Intestinal helminths were collected from a total of 233 striped dolphins stranded along the east coast of Spain between 1982 and 2016. Permission to collect stranded dolphins was given by the wildlife services of the regional governments, which are the official institutions in charge of managing and protecting wildlife. During the 1990 epizootic, the geographic coverage included the coasts of Catalonia, Valencian Community, Murcia and eastern Andalusia; in the remaining years, sampling was restricted to the Valencian Community ( Figure 1 ). The sample included the 52 animals examined by Mateu et al. (2014), as well as 6 of the 13 dolphins examined by Raga and Carbonell (1985), for which at least information on occurrence of all intestinal helminths was available. The remaining 175 dolphins were examined in this study, and included dolphins from every year within the time series, in order to avoid significant gaps that could affect the results. All dolphins in the final sample were > 100 cm long to exclude calves that had consumed milk only.

Before necropsy, total body length was measured and sex was determined for all animals. Intestines were removed and stored at -20°C. After thawing, intestines were weighed, measured and divided into 20 sections of equal length. Each section was opened and washed over a 0.02 mm sieve using tap water. The solid content was collected in a Petri dish and examined under a stereomicroscope. The intestinal wall was also thoroughly examined for attached helminths. Acanthocephalans were left in tap water for 24 hours at 4°C to allow the proboscis to evert. All parasites were washed in saline (9 g/l), fixed and preserved in 70% ethanol, and later identified to the lowest taxonomic level possible. Cestodes were stained with iron acetocarmine (Georgiev, 1986) and identified following Delyamure (1955) and Raga, (1985). Acanthocephalans were cleared in lactophenol, drawn using a light microscope connected to a drawing tube, and identified following Petrochenko (1958); Amin and Margolis (1998) and Costa et al. (2000). All parasites were counted except in 12 dolphins stranded before 1988, for which only presence/absence data were recorded. ‘Small’ and ‘large’ plerocercoids of tetraphyllidean cestodes were not considered members of the intestinal assemblage and were not included in this study (see Aznar et al., 2007).

All individuals identified as B. vasculosum based on morphologic traits were juveniles (see the Results section), thus raising the possibility that they actually represent immature forms of other Bolbosoma species. We investigated this possibility using molecular barcoding. Two species of Bolbosoma have been recorded as adults in the Mediterranean, i.e., Bolbosoma balaenae in fin whales, Balaenoptera physalus, and Bolbosoma capitatum in long-finned pilot whales, Globicephala melas (Delyamure, 1955; Yamaguti, 1963; Raga and Balbuena, 1993; Santoro et al., 2021). We sequenced a partial (479 bp) region of the mitochondrial cytochrome c oxidase subunit I (cox1) of (i) 5 specimens of B. vasculosum from our sample (DNA extraction was attempted on a number of additional specimens, but it was unsuccessful due to poor preservation); (ii) two juvenile specimens of putative B. balaenae also collected from our sample of striped dolphins, and (iii) one adult specimen of B. capitatum from a long-finned pilot whale, Globicephala melas sampled in the study area, which was available at the collection of the Marine Zoology Department of the University of Valencia, Spain (voucher specimen ID: 13653). Available cox1 sequences from GenBank from adults of B. balaenae infecting Mediterranean fin whales, as well as from two other Bolbosoma species, were also used for comparison ( Table 1 ).

Total genomic DNA was extracted from specimens using the DNeasy Blood and Tissue Kit (Qiagen), following manufacturer’s instructions. The cox1 region was amplified using primers forward LCO1490F (5’-GGTCAACAAATCATAAAGATATTGG-3’) and reverse HC02198 (5’-TAAACTTCAGGGTGACCAAAAAATCA-3’) (Folmer et al., 1994). The concentration and quality of the extracted DNA was measured using a NanoDrop™ One (Thermo Scientific). PCR amplification reactions were performed in a total volume of 20 μL, including 1.6 μL of both forward and reverse primers (final concentration of 5 μM), 2 μL of template DNA, 4.8 μL of PCR water and 10 μL of MyFi™ DNA Polymerase (BioLine, Meridian Life Science Inc., Taunton, MA, USA). Cycling conditions were 94°C for 5 min for an initial denaturation, then 38 cycles of 94°C for 45 s, 48°C for 45 s, and 72°C for 80 s, followed by a final extension at 72°C for 7 min. Positive and negative (no template DNA) controls were used in each PCR reaction. PCR products were purified using the Nucleospin^® PCR and Gel Purification Clean-up kit (Machery-Nagel, Düren, Germany). Purified amplicons were sent to Macrogen Europe (Amsterdam, The Netherlands) for sequencing with the same primers used in the PCR amplifications.

