Introduction

Symbiosis is a ubiquitous phenomenon in nature (Herre et al. 1999). Though symbiotic relationships can be subdivided into commensalism, mutualism, and parasitism, it is often difficult to correctly determine which type of relationship they are in (Bronstein 1994). The difficulty in determinations stems from several reasons, including the difficulty in measuring the cost and benefit for both symbiont and host in the relationship. Furthermore, relationships can change over time or depending on the situation; once beneficial symbionts can become harmful, e.g. when numbers are too large (e.g., Cunning and Baker 2013; Wang et al. 2023), which also makes it difficult to accurately determine their relationship.

Cost and benefit for both symbiont and host substantially come from their various host-symbiont interactions such as providing food, shelter, dispersal (phoresis), pollination, mycorrhizal interaction, cleaning, and so on (Smith and Douglas 1987; Paracer and Ahmadjian 2000; Varma and Kharkwal 2009; Baeza 2015; White et al. 2017; Ollerton 2021), and these interactions often occur at the same time. The trophic relationship between symbionts and their hosts is one of the key criterions that determines the symbiotic relationship. For example, in a cleaning symbiosis, a cleaner symbiont is beneficial when feeding on host ectoparasites, but harmful when feeding on host tissue (Poulin and Vickery 1995). Thus, an accurate understanding of the feeding ecology of symbionts is necessary for the proper discrimination of some kinds of symbiont-host relationships, though measuring the trophic relationship of the symbiont to their host and vice versa is sometimes complex (Alfaya et al. 2015).

Sea urchins are known to host a wide diversity of symbionts from at least ten animal phyla (Hayes et al. 2016), including parasites, commensals, and mutuals. Some symbionts, such as eulimid gastropods and Echinoecus crabs, are considered to have a significant negative impact on their host urchins by feeding on the urchins’ tissue or drilling a hole in them (Castro 1971; Warén 1983; Dgebuadze et al. 2020). On the other hand, some other symbionts, such Mithrax crabs are considered to have a positive effect by removing parasites from the host urchins (Sonnenholzner et al. 2011). While some of these symbionts have been well studied, there are however still a number of symbionts whose ecology and effects on host urchins are poorly understood.

Amphipods of the genus Dactylopleustes Karaman and Barnard (1979) are known as symbionts of sea urchins and thought to have adapted to inhabit the surface of urchins (Bousfield and Hendrycks 1995). However, despite their strong symbiotic relationship, the ecology of Dactylopleustes is almost unknown (Vader 1978). A member of the genus, D. yoshimurai Tomikawa et al. 2004 was recently reported to aggregate on diseased areas of host urchins, while aggregations were not found on healthy urchins (Kodama et al. 2020). These observations suggested that the aggregated Dactylopleustes individuals fed on the diseased tissues of urchins, implying that they may use the urchin not just as a habitat but also as prey when the host urchin is diseased (Kodama et al. 2020). Moreover, these aggregations of Dactylopleustes on the diseased urchins are usually dominated by small juvenile individuals of Dactylopleustes while small juveniles of Dactylopleustes were rarely found associated with healthy urchins (Yamazaki unpublish). It is also suggested that large individuals and small juveniles might have different ecological niches including feeding behaviors. Accordingly, it is possible that the trophic relationship between Dactylopleustes and their host urchins vary depending on the urchins’ condition (healthy or diseased) and on the growth stage of Dactylopleustes.

It is however challenging to examine the gut contents of small amphipods like Dactylopleustes, making it difficult to determine their feeding ecology. In addition to the small size of prey organisms, amphipods intensively comminute their prey before ingestion, making it sometimes impossible to identify the prey species under microscopic examinations (Koester and Gergs 2017). In particular, indeterminate soft materials like urchin’s tissues are thought likely to be missed or can not be identified, and thus likely to be underestimated in microscopic examinations (Redd et al. 2008; Koester and Gergs 2017). Hence, elucidation of the trophic relationship between Dactylopleustes and its host urchins is challenging and has not been detailed so far.

