Sophistication in a seemingly simple creature: a review of wild holothurian nutrition in marine ecosystems

ABSTRACT Holothurians are marine invertebrates that are among the most widespread benthic megafauna communities by both biomass and abundance in shallow‐water and deep‐sea ecosystems, their functions supporting important ecological services worldwide. Despite their simple appearance as sea cucumbers, holothurians show a wide range of feeding practices. However, information on what and how these animals eat is scattered and potentially confusing. We provide a comprehensive review of holothurian nutrition in coastal and deep‐sea ecosystems. First, we describe morphological aspects of holothurian feeding and the ultrastructure of tentacles. We discuss the two processes for food capture, concluding that mucus adhesion is likely the main method; two mucous cells, type‐1 and type‐2, possibly allow the adhesion and de‐adhesion, respectively, of food particles. Secondly, this review aims to clarify behavioural aspects of holothurian suspension‐ and deposit‐feeding. We discuss the daily feeding cycle, and selective feeding strategies. We conclude that there is selectivity for fine and organically rich particles, and that feeding through the cloaca is also a route for nutrient absorption. Third, we provide a wide description of the diet of holothurians, which can be split into two categories: living and non‐living material. We suggest that Synallactida, Molpadida, Persiculida, Holothuriida and Elasipodida, ingest the same fractions, and emphasise the importance of bacteria in the diet of holothurians.


I. INTRODUCTION
Dating back 460 million years, holothurians (commonly called sea cucumbers) are ubiquitous marine echinoderms belonging to the class Holothuroidea. An initial classification was established by Pearson (1914), who attempted to organise the species, under the name Holothuria L., into groups based on their gross morphology. Pawson & Fell (1965) subsequently proposed a classification based on tentacle, body, and calcareous ring morphology, defining five orders: Dendrochirotida, Apodida, Molpadida, Elasipodida, and Aspidochirotida. Until recently, this was the most widely used classification for holothurians. When the first broad-scale molecular phylogenetic analyses of Holothuroidea were completed, these deeply altered the previous classification through the suppression of the order Aspidochirotida, which was revealed to be polyphyletic. Species previously under Aspidochirotida have since been placed into three new orders: Synallactida, Persiculida (in part), and Holothuriida (Miller et al., 2017). There are more than 1,752 accepted holothurian species (WoRMS, 2020), with new species being described each year.
Some species of holothurians are considered as luxury food (bêche-de-mer, trepang or hai-som), medicines, and aphrodisiacs in many Asian countries (Conand, 1990;Lovatelli et al., 2004;Shiell & Uthicke, 2005; Toral-Granda, Lovatelli & Vasconcellos, 2008). More than 70 species of holothurians are commonly harvested (Purcell et al., 2016), predominantly from the Indo-Pacific region (Kinch et al., 2008;Conand, 2018). In some locations, populations of highly commercially valuable species have been decimated to a point that fishing regulations and regulatory measures alone may be insufficient to restore populations (Friedman et al., 2011). This overexploitation is linked to a shift from traditional to semi-industrial fisheries (Conand, 2001). Countries of the Indian Ocean, West Pacific and Latin America have active fisheries (Conand, 2018), and South-East Asia is considered the main world market (Rahman & Yusoff, 2017). The world fishery of holothurians quadrupled between 1955 and 2012 to satisfy the increasing Asian market for 'bêche-de-mer' (Rahman, Yusoff & Arshad, 2015). Consequently, many countries in the Indo-Pacific have prioritised sea cucumber aquaculture in their development plans (Jimmy, Pickering & Hair, 2012). In addition, restocking, sea ranching, and sea farming have been described as potential alternatives to reduce pressure on wild holothurian populations and their worldwide overexploitation. Google Scholar using the key words 'sea cucumber' for general publications and 'bêche-de-mer' for those focusing on fisheries and aquaculture, shows that the number of articles on both topics has increased considerably since the 1990s (Fig. 1).