Nucleotide sequences from both strands were used to assemble consensus sequences using Geneious R7 (https://www.geneious.com). The identity of the assembled sequences was verified using the NCBI Basic Local Alignment Search Tool (BLAST). The obtained cox1 sequences, together with all cox1 sequences of Bolbosoma available in GenBank, were aligned and trimmed using Geneious R7. The accuracy of the alignment was inspected by checking the amino acid translation, using the invertebrate mitochondrial code. The total number of sequences included in the final alignment was 23. Pairwise genetic distances were calculated using the Maximum Composite Likelihood model (Tamura et al., 2004) in MEGA v.11 (Tamura et al., 2021).

Ecological terms follow Bush et al. (1997). Infection parameters were determined using the online software Quantitative Parasitology 1.0.15 (Reiczigel et al., 2019). Prevalence is the percentage of hosts that are infected by a specific helminth species, mean abundance is the average number of individuals of a helminth species per examined host, and mean intensity is the average number of individuals of a helminth species per infected host. The 95% confidence interval (CI) for prevalence was calculated with Sterne’s exact method (Reiczigel, 2003). The 95% CI for mean abundance and mean intensity of each helminth species were obtained with the bias-corrected and accelerated bootstrap method using 10000 replications (Rózsa et al., 2000).

Our dolphin sample was not a random sample of the wild striped dolphin population, but a subset of stranded animals that died from varied causes. Thus, a pertinent question is to what extent our sample could be biased regarding helminth infections if, e.g. it is enriched with heavily parasitized dolphins. We carried out a preliminary exploration of this question based on the comparison of two dolphin subsamples collected in the same period (2010-2015) for which the putative cause of death could reliably be determined based on a thorough veterinary analysis. The first one (N = 10) was composed of striped dolphins that were killed by common bottlenose dolphins, Tursiops truncatus or as a result of fisheries interactions (Crespo-Picazo et al., 2021; Izquierdo-Serrano et al., 2022). In principle, these causes of death are not related with the health condition of dolphins and should be little influenced by the infections of intestinal helminths, which were light (see the Results section). In fact, veterinary analyses indicated good body condition and no pathologies were detected. The second subsample (N = 20) included dolphins that died from diverse infectious syndromes (Unpub. data). Therefore, we hypothesized that these animals would be more prone to concomitant recruitment of TTPs due to a compromised immune response. Potential differences in abundance of each helminth species between the two subsamples were tested with Mann-Whitney tests.

The geographic analysis was carried out using 74 dolphins that died during the DMV outbreak in the summer and autumn of 1990 and stranded along the coast from Andalusia to Catalonia ( Figure 1 ). This subsample was considered suitable for three reasons. First, it was the only sample of dolphins collected across a wide geographical range; in other years (including 2007, when the second DMV epizootic occurred), sampling was restricted to the Valencian Community; second, it was composed of animals that died from the same disease; and third, it was temporally restricted, thus reducing potential confounding seasonal or inter-annual effects.