DNA metabarcoding analysis has recently been applied to gut contents analysis to study the feeding ecology of various animals (e.g., Hirai et al. 2017; Kodama et al. 2017; Kobari et al. 2021). This methodology has an advantage in its independence to the morphological features of the prey. The DNA metabarcoding method allows us to detect a high taxonomic coverage of prey items including organisms that are undetectable by microscopic analysis (O’Rorke et al. 2012; Albaina et al. 2016). Thus, we expected this DNA metabarcoding method to provide new insights into the feeding ecology of Dactylopleustes as well as the trophic relationship between Dactylopleustes and their host urchins.

In this study, we examined the feeding ecology of the urchin symbiont Dactylopleustes yoshimurai by DNA metabarcoding analysis. The aim of this study was to test whether D. yoshimurai feed on their urchin hosts and if this trophic relationship is different with host disease status and developmental stage of D. yoshimurai. For this purpose, we compared prey composition (1) between D. yoshimurai individuals on diseased and healthy urchins, and (2) between their different development stages.

Materials and methods

Sample collections

Sample collections of D. yoshimurai were carried out in a rocky subtidal area at Akahama in Otsuchi Bay, Japan (Figs. 1: 39°21’00” N, 141°56’10” E, 2–3 m deep) from January to March 2021.

Fig. 1
figure 1

Map showing (A) Japan, (B) Otsuchi Bay and the study site

Full size image

Diseased and healthy individuals of short-spined urchins Strongylocentrotus intermedius were collected at the study site and gently placed in plastic bags to prevent escape of symbiotic organisms (diameter of diseased urchins, mean ± SD = 40.4 ± 5.5 mm, n = 18; diameter of healthy urchins, 38.7 ± 4.7 mm, n = 18). The urchin’s disease found in Otsuchi Bay appears to be a kind of bacterial diseases called ‘bald sea urchin disease’ (Johnson 1971; Shaw et al. 2023). However, it is impossible to identify the causes of disease in the field. In this study, therefore, a ‘diseased urchin’ was defined as an urchin with the following symptoms: the spines are observed to have fallen off in an area of the test, exposing the tissue, which is black and discolored. These symptoms are typical of bald urchin disease.

As reported in Kodama et al. (2020), D. yoshimurai were found to congregate in large numbers on diseased urchins, but only found alone or in small numbers on healthy urchins. The urchins and its symbionts were all anaesthetised with iso-osmotic 7.5% weight% magnesium chloride (MgCl2.6H2O). Then, D. yoshimurai detached from the urchins by anaesthesia were fixed and preserved in 80% ethanol. Dactylopleustes yoshimurai were detected from all the 18 diseased urchins (100%) and from 11 out of the 18 healthy urchins (61.1%).

Metabarcoding analysis of gut contents

A total of 41 amphipods of the following four groups were selected and their gut contents were used for DNA metabarcoding analysis: large individuals (body length > 4.0 mm) aggregated on diseased sea urchins (n = 10); small juveniles (body length < 2.5 mm) aggregated on diseased urchins (n = 14); large individuals collected from healthy urchins (n = 10); small juveniles collected from healthy urchins (n = 7). In our preliminary surveys, the individuals smaller than 2.5 mm in body length were always juveniles with no reproductive traits developed at all; individuals larger than 4.0 mm were always mature or sub-mature with developed (or developing) reproductive traits (Yamazaki unpublished).