The increasing literature provides scattered data about holothurian feeding, making it difficult to obtain a broad view on this topic. This review intends to assess our understanding of holothurian feeding. The nutritional aspects considered herein include feeding (the act of consuming food), digestion and absorption. To our knowledge, only one review has been published previously on the nutrition of holothurians (edited by Jangoux & Lawrence, 1982), about 40 years ago. There is no comprehensive review covering the nutrition traits of holothurians in relation to their ecology, and collating information on their feeding patterns. While much information exists on the feeding of farmed holothurians, this relates only to controlled conditions. Thus, our aim herein is to analyse holistically the nutrition of wild holothurians.

II. MORPHOLOGICAL AND PHYSIOLOGICAL ASPECTS
(1) Tentacle structure and movements during ingestion The number, size, and structure of tentacles varies among holothurian orders. Based on their structure, Massin (1982) defined five types of tentacles: dendritic, peltate, pinnate, digitate, and peltodendritic. The first four of these are common and widely used in the description of tentacle morphology of holothurians. Species within Dendrochirotida possess dendritic tentacles whereas those of Molpadida possess digitate tentacles (Levin, 1989). Peltate tentacles, described as "cauliflower-like structures" (Bouland, Massin & Jangoux, 1982, p. 134) or "a nasturtium leaf with a central short stalk giving off horizontal branches" (Hyman, 1955, cf. Cameron & Fankboner, 1984, are found within Elasipodida, Synallactida, Persiculida, and Holothuriida. In some cases, the tentacle structure shows variations among species within the same order, such as for Apodida species, which are found in coastal and deep-sea ecosystems and possess pinnate or digitate tentacles. Combinations of these four main types of tentacle can be found in the literature: (i) a 'peltodendritic tentacle' which is, according to Massin (1982), a combination of the shaft of a peltate tentacle and distal end of a dendritic tentacle; (ii) a 'peltatodigitate tentacle', which is a combination of the shaft of a peltate tentacle and distal end of a digitate tentacle (Miller et al., 2017). Peltodentritic tentacles are uncommon, while peltatodigitate tentacles are found in Apodida (Miller et al., 2017).
Tentacle morphology may also differ between early and adult stages of the same species (Cameron & Fankboner, 1984). Generally, a tentacle possesses several shafts that end with one or more discs, composed of numerous apical papillae (Fig. 3A, B). The apex of each papilla is characterised by the presence of buds ( Fig. 3C) with various cell types. The main difference in tentacle structure is the degree of branching of the main tentacle stalk (Fig. 2), from an unbranched (digitate form), to a slightly branched (peltate form), to a highly branched (pinnate form) to an ultrabranched (dendritic form) structure. These differences reveal an adaptive radiation facilitating habitat and feeding specialisations (Sokolova, 1958;Hansen, 1975). From an ecological perspective, differences in tentacular morphology among sympatric holothurian species with overlapping bathymetric ranges may allow resource partitioning to avoid competitive interactions (Roberts & Moore, 1997).
A comprehensive review by Massin (1982) described the feeding mechanisms of deposit-feeding holothurians. When the animal initiates feeding, the tentacles expand into the water column, and are pressed onto or into the sediment surface where they perform investigatory movements (Bouland et al., 1982). The tentacular movement is thought to be driven by the hydrostatic pressure of the ambulacral fluid of the water-vascular system and the mesothelial muscles (Bouland et al., 1982). Tentacles can also cooperate to capture large debris such as fragments of seagrass or macroalgae. In this case, when the desired food elements are located, the extremities of the tentacle retract first, thus entrapping the particles between the buds (Fankboner, 1978;Levin, 1989). Sokolova (1958) described how tentacle morphology can influence the collection of particulate food in deep-sea holothurian species; well-developed marginal processes on discs are capable of picking up single food particles from the sediment, whereas species with undifferentiated discs indiscriminately ingest the upper layer of the sediment. After collection, the tentacle continues to contract, bending towards the mouth and penetrating it. At this ingestion stage, mouth size is the limiting factor (Myers, 1977), although a proportion of captured particles is lost during transport to the mouth (Powell, 1977;Levin, 1989). As the tentacle penetrates the pharyngeal cavity, particle removal is facilitated by wiping Fig 1. Cumulative number of articles on "sea cucumber" for general publications and "bêche-de-mer" for those focusing on fisheries and aquaculture since 1950. Note the different scales for the two axes. Sophistication in a seemingly simple creature the tentacle against the bulging pharynx wall (Cameron & Fankboner, 1984). Finally, the tentacle withdraws from the mouth and extends again to continue investigating for food sources.