We used two approaches to investigate latitudinal variability in helminth infections in the stranded dolphins. First, the coast was divided into four regions based on latitude, distance from coastline to continental slope, and current circulation schemes (Martínez et al., 2022) ( Figure 1 ). The numbers of dolphins from regions 1 to 4 were 9, 33, 16 and 16, respectively ( Figure 1 ). Differences in prevalence of each helminth species among regions were tested with exact Chi squared tests. Differences in helminth community structure were investigated using PERMANCOVA with the software PERMANOVA + for PRIMER (Anderson et al., 2008). Abundances of parasites were firstly fourth-root transformed and a Bray-Curtis similarity matrix was obtained. The full model included ‘Region’ (fixed factor) and two relevant predictors associated with individual dolphins, i.e. ‘Sex’ (fixed factor) and ‘Length’ (fixed covariate), as well as their interactions. Pseudo-F statistics under a true null hypothesis were obtained by using 20,000 permutations. Akaike Information Criterion (AIC) values were used to compare competing models with different numbers of parameters. The model with minimum AIC for small sample sizes (AIC_c) was considered the best model, and the rest of the models were ranked according to increasing AIC_c values (Johnson and Omland, 2004; Anderson et al., 2008; Aho et al., 2014). Models with values of ΔAIC_c ≤ 2 with respect to the best model were considered to have substantial empirical support, whereas those with ΔAIC_c > 4 were considered to have considerably less support (Anderson and Burnham, 2004). Akaike weights (w_i) were calculated according to Anderson and Burnham (2004). It was assumed that models with w_i ≤ 0.05 were unlikely to be the ‘true’ models (Anderson and Burnham, 2004).

Second, we used Generalised Additive Models (GAMs) (Zuur et al., 2009; Wood, 2017) to explore non-linear patterns in the abundance, or likelihood of occurrence, of each helminth taxon with latitude of stranding. Abundance and occurrence (presence/absence) of each parasite species were used as the response variable for each type of model, respectively, and ‘Latitude’, ‘Dolphin sex’ and ‘Dolphin length’ as the explanatory variables. For abundance data, negative binomial GAMs were fitted using a log link function; for occurrence data, binomial GAMs were fitted using a clog-log link function, which is recommended when there are considerably more zeros than ones or vice versa (Zuur et al., 2009). Model selection followed the same criteria used for the PERMANCOVA models. Collinearity between variables was checked by calculating the Variance Inflation Factor (VIF); values in between 1 and 5 were considered to indicate low collinearity (Goldstein, 1993). We further ensured the lack of dependency between covariates and factors (‘Sex’) with Mann-Whitney tests, and between the two covariates with Spearman correlation. Models were validated following Zuur et al. (2009), i.e., checking normality, homogeneity of variance, autocorrelation and over-dispersion of residuals.

GAMs were also used to investigate long-term infection changes of each helminth species. The initial models included two alternative response variables (helminth abundance or occurrence) and five predictors, i.e. ‘Group’ (with two levels, i.e., epizootic and non-epizootic)’, ‘Dolphin sex’ and ‘Season’ as fixed factors; and ‘Year’ and ‘Dolphin length’ as covariates for which smoothers were applied. Alternative models excluding dolphins affected by the DMV, or using ‘Julian day’ instead of ‘Year’, were also fitted, but the results were nearly identical to those of the models described above and are not shown. Note that ‘Latitude’ was not included in the temporal analysis because, in years other than 1990, all stranded dolphins were collected on the coasts of the Valencian Community ( Figure 1 ).

For abundance data, negative binomial GAMs were fitted using a log link function; for occurrence data, binomial GAMs were fitted using a clog-log link function. Model selection and validation, as well as collinearity analysis between predictors, were carried out as indicated in the previous section. However, AIC was not corrected for small samples in this case because sample size (n) was large compared with the number of parameters (k), i.e. n/k > 40 (Anderson and Burnham, 2004). The lack of dependency between factors was further examined with Chi Squared tests; between covariates and factors with Kruskal-Wallis (‘Season’) or Mann-Whitney (‘Group’, ‘Sex’) tests, and between the two covariates with Spearman correlation.

All GAM models were fitted using the mgcv package (Wood, 2011) available in R version 4.2.2 (R Core Team, 2022).