The DNA metabarcoding analysis was carried out generally following Kobari et al. (2021) with minor modification. All the individuals of D. yoshimurai were transferred into glycerol and dissected under a stereomicroscope (SMZ1800, Nikon). The whole gut of each individual was transferred into a 1.5 ml tube containing 50 µl of 5% Chelex buffer (BIO-RAD) for the DNA extractions. These samples were homogenized with a pellet pestle and heated at 95 °C for 20 min. The samples were then centrifuged at 10,000 × g for 1 min. DNA concentrations in the supernatants were measured using a Qubit Assay Kit (Termo-Fisher Scientific) and a library was prepared using high-throughput sequencing based on the three-step PCR method. The 18 S rRNA V9 region was amplified using eukaryotic universal primers 1389 F and 1510R (Amaral-Zettler et al. 2009), and adaptor and dual-index sequences were attached during second and third PCRs for sequencing runs on an Illumina MiSeq. Each PCR sample was prepared in a 15 µL reaction volume using KOD Plus version 2 (Toyobo Inc.). The PCR protocols and settings followed Kobari et al. (2021). PCR products from the target region were confirmed by electrophoresis using a 2.0% agarose gel after first and third PCRs. Final PCR products were purified with a QIAquick PCR Purification Kit (Qiagen), and the concentrations of purified PCR products were measured with a Qubit Assay Kit. The quality of final PCR products was confrmed by using a Bioanalyzer (Agilent), and high throughput sequencing runs were performed using the MiSeq Reagent Kit v2 (Illumina) on an Illumina MiSeq to obtain 2 × 250 bp paired-end sequence reads (SRs). The raw data of sequence reads are deposited in the NCBI Sequence Read Archive (SRA) database (BioProject ID: PRJNA1132487).

Bioinformatics analysis

Bioinformatics analysis was carried out generally following Kobari et al. (2021). Raw SRs were quality-filtered using Trimmomatic (Bolger et al. 2014) and paired-end sequences were merged and further quality-filtered in mothur version 1.39.1 (Schloss et al. 2009). After sequence alignment against the SILVA 132 database (Quast et al. 2013), single-linkage pre-clustering (Huse et al. 2010), and chimera removal using UCHIME (Edgar et al. 2011) in mothur, taxonomic classifications were performed based on PR2 version 4.14.0 (Guillou et al. 2013) using a naïve Bayesian classifier with a threshold greater than 99%. As this study focused on eukaryotic organisms, only sequences classified as “Eukaryota” were selected. The final quality-filtered sequences were clustered into operational taxonomic units (OTUs). We used the 99% similarity threshold for high taxonomic resolution based on the average neighbor algorithm.

OTUs from freshwater or terrestrial organisms were considered contaminants, and thus were excluded from the dataset. The taxonomic group ‘Craniata’, which includes vertebrates, and ‘Fungus’ were also excluded to avoid sequence reads as human or fungal contaminants. To avoid possible OTUs for the host (i.e., Dactylopleustes), “Amphipoda unclassified”, “Malacostraca unclassified” or some other discernible OTUs were also excluded. The remaining OTUs were treated as the prey OTUs.

Data analysis

We examined whether the diet composition of D. yoshimurai differs among four groups (large/small Dactylopleustes individuals on diseased/healthy urchins). Diet composition was estimated by calculating the proportion of each prey OTU in each D. yoshimurai individual, meaning standardized by the total prey SR for each D. yoshimurai individual: proportion based on percentage of (number of SR of each prey taxon) / (total number of prey SR). Two-dimensional non-metric multidimensional scaling (nMDS) was used to visualize patterns of similarities in the diet composition among groups. Bray-Curtis similarity index was used for building the resemblance matrix. For more detailed comparison, the similarities in the diet composition were also visualized using nMDS in the following four pairs of groups: (1) large individuals on diseased urchins vs. those on healthy urchins, (2) small juveniles on diseased urchins vs. those on healthy urchin, (3) large individuals vs. small juveniles on diseased urchins, (4) large individuals vs. small juveniles on healthy urchins. In each pair of two groups, differences in the diet composition between two groups were assessed using permutational analysis of variance (PERMANOVA; Anderson 2001). Bray-Curtis similarity index was used for building the resemblance matrix. The p values were obtained using 9999 permutations. In the PERMANOVA and nMDS, the OTUs from each prey taxon were pooled at the order or higher taxonomic level to avoid different data resolutions for different prey taxonomic groups due to the bias in the database used. The PERMANOVA and nMDS were carried out using the functions ‘adonis2’ and ‘metaMDS’, respectively, in the ‘vegan’ package in R.

