Material and Methods
Collection details and sampling were explained by Martin et al. (2015). Specimens of the host bivalves were collected: monthly, from April 2011 to May 2012 and in January 2013, at Río San Pedro, 36°31’56.28”, 6°12’53.28” (S. plana); in January 2013 at Río San Pedro mouth, 36°31’47.28”, 6°14’52.80” (M. pellucida). The specimens of S. plana from May 2011 were damaged during sampling. All applicable international, national, and/or institutional guidelines for the care and use of animals were followed. Possible non-biotic influences in the observed patterns could not be determined because no environmental data were recorded during the study period.
In the laboratory, the collected specimens of S. plana and M. pellucida were opened to estimate the infestation intensity and prevalence. All obtained worms, and the bivalves harbouring them, were counted. The longest shell diameter of the hosts was measured (length, in mm) using callipers. Samples of S. plana (i.e. 100 to 300 specimens) were collected monthly and measured (independently of the presence of the symbiont) to define the seasonal size structure of the host population.
The width (in µm) of the tenth segment (parapodia included) was used as a proxy for symbiont size (WW). WW was measured under a Nikon SMZ645 stereomicroscope equipped with a micrometric ocular. Ripe females were identified by the presence of intracoelomic oocytes, which were usually visible through the body wall. In preserved specimens, the worm tissues become more opaque than in living ones, which usually hindered observation of the oocytes. Their presence was confirmed by placing the worms in a Petri dish and then cutting their body wall and applying a gentle pressure with the forceps so that present oocytes would spill from the coelomic cavity for observation. Digital images were taken with a CMEX camera, linked to a Zeiss Stemi 2000-c, stereomicroscope, using the ImageFocus 4.0 software by Euromex.
The sizes of the Iberian symbionts and their hosts were used to estimate their monthly size-class frequencies. The frequency of infested bivalves and that of the whole population was estimated separately and then expressed as percentages for comparison. The worm/host size and the ripe females/prevalence relationships were assessed by correlation analyses, while differences in percentage of females and prevalence during warm (i.e. March to September) and cold (i.e. October to February) periods were assessed by one-way Analysis of Variance (one-way ANOVA).
For morphometric purposes, preserved specimens of the Congolese population were kindly loaned by the United States National Museum of the Smithsonian Institution (USNM). Twenty-five specimens from the Iberian population, representative of all available sizes, were selected for the analyses, while for the Congolese population, 25 specimens were selected on the basis of the quality of the available specimens (as all of them were similar in size). The following characters were selected and measured (Fig. 1): WW, worm width without parapodia (WWP, µm), worm length (WL, µm), number of segments (NS), head width (HW, µm), head length (HL, µm), length of lateral antenna (LA, µm), length of palpophore (PP, µm), length of palpostyle (PS, µm), distance between anterior eyes (DAE, µm), distance between posterior eyes (DPE, µm), distance between anterior and posterior eyes (DAPE, µm), length of dorsal lobe (DL, µm), length of dorsal cirrophore (DCP, µm), length of dorsal cirrostyle (DCS, µm), length of posterior neurochaetal lobe (PNCL, µm), and length of ventral cirri (VC, µm) (measured at the 10th parapodia level). As cirrostyles alternate short and long along the body (Martin et al., 2015), the measurements were recorded for two parapodia (from chaetiger 10 to 30, depending on the specimen) bearing long and short cirrostyles, respectively, and indicated by adding L and S to the end of the acronym (e.g. “DCSL” and “DCSS” meaning dorsal cirrostyle from parapodia having long and short cirrostyles, respectively). Special care was addressed to avoid measurements on damaged appendages. The eyespot diameter, which is highly variable among the specimens of the two species (probably as a result to different individual responses to fixation) was not considered a valid taxonomic character and was not used in the statistical analyses. This also led us to measure all characters based on the eyes at the centre of each eyespot (Fig. 1). WWP, WL and NS were measured as WW, while the remaining characters were measured under a Motic BA210 binocular microscope equipped with a TOUPCAM™ U3CMOS digital camera, managed through the ToupView 3.7 software.
The inter-population differences were analysed for three different datasets: 1) raw data (direct measurements), 2) size-independent data; 3) taxonomically relevant character proportions (i.e., WL/WW, NS/WW, WWP/WW, DCPL/DLL, DCSL/DLL, DCSL/DCPL, VCL/PNCLL, DCPS/DLS, DCSS/DLS, DCSS/DCPS, VCS/PNCLS, HL/HW, LA/HL, PP/HL, PS/HL, PS/PP, DAE/HW, DPE/HW, DAPE/HL, DAE/DPE). Size-dependency of both measured characters and character proportions was assessed by Pearson correlation. Size-independent measurements were used without transformation, while size-dependent ones were divided by worm size (WW). The inter-population differences for the averaged measurements were estimated by one-way ANOVAs. Since multiple F-tests were carried out within each data set, the level of statistical significance was set at p < 0.05 and adjusted according to false discovery rate procedure (Benjamini and Hochberg, 1995). To avoid biases due to size differences, the description of the new species was based on the size-independent data as well as on the character proportions.
All character sets (i.e., raw data, size-independent data, proportions) were analysed by Principal Component Analysis (PCA) based on normalised data. The significance of the inter-population differences was explored by one-way analysis of similarity (ANOSIM) based on Euclidean distance resemblance matrices, while the contribution of each measured character to the distance within and between the two populations was assessed by the Similarity Percentages analysis (SIMPER) based on Euclidean distance. Forward stepwise discriminant analyses (FSDA) were used to obtain discriminant functions for the two populations under study, using the three datasets and assuming equal intra-population covariance matrices and threshold significance levels of 0.05 both to include and to remove variables. The similarity of the intra-population covariance matrices was assessed by the Box test (based on an asymptotic approach to the Fisher’s F index) and the differences between population-averaged vectors were assessed by the Rao approach to the Wilks’ Lambda test. The classification matrices were built by considering the sensitivity percentage (i.e. well classified positive events) as belonging to the Iberian population, and the specificity percentage (i.e., well classified negative events) as belonging to the Congolese population. Furthermore, to determine the probability that one of the observed specimens would effectively belong to one of the two populations, we also performed a cross-validation in which each observation was removed from the original matrix and the model and the forecast successively estimated, allowing a better adjust to the model (Huberty, 1994).
Pearson correlation analyses, one-way ANOVAs and discriminant analyses were performed with the XLSTAT software (2015.5.01.23039, copyright by Addinsoft 1995-2016). All remaining analyses were conducted using the PRIMER software, version 6.1.11, copyright by PRIMER-E Ltd. 2008 (Clarke and Warwick, 2001; Clarke and Gorley, 2006).