Six helminth species were found in the intestine of the striped dolphins analysed. Three tetrabothriid cestodes included adult specimens, i.e. Trigonocotyle globicephalae, Tetrabothrius forsteri and Strobilocephalus triangularis, but only the two latter exhibited a prevalence > 10% and mean intensity > 10 individuals/dolphin ( Table 2 ). A single juvenile cestode belonging to the genus Diphyllobothrium was detected in a dolphin from 1987, but identification at species level was not possible due to poor preservation ( Table 2 ). Finally, two species of acanthocephalan were found as juvenile forms. Bolbosoma vasculosum showed an overall prevalence >15%, whereas single specimens of B. balaenae were found in two dolphins from 2011 and 2013, respectively ( Table 2 ).

The ranges of intraspecific genetic distances of Bolbosoma spp. based on cox1 sequences were as follows: 0.006 - 0.024 (B. vasculosum); 0 - 0.017 (B. balaenae); 0.004 – 0.109 (B. turbinella) and 0.004 (B. nipponicum) ( Table S1 ). At interspecific level, specimens of B. vasculosum were indistinguishable from the specimen of B. capitatum (range: 0.006 - 0.024), showing higher distances with respect to other Bolbosoma spp. (minimum distance: 0.194) ( Table S1 ).

Infection parameters of helminths from striped dolphins that were killed by interactions with bottlenose dolphins or fisheries were closely similar to those from dolphins that died from infectious diseases ( Table 3 ). In fact, we did not find significant differences in abundance for any of the helminth species (Mann-Whitney tests, all p >> 0.680).

Of the four species found in 1990, T. forsteri, S. triangularis and B. vasculosum were detected throughout the whole study area, whereas Tri. globicephalae was apparently restricted to its southern part ( Figure 2 ). However, we failed to detect differences in prevalence of this species between regions (exact Chi-Squared, χ2 = 2.308, 3 df, p = 0.538), nor a significant latitudinal trend (χ2 = 1.934, 1 df, p = 0.121).

There were four PERMANCOVA models that had substantial empirical support (ΔAIC_c ≤ 2), and two that had less support (2 < ΔAIC_c < 4). Among them, the simplest and with highest w_i value included the intercept only ( Table 4 ). In the rest of the models, none of the individual predictors was found to be significant ( Table 4 ).

GAMs could be fitted for T. forsteri, S. triangularis and B. vasculosum. In the case of T. forsteri, the model with lowest AIC_c, highest w_i and lowest number of predictors included ‘Latitude’ only. There was a significant but modest positive effect of this variable on the abundance of T. forsteri (deviance explained = 13.3%, edf = 3.071, p = 0.0154) ( Figure 3 ). Three additional negative-binomial GAMs also had substantial empirical support (ΔAIC_c ≤ 2), namely, those including ‘Latitude’ and ‘Dolphin length’ (ΔAIC_c = 0.385), ‘Latitude’ and ‘Dolphin Sex’ (ΔAIC_c = 1.321) or the three predictors (ΔAIC_c = 1.501). In all of them, the only significant variable was ‘Latitude’ (p < 0.03). In the GAMs for the abundance of S. triangularis and B. vasculosum, none of the predictors were found to be significant, and the same occurred in the models involving occurrence data (data not shown).

The cestodes T. forsteri, S. triangularis and Tri. globicephalae were detected throughout the whole study period, but the acanthocephalan B. vasculosum was almost exclusively found in 1990; only two additional individuals were found in 1989 and 1997, respectively ( Figure 4 ).

Long-term trends could be investigated with GAMs in the case of T. forsteri and S. triangularis. Binomial GAMs could not meaningfully be fitted for T. forsteri due to its high prevalence (> 94%). There were four negative-binomial GAMs for abundance with substantial empirical support (ΔAIC ≤ 2) and two with less support (2 < ΔAIC < 4) ( Table 5 ). The two simplest models included two predictors ( Table 5 ). Among them, the one with the highest w_i value explained < 10% of deviance and included the predictors ‘Year’ and ‘Dolphin length’ ( Table 5 ). In this model, the effect of ‘Dolphin length’ was highly significant, and that of ‘Year’ just marginally so ( Table 5 ). The abundance of T. forsteri experienced a slight decrease with a subsequent recovery throughout the study period ( Figure 5A ), and peaked at dolphin lengths between 130 and 150 cm, then decreased ( Figure 5B ). All the other competing models included ‘Length’ as a highly significant predictor; ‘Year’, ‘Group (epizootic vs. non-epizootic)’ or ‘Dolphin sex’ were included in some models, but their effects were only marginally significant, or not significant ( Table 5 ).