We also examined whether D. yoshimurai feed on urchins’ tissue at different proportions depending on the host disease status and developmental stage of D. yoshimurai themselves. The proportion of genus Strongylocentrotus, the host urchin, to the total prey SRs was compared between four groups (large/small Dactylopleustes individuals on diseased/healthy urchins) by pairwise Wilcoxon signed-rank test with Bonferroni adjustment. The pairwise Wilcoxon signed-rank test was carried out using the function ‘pairwise.wilcox.test’ in R.

Results

Prey sequence data

After a quality check of all sequence data as well as contamination removals, a total of 13,244 SRs were obtained for prey organisms in the gut content DNA. Data for all Dactylopleustes individuals obtained in this study are shown in Fig. 2A and those averaged per group are shown in Fig. 2B. At the individual level (Fig. 2A), it is shown that SRs of Maxillopoda and Echinoidea are most common in gut contents in all the individuals. Notably, the proportion of Echinoidea varied widely among individuals, from 0% in individuals with the the lowest values to over 75% in individuals with the highest values. Though OTUs were labelled at higher taxonomic levels, some OTUs were discernible at genus or species level. It should be noted that all the OTUs labelled as “Echinoidea” in this study were actually detected as “Strongylocentrotus” in the metabarcoding analysis. The results averaged per group showed small differences among groups (Fig. 2B). The class Maxillopoda was the most common in the prey taxa in all the four groups, followed by Echinoidea, Mollusca and/or Tunicata. Although differences were only small, Echinoidea was found to have a smaller proportion in large amphipod individuals on healthy urchins than in the other groups (Fig. 2B), and Tunicata was found to have a higher proportion in small amphipod juveniles on healthy urchins than in the other groups.

Fig. 2
figure 2

Proportions of prey sequence reads of Dactylopleustes yoshimurai. (A) Data for all individuals of D. yoshimurai obtained in this study, not averaged by group. (B) Data averaged per group

Full size image

A detailed result at the family level for the common prey taxon, the class Maxillopoda, is shown in Fig. 3. Families in the order Calanoida were found to be common; Cyclopoida and Harpacticoida were also detected but in smaller proportions. There was also a considerable proportion of “Maxillopoda unidentified”, which were undiscernible below the class level. Apart from the “Maxillopoda unidentified”, the most common taxon was Paracalanidae, followed by Calanidae, and Eucalanidae. These are all planktonic copepod families. Benthic or parasitic copepod families were rarely found from the prey SRs.

Fig. 3
figure 3

Proportions of sequence reads of common prey taxa, class Maxillopoda. OTUs are pooled at the family or higher level

Full size image

Difference in diet composition of Dactylopleustes yoshimurai in relation to the ontogenetic stages and host urchin’s disease

In the nMDS plots with all the four groups (Fig. 4), no clear separation among groups was found, which indicates that this species does not significantly change its diet depending on the host disease status or the developmental stage. More detailed comparisons between groups also show similar results (Fig. 5; Table 1). Except for the comparison between large individuals on diseased and healthy urchins (Fig. 5A), the other three comparisons did not show separations between groups in the nMDS plots, and the prey SRs also did not differ significantly between groups (Table 1; PERMANOVA, p > 0.05). Only in the comparison between large individuals of D. yoshimurai on diseased and healthy urchins, the nMDS plots were placed separately between groups (Fig. 5A). Furthermore, the composition of prey SRs was also significantly different between individuals on diseased and healthy urchins (Table 1; PERMANOVA, p = 0.014). In the nMDS of large individuals (Fig. 5A), the arrow of “Echinoidea” extends to the direction in which the plots of individuals on diseased urchins were placed, whereas the arrows of “Calanoida”, “Maxillopoda unidentified”, and “Rhizaria” extend to the direction in which the plots of individuals on healthy urchins were placed. This indicates that large Dactylopleustes individuals on diseased urchins feed more on urchins, and those on healthy urchins feed more on Calanoidea, Maxillopoda, and Rhizaria.

