 45 . The analysis was performed for a total of 5,000,000 generations. Visual inspection of log-likelihood scores against generation time indicated that the log-likelihood values reached a stable equilibrium before the 100,000th generations. Thus, a burn-in of 1000 samples was conducted; every 100th tree was sampled from the MCMC analysis, obtaining a total of 100,000 trees, and a consensus tree with the 50% majority rule was calculated for the last 59,000 sampled trees. Support for nodes in the BI tree topology was obtained by posterior probability. For the ML analysis, we used the default options in IQ-TREE run through the Cypress Science Gateway 47 . The robustness of the ML tree topology was assessed by bootstrap iterations of the observed data 1000 times. Phylogenetic trees were visualized and edited in Figtree 1.4.4.
Pairwise genetic distances (intra-and inter-specific) between the sequences of cox1 gene were calculated in MEGA 6 48 using the Kimura 2-Parameter model 49 .