In the case of Binomial GAMs accounting for the occurrence of S. triangularis in striped dolphins, there were three models with substantial empirical support (ΔAIC ≤ 2) and two with less support (2 < ΔAIC < 4) ( Table 6 ). The model with the highest w_i value included the predictors ‘Year’, ‘Dolphin length’ and ‘Dolphin sex’, while the simplest model included ‘Year’ and ‘Dolphin length’ ( Table 6 ). The deviance explained for these models was 12.5% and 10.6%, respectively ( Table 6 ). The effects of ‘Year’ and ‘Dolphin length’ were statistically significant in both models ( Table 6 ). The likelihood of occurrence of S. triangularis strongly decreased from ca. 1990 (note the wide 95% CI in previous years) to 1997-1999, then increased, but without reaching the initial levels of the time series ( Figure 5C ). The likelihood of occurrence also decreased monotonically at increasing dolphin length ( Figure 5D ). In the other models; ‘Year’ and ‘Dolphin length’ remained as significant predictors; ‘Group (epizootic vs. non-epizootic)’, ‘Dolphin sex’ and ‘Season’ were included only in some models, having just marginally significant, or not significant effects ( Table 6 ). None of the negative binomial GAMs for the abundance of S. triangularis had a suitable fit according to model validation plots, thus they are not considered further.

The sample used in this study is composed of stranded dolphins, thus a potential concern is whether they represent an unbiased subset of animals from the wild population. First, the DMV that caused the epizootics in 1990 and 2007 is known to impair the host’s immune system (Bossart et al., 2011; Van Bressem et al., 2014). Although this effect could facilitate the establishment of parasites, thus abnormally increasing their loads (Aznar et al., 1994; Aznar et al., 2005), this is unlikely to occur in the case of TTPs. Mammals affected by other morbillivirus diseases have been reported to die between 8 hours and 2 weeks after the first clinical symptoms (Harder et al., 1990; Evermann et al., 2001; Mahy and Van Regenmortel, 2008). If the disease caused by DMV also runs its course quickly (Kennedy, 1998), the temporal window for recruitment or senescence of the helminths infecting striped dolphins would be too narrow to generate significant alterations of previous parasite loads (Mateu et al., 2014).

Second, the cause of stranding for other dolphins was not always easy to assess but included, inter alia, by-catch (Cuvertoret-Sanz et al., 2020; Izquierdo-Serrano et al., 2022), deadly interactions with bottlenose dolphins (Crespo-Picazo et al., 2021) and infectious diseases (Isidoro-Ayza et al., 2014; Cuvertoret-Sanz et al., 2020). Our results indicate that infection parameters were closely similar for ill vs. non-ill dolphins. One could wonder whether the striped dolphins belonging to the latter category exhibited non-lethal pathological conditions that made them prone to accidental catch or aggression by bottlenose dolphins; however, no such conditions were detected in any of them during the necropsies. Besides, none of the intestinal parasites found, except perhaps Diphyllobothrium sp., have pathological significance (Geraci and St. Aubin, 1987; Cuvertoret-Sanz et al., 2020), thus they were unlikely to be involved in dolphin stranding. In addition, post-mortem infection is not a possibility, since TTPs are transferred to dolphins through the consumption of infected prey. Accordingly, the load of intestinal helminths in our dolphin sample is not expected to be significantly higher than that of the wild population.