Fig. 4
figure 4

Two-dimensional non-metric multidimensional scaling plots with convex hulls (shaded areas) showing composition of prey sequence reads of Dactylopleustes yoshimurai. Each point represents an individual of D. yoshimurai. Open circles and solid triangles represent large individuals and small juveniles of D. yoshimurai, respectively. Black and red represents D. yoshimurai on healthy and diseased urchins, respectively. Arrows indicate vectors that represent the strength and direction of the correlation of each prey taxon with the nMDS space. Only taxa with significant loadings (p < 0.05) are indicated

Full size image
Fig. 5
figure 5

Two-dimensional non-metric multidimensional scaling plots with convex hulls (shaded areas) showing composition of prey sequence reads of Dactylopleustes yoshimurai: (A) large individuals on diseased urchins vs. those on healthy urchins, (B) small juveniles on diseased urchins vs. those on healthy urchin, (C) large individuals vs. small juveniles on diseased urchins, (D) large individuals vs. small juveniles on healthy urchins. The p value on upper right on each nMDS space was calculated by PERMANOVA based on the differences in the composition of prey sequence reads between each two groups. Each point represents an individual of D. yoshimurai. Arrows indicate vectors that represent the strength and direction of the correlation of each prey taxon with the nMDS space. Only taxa with significant loadings (p < 0.05) are indicated

Full size image
Table 1 Results of PERMANOVA testing differences in diet composition between Dactylopleustes individuals on diseased and healthy urchins
Full size table

The proportion of the host urchin Strongylocentrotus to the total prey SRs also showed different results between large individuals and small juveniles of D. yoshimurai (Fig. 6). In the large individuals, proportion of Strongylocentrotus was significantly greater in individuals on diseased urchins than those on healthy urchins (Fig. 6; Wilcoxon signed-rank test, Z = 2.97, p = 0.020). On the other hand, in the small juveniles, there was no significant difference in the proportion of Strongylocentrotus between individuals on diseased and healthy urchins (Fig. 6; Wilcoxon signed-rank test, Z = -0.843, p = 1.000).

Fig. 6
figure 6

Box plots with jitter plots showing proportion of sequence reads of the genus Strongylocentrotus, the host urchin, to the total sequence reads of all the prey taxa. Each box indicates the upper and lower quartiles. Black bold line in each box indicates the median. Whiskers indicate minimum and maximum values excluding outliers. Black dots represent all the raw data. Values outside two times the interquartile range above the upper quartile and below the lower quartile are treated as outliers. The p values were calculated by pairwise Wilcoxon signed-rank test with Bonferroni adjustment. The significant p value was shown in bold

Full size image

Discussion

Despite the strong symbiotic relationship between Dactylopleustes amphipods and their host urchin, the ecology of Dactylopleustes including their feeding behavior is almost unknown to date. The following two possibilities for their feeding behavior have been speculated based on the morphology of their mouth parts (Tomikawa et al. 2004): amphipods feed on (1) small organic food items around the urchin’s spines, (2) urchin test, such as pieces of skin or diseased surface tissue. Due to analytical difficulties, no detailed study on their feeding ecology has been carried out to date. By using the DNA metabarcoding method, the present study provides the first analysis of the feeding ecology of the genus Dactylopleustes. Our results generally showed that planktonic copepods were predominantly found from the gut contents of Dactylopleustes yoshimurai, regardless of their body size or whether the host urchin was diseased or not, suggesting that they do not have a strong trophic linkage with the host urchin regardless of circumstances. We herein discuss the feeding strategy of Dactylopleustes yoshimurai as well as the methodological considerations of the DNA metabarcoding analysis.