Out of the 6 helminth species found, 4, i.e., T. forsteri, Tri. globicephalae, S. triangularis and B. vasculosum, had previously been reported by Mateu et al. (2014) based on a subset of 52 striped dolphins from 1990, now included in the present study. Mateu et al. (2014) identified the juvenile specimens of Bolbosoma as B. vasculosum, as we do here, based on the most detailed descriptions of this taxon (Van Cleave, 1953; Costa et al., 2000). However, the absence of adults in a large sample of worms (n = 84, collected from 37 dolphins), which agrees with previous reports in cetaceans (Van Cleave, 1953; Fernández et al., 2004; Fraija-Fernández et al., 2016), indicates that these specimens must be juveniles of another Bolbosoma species. We cannot rule out the possibility that they belong to a new species associated with striped dolphins, but the absence of adults rather suggests that they correspond to an existing species present in other cetacean hosts. In the western Mediterranean, the only two additional species of Bolbosoma are B. balaenae and B. capitatum (Raga and Balbuena, 1993; Santoro et al., 2021). Molecular barcoding showed that our specimens of B. vasculosum were indistinguishable from B. capitatum. Nevertheless, further research using a higher number of markers and Bolbosoma specimens should be conducted to confirm our results. Since morphological similarities are also apparent between both species (Amin and Margolis, 1998; Costa et al., 2000), a taxonomic re-analysis of the concept of B. vasculosum is warranted.

Bolbosoma balaenae and Diphyllobothrium sp., were not detected by Mateu et al. (2014) and constitute exceptional records in the present survey. The former species typically infects baleen whales worldwide (Fraija-Fernández et al., 2016), with only two previous records in Stenella spp. (Dailey and Perrin, 1973). In the Mediterranean Sea, adults of B. balaenae have been reported in the only baleen whale species regularly present in this area, the fin whale Balaenoptera physalus (Santoro et al., 2021). The occurrence of juvenile specimens in striped dolphins could be interpreted as an accidental infection through shared prey.

With regard to Diphyllobothrium sp., a single immature specimen was found in our sample. The only previous record of species of Diphyllobothrium in striped dolphins corresponds to an adult of an unidentified species in the Atlantic Ocean (Jaber et al., 2006). In the study area, there are records of adults of an unidentified species of Diphyllobothrium in common bottlenose dolphins Tursiops truncatus (Raga, 1985; Quiñones et al., 2013). Since at least 7 species of this genus have been reported in odontocetes (Fraija-Fernández et al., 2016), further molecular studies are needed to determine the identity of the species occurring in our study area.

Mateu et al. (2014) did not detect a significant effect of host body length or sex on the abundance of any intestinal helminth taxa. Using now a larger sample size, we neither detected a sex effect, nor seasonal changes in infections, but we did find a significant effect of dolphin length on the abundance of T. forsteri and the prevalence of S. triangularis. Admittedly, the current sample also covers a large number of years and, therefore, inferences are more prone to confounding effects of time. However, there are also contrasting differences in the host length composition between samples that could readily explain the difference. Only 4 of 52 dolphins (7.7%) from the sample analysed by Mateu et al. (2014) could be considered sexually immature (<180 cm long, see Calzada et al., 1997), vs. 72 of 233 (30.9%) in the present study. Thus, it is likely that host length effects on parasite infections are now being detected because there is a suitable representation of small dolphins; among adult dolphins, infection parameters would vary individually regardless of host’s body length (Mateu et al., 2014).

The occurrence of S. triangularis was found to be negatively related with host length, whereas the abundance of T. forsteri peaked at ca. 140 cm, then decreased in larger dolphins. Interestingly, comparable trends appear to exist for the same parasites in other Stenella spp. Dailey and Perrin (1973) reported raw data on the abundance of T. forsteri and the occurrence of S. triangularis in a sample of pantropical spotted dolphins, S. attenuata from the eastern tropical Pacific. When we examined the relationship of both variables with host length (excluding calves), it was negative and significant, as expected (T. forsteri: Spearman correlation, r_s= -0.235, n= 75, one-tailed p< 0.021; S. triangularis: logistic regression, b= -3.561, 1 df, one tailed p= 0.025).

The factor(s) accounting for such trends are difficult to ascertain. In theory, larger (older) dolphins have more chances to get infected because they are exposed to infected prey for longer and/or consume more infected prey due to their higher metabolic demands (Mateu et al., 2014; Pool et al., 2020). Since this does not occur in the striped dolphins of our study, one could argue that smaller (younger) dolphins have a weaker immune system (thus they offer more opportunities for parasite establishment) and/or feed more on specific prey that serve as intermediate hosts for T. forsteri and S. triangularis. The latter possibility opens future avenues to pinpoint relevant prey taxa for the transmission of these cestodes.

At a spatial scale of ca. 800 km (from the highest to the lowest latitude), we did not detect obvious geographical differences in the intestinal helminth assemblages of striped dolphins, except for a modest positive effect of latitude on the abundance of T. forsteri. The dolphins used for this analysis were collected in a narrow time window (summer and autumn, 1990) that minimises confounding time effects, but these animals were ill and could have performed unusual movements or suffered post-mortem transport by currents prior to stranding. However, it seems unlikely that such movements occurred at the spatial scale of the study as we compared regions ca. 200 km long on average.

Differences in helminth infection levels between localities have been observed when cetacean populations are sedentary within the spatial scale covered by the surveys (e.g., Aznar et al., 1995; Mead and Potter, 1995). In contrast, it is likely that, in the study area, striped dolphins perform large-scale movements, perhaps facilitated by the lack of obvious physical or oceanographic barriers, and this would tend to homogenise infections across the defined regions. Currently, there are no precise data on movements of striped dolphins based on satellite tagging, but evidence obtained for other Stenella spp. indicate that individuals can perform movements of > 100 km/day along the continental slope (Scott and Chivers, 2009). Mediterranean striped dolphins appear to be also highly mobile; there are reports of diel movements covering dozens of km (Gannier, 1999), and seasonal changes in abundance that would suggest migrations at the scale of hundreds of kilometres (Gómez de Segura et al., 2006; Laran and Drouot-Dulau, 2007; Panigada et al., 2011; Arcangeli et al., 2017).

Interestingly, a parasitological survey of 14 striped dolphins that were collected on the coasts of the Ligurian Sea during the 1990 epizootic reported a similar composition and comparable prevalence of helminth fauna as that of the present study (Manfredi et al., 1992). Whether or not this homogeneity was caused also by extensive dispersal of striped dolphins between these two regions is difficult to assess in this case. There is evidence of significant genetic differentiation between dolphins from the Spanish vs. Ligurian populations (Gaspari et al., 2007), thus the parasitological similarity could have resulted just from ecological convergence.

Infections of Tri. globicephalae in striped dolphins were sporadic and scattered throughout several decades. This cestode species seems to be specific to members of the subfamily Globicephalinae and, in the western Mediterranean, it has been further detected in long-finned pilot whales Globicephala melas and Risso’s dolphins Grampus griseus ( Supplementary Tables S2 , S3 ). Our findings suggest that the striped dolphin is a secondary host for this cestode, its unpredictable occurrence perhaps resulting from unusual host-parasite contacts.

On the contrary, T. forsteri and S. triangularis are typical cestode species of striped dolphins ( Supplementary Tables S2 , S3 ). In the study area, their long-term infection trends were contrasting, with infection levels decreasing slightly (T. forsteri) or strongly (S. triangularis) up to year 2000 and subsequently recovering. To understand why such decrease occurred, and why each species was affected differently, we would need detailed knowledge on the life-cycle of both parasites. Unfortunately, no such information exists, although the general life-cycle for tetrabothriid cestodes is suggested to be complex, involving zooplanktonic organisms as first intermediate hosts, and fish or cephalopods as second intermediate hosts (Hoberg, 1987; Hoberg, 1996; Hoberg and Soudachanh, 2021). Thus, a potentially vast number of both biotic and abiotic factors could influence transmission dynamics of T. forsteri and S. triangularis, as evidenced for other TTPs infecting marine mammals (Des Clers and Wootten, 1990; Des Clers and Andersen, 1995; Stobo and Fowler, 2001).