Methodological considerations

The DNA metabarcoding allows researchers even without taxonomic expertise to identify prey taxa with rapid and solid protocol. Furthermore, it can detect a high taxonomic coverage of prey items including organisms that are undetectable by microscopic analysis. Hence, it is highly expected to provide new insights into the feeding ecology of small organisms like Dactylopleustes. However, it should be noted that a few drawbacks of the methodology need to be considered in order to appropriately interpret the results.

First, the quantitative limitations of DNA metabarcoding method have been pointed out from several points of view (Deagle and Tollit 2007; Troedsson et al. 2009; Valentini et al. 2009). Recent studies concluded quantitative relationship between the biomass and SRs may not be strong (e.g. Lamb et al. 2019). Second, the copy numbers of 18 S rRNA genes can vary among taxonomic groups. For example, dinoflagellates have a relatively large number of copies of the 18 S rRNA genes (de Vargas et al. 2015), whereas appendicularians have a smaller number (Bucklin et al. 2018), and thus these variations may result in overestimation or underestimation of SRs for some taxa. However, since the overestimated dinoflagellates are still only less than 2.5% of total prey SRs on average, this overestimation does not seem to be a major problem, at least in the present study. Third, the DNA metabarcoding method cannot detect prey SRs from cannibalism and closely related taxa to Dactylopleustes. There are no adequately high-resolution gene databases on benthic amphipods including Dactylopleustes, and as a result, we detected a certain number of SRs that have been labeled as ‘amphipoda unclassified’. We could not confidently distinguish between contamination of the tissue of D. yoshimurai and predating other amphipod individuals (including cannibalism) in our genetic dataset. Thus, we conservatively treated these sequences as a contamination of the tissue of D. yoshimurai and disregarded them in our analysis, which may potentially result in underestimation of SRs from the order Amphipoda and its close taxa.

We acknowledge that our results contain a certain amount of bias and imprecision, as it is currently difficult to correct for these over- or underestimations for technical reasons. Nevertheless, the metabarcoding method is still a powerful tool to perform a comprehensive dietary analysis, especially for small organisms for which microscopic observations are practically impossible. The following discussion are based on the results obtained, but we would like to note that the biases discussed here should be considered in the interpretation.

Trophic relationship between Dactylopleustes yoshimurai and the host urchins

OTUs of planktonic copepods were found in a high proportion from the gut contents of Dactylopleustes yoshimurai, regardless of their body size or whether the host urchin was diseased or not. The most common taxon in the gut contents was the family Paracalanidae, which has been reported to occur abundantly in the water column of Otsuchi Bay (Nishibe et al. 2016). Most of the other copepod taxa detected in the gut contents have also been reported to occur in the water column of Otsuchi Bay, such as Calanidae, Clausocalanidae, Corycaeidae, Eucaranidae and Oithonidae (Nishibe et al. 2016). It is not possible to distinguish whether the SRs of copepods in the gut contents were derived from direct feeding on copepods in the water column or feeding on copepod fecal pellets. Morphologically, this species does not appear to be a suspension feeder, but a deposit feeder. Suspension-feeding amphipods generally have heavily setose antennae, gnathopods, and mouthparts that are adapted to collecting suspended particles (Caine 1974, 1977; Dixon and Moore 1997; Poltermann 2001) or have an ecology that they build a tube-like nest that are adapted to collecting suspended particles (Dixon and Moore 1997; Moore and Eastman 2015). As members of the genus Dactylopleustes do not have suspension-feeding appendages or tubes, they are unlikely to be able to feed directly on copepod individuals from the water column as a main food source. One possible interpretation is that the planktonic copepod OTUs detected in their gut contents were not derived from copepod individuals, but from sinking particles deposited on the sediments like fecal pellets or exuviae of copepods. A previous study suggests that gut content DNA from copepods resulted from coprophagy on copepod fecal pellets (e.g., Kobari et al. 2021). It is also well known that a large amount of copepod fecal pellets is produced in the water column (Turner 2002, 2015). Although much of them are reprocessed in the water column by microbial decomposition and coprophagy, they are known to be partly sedimented (Turner 2002, 2015). Indeed, copepod fecal pellets sampled from 50 m depth in neighboring waters of the study site, were identified with a molecular method (Suzuki et al. 2000). In addition, D. yoshimurai were generally observed attaching on the oral side rather than the dorsal side of the host urchin, which is a suitable position for feeding on the sediment. Not only OTUs of specific organisms such as urchins and copepods, but also OTUs of various organisms were detected from the gut contents, which would also support that the species is a sediment feeder. Accordingly, we herein propose a hypothesis about the feeding ecology of this species; this species mainly attach to the oral side of the host urchin and feed on the sediment under the host urchin where copepod fecal pellets and exuviae are deposited in a high density.

Another important point to discuss is whether the species feeds on the host urchin tissue. Firstly, it should be noted the impossibility to differentiate whether the amphipods feed on living urchin tissue, dead urchin tissue, or deposited urchin faeces. Though a few individuals showed a high proportion of urchins up to about 80% of the total prey SRs, on average the SR of urchins only accounted for less than 20% of the total prey SRs. We therefore conclude that this species does not essentially feed on their host urchin tissue as a main food source. Further metabarcoding analyses for sediments may be necessary in the future study to examine the possibility of the species feeding on faeces in sediments.

Effect of the disease status of host urchin and developmental stage of Dactylopleustes yoshimurai on its feeding ecology

In large individuals of D. yoshimurai, their feeding ecology differed between those on diseased and healthy urchins. Large individuals on healthy urchins almost did not feed on urchins, whereas those on diseased urchins fed more on urchins. In combination with the previous observation that this species aggregated on diseased areas of host urchins (Kodama et al. 2020), our results suggest the following feeding strategies of the species in response to diseased host urchin: when the host urchin is healthy, they are feeding only on sediment including copepod fecal pellets, but when diseased urchins appear, they aggregate at the diseased area and feed more on urchin tissues. Small juveniles were also observed at a high abundance amongst individuals aggregating on the diseased area (Fig. 7). This would suggest that the aggregating large individuals may be reproductively active and nourished, and thus potentially imply that diseased urchins may be a suitable breeding habitat for the species.

Fig. 7
figure 7

A–C1, Dactylopleustes yoshimurai aggregated on the diseased area of their host urchin, including a number of small juveniles; C2, magnified photograph for C1. Body colour of small juveniles is pale purplish white or translucent

Full size image

In small juveniles, the dietary composition did not differ between those on diseased and healthy urchins contrary to our expectations. We had anticipated small juveniles on diseased urchins would feed to a greater amount on the urchin’s tissue than those on healthy urchins, since they were found to be highly abundant on diseased urchins but seldom found on healthy sea urchins. A possible explanation for the result is that healthy urchins are not an appropriate habitat for small juveniles of this species. This species might have a life history of breeding on diseased urchins as discussed above, small juveniles basically live on diseased urchin and then occasionally disperse to surrounding urchins. It is possible that the small juveniles on healthy urchins analyzed in this study may have migrated from diseased sea urchins just before collection.

These results also indicate that their feeding strategies differ between large individuals and small juveniles. Large individuals change their feeding behavior in response to disease of the host, whereas small juveniles do not change their feeding behavior regardless of host disease. Furthermore, it is also suggested that in large individuals that the symbiotic relationship may change from commensalism to parasitism depending on the condition of the host urchin. Symbiotic relationships are often context-dependent (Daskin and Alford 2012; Bedgood et al. 2020), and known to change among mutualism, commensalism, and parasitism depending on circumstances (Lee et al. 2009; Stoll et al. 2013). However, in our results the main food source of D. yoshimurai still remains as copepods regardless of their body size and also the host urchin disease status, implying that the host disease has only a small effect on the feeding behavior of D. yoshimurai and on their symbiotic relationship.