It is worth noting, however, that three aspects of the patterns found in this study are compatible with the potential effects of a sudden significant reduction of the striped dolphin population in 1990. First, the dolphin mortality decreased the number of adult worms that successfully released infective stages, which could result in lower infections in dolphins in the following years (see MacDonald and Brisson, 2022). Second, the effect should be expected to be more pronounced in S. triangularis, at least for one reason, i.e., there seems to be a smaller number of other cetacean hosts supporting this cestode population. Out of the odontocete species analysed for parasites in the study area, S. triangularis has only been further detected in bottlenose dolphins, with low prevalence, whereas T. forsteri infects three other cetacean species, some with moderate prevalence ( Table S2 ). Finally, the increase in infection levels of T. forsteri and, particularly, S. triangularis after year 2000 would be not surprising if we consider that the striped dolphin population also experienced a fast population recovery in just a decade (Gómez de Segura et al., 2006). Note that some decoupling is expected between the recovery of the host and the parasite populations (Anderson and May, 1978). This is also congruent with the absence of a significant effect of the variable ‘Group (epizootic vs. non-epizootic)’ in our models, since the effect of the first morbillivirus outbreak seems to have extended over the next years, rather than being restricted to year 1990. In addition, this outbreak was far more severe and affected a much higher number of dolphins than the following ones in 2007 and 2011 (Raga et al., 2008; Rubio-Guerri et al., 2013), which would explain the absence of an observable effect of the other epizootic events in the abundance/prevalence of TTPs.

Long-term trends of the acanthocephalan B. capitatum (=B. vasculosum) are striking. This parasite was detected just once in striped dolphins before and after 1990, respectively, but in nearly half of those analysed in 1990 (34 of 74 dolphins), when infections were widespread across the study area. Moreover, as noted above, the striped dolphin acts as an accidental host for this parasite and likely acquired infections via prey shared with their ‘true’ cetacean hosts (see Hoberg et al., 1993; Costa et al., 2000; Gregori et al., 2012). Acanthocephalans identified as B. vasculosum have never been reported in other cetacean species from the study area, and B. capitatum only in the long-finned pilot whale, Globicephala melas, as adult (Raga and Balbuena, 1993; see also references in Table S2 ). The sperm whale, Physeter macrocephalus could also be a host for B. capitatum (Amin and Margolis, 1998) but has not been analysed for parasites in the study area. Therefore, these two odontocetes could be considered as the putative source of the infections observed in striped dolphins.

The reason(s) for such puzzling temporal pattern of B. capitatum (=B. vasculosum) in striped dolphin is(are) difficult to ascertain. First, we could hypothesize that, for some reason, the production of infective stages was exceptional in 1990 but, if so, the effects should last longer than a year in the trophic web. Species of Bolbosoma are relatively long-lived in paratenic hosts (Costa et al., 2000), similarly to other TTPs, for which a lifespan of several years is not uncommon (Kuhn et al., 2016). Second, we could postulate that, in 1990, striped dolphin fed on infected prey that were seldom exploited in other years. However, there is no evidence for this; although striped dolphins shifted diet from more oceanic to more neritic prey during the period 1990-2012, the change was gradual and did not entail the disappearance of any of the common fish prey taxa (Aznar et al., 2017). Third, we could argue that, in 1990, the dolphins immunocompromised by the DMV provided more favourable conditions for a longer permanence of B. capitatum in the intestine, thus significantly increasing their likelihood of detection. Yet, we did not find B. capitatum in dolphins likewise affected by the DMV in the 2007 or 2011 epizootics. Finally, we could speculate that migratory prey brought infections from elsewhere but, again, why this did occur only in 1990? In summary, the singular infection pattern of B. capitatum is hard to explain and was likely multi-causal in nature. Further monitoring of the intestinal helminth fauna of striped dolphins in the western Mediterranean could shed more light on this enigmatic parasite.