Ecological interactions between Dactylopleustes yoshimurai and the host urchins

The effects of their feeding activities on the host urchin were not detailed in this study but thought to be limited, as they feed only on a small amount of urchin. However, even if the impact of a single Dactylopleustes individual is small, the entire Dactylopleustes population on an urchin may have a strong impact on the urchin, as large numbers of individuals often congregate on diseased urchins. It is possible that their feeding activities cause damage and negative effects on the urchin. Besides, it is also still possible that their feeding activities remove the diseased tissue and thus have potential positive effects, or that they just feed on tissues that have naturally detached from the diseased area and not have any particular effect.

It is also still not clear why this species aggregates on diseased urchins (Kodama et al. 2020). We had thought that they aggregated in order to feed more on diseased urchin tissues, however, our results are contrary to this hypothesis since they feed on urchins in only a small amount. Another hypothesis herein arisen is that the quality of the diseased urchin tissue, rather than quantity, may be important in their diet. For example, some fatty acids were implicated as being important in crustacean reproductive physiology and the subsequent viability of juveniles (Alava et al. 1993; Xu et al. 1994; Glencross 2009; Titus et al. 2019). It is possible that nutrients that are required for egg production and juvenile growth (e.g., fatty acids) may be less available from the sediment and more available from the urchins’ tissue, meaning that they might need to eat urchin tissues (if only a little) to reproduce. Further research about their nutritional requirements are needed to examine this hypothesis.

This study partially points to the possibility that diseased urchins are not so important as food for Dactylopleustes. Therefore, it suggests that diseased urchins may be utilized by Dactylopleustes in other ways than as food. For example, diseased surfaces may act as a refuge for Dactylopleustes from urchin attacks. Symbionts of urchins are known to be eliminated and sometimes suffered fatal damages by the pedicellariae that urchins usually use for defense (Campbell 1973; Campbell and Rainbow 1977). It is possible that the effect of healthy urchins’ defenses against foreign bodies (e.g. pedicellariae) is excluding small amphipods from thriving on them, which may explain their greater abundance on diseased urchins which may have less or less active defenses. Though it is currently unclear if they are only found on urchins or also in other environments, their specialized pereopods, modified antennae, and some other morphological features indicate that members of Dactylopleustes are strongly adapted to life on urchins (Bousfield and Hendrycks 1995; Tomikawa et al. 2004), and they are sometimes considered to be obligate symbionts of urchins (e.g. Bousfield and Hendrycks 1995). It is at least likely that the surface of sea urchins is their primary habitat, and thus such avoidance of host urchin attacks may be important for their symbiotic life history.

Conclusion

As host-symbiont interactions, and hence their symbiotic relationships, can easily change depending on the condition of the host or symbiont, it is important to examine the plasticity or stability of symbiotic relationship for a better understanding on their ecology. Our study demonstrates that D. yoshimurai does not have a strong trophic linkage with the host urchin, regardless of their body size or whether the host urchin is diseased or not, and thus their impact on the host urchin would be limited. Though further research is needed, these results generally support that D. yoshimurai is at least not a strong parasite but more of a commensal of urchins. To date, our knowledge on the feeding ecology of Dactylopleustes has been severely limited because of the difficulty of traditional microscopic examination methods. We herein demonstrated the effectiveness of molecular-based approach in evaluating the feeding ecology of small crustaceans like Dactylopleustes as well as the trophic linkage between small symbionts and their hosts. This kind of approach can provide further understanding of the feeding ecology of Dactylopleustes amphipods as well as many other small symbiotic crustaceans of which the feeding ecology is still largely unknown.