(2) Food-capture mechanism Understanding the mechanism(s) of capture of food particles in holothurians was approached via studies in functional morphology integrating data concerning the ultrastructure of the tentacles to determine the composition of the tentacle parts in contact with the ingested sediment. These studies, mainly carried out in the 1980s and 1990s, are unfortunately few in number. Sixteen species were investigated, mainly Dendrochirotida (13 species ;Smith, 1983;McKenzie, 1987), two former Aspirochirotida (Bouland et al., 1982;Cameron & Fankboner, 1984) and one Apodida (Flammang & Conand, 2004). Roberts & Moore (1997) and Fankboner (1981) used scanning electron microscopy to study the fine external structure of tentacles of four species of Elasipodida and five species of Dendrochirotida without detailing their cellular composition. Based on these studies and behavioural experiments, several authors have proposed mechanisms for the capture of food particles in several species (Roberts, 1979;Hammond, 1982;Bouland et al., 1982;Cameron & Fankboner, 1984). Holothurians appear to use two methods for food collection: food particle ensnarement (a mechanical process) and food particle adhesion (a chemical process). Food particle ensnarement was proposed in early studies about the functioning of the tentacles, since when adhesion has gained support as the principal mechanism involved in the capture of food particles. (a) Food particle adhesion Several authors described a mucus-like secretion on the oral tentacles and suggested a primary role in holothurian food capture (Bouland et al., 1982). Roberts & Bryce (1982) described mucus-secretory cells in the tentacular epidermis of several tropical species and stated that the adhesive material would play a role in collecting food particles. Levin (1982), studying Apostichopus japonicus (Selenka), presumed that adhesion was a primary function of the tentacle during feeding. Similarly, Hammond (1982) observed mucus coating the tentacular surface and stated that adhesion is a significant factor in food collection for Synallactida, Persiculida, and Holothuriida, a statement reiterated by other  authors working on temperate species (Smith, 1983;Costelloe & Keegan, 1984;Holtz & MacDonald, 2009).
Transmission electron microscopy studies have shown buds to be the sites of secretion. The bud epidermis can include six types of cells in the species investigated to date, with 3-5 of these cell types usually being present (Table 1). The functions of these cells have been deduced by drawing parallels with the roles of cells in the adhesive disc of podia of echinoderms. (i) 'Support cells' have been described in some Dendrochirotida (McKenzie, 1987) and in the only species of Apodida investigated (Flammang & Conand, 2004) (Table 1). (ii) A 'vesicular cell' type has only been described in Apodida (Flammang & Conand, 2004) with unknown functions. (iii) Ciliated cells, also named 'uniciliated cells', 'ciliated cells' or 'uniciliated sensory cells' have been observed in all species investigated (Table 1). Two roles have been suggested for this cell type: they could mechanically disengage particles (Fankboner, 1978) or they could be sensory (Flammang & Conand, 2004). Bouland et al. (1982) associated the bud structure of Holothuria (Panningothuria) forskali Delle Chiaje with Laverack's (1974) description of chemosensory organs in marine invertebrates, where cilia are proposed to function as olfactory receptors, while microvilli are gustatory. Cilia are generally short and non-motile (Dorsett & Hyde, 1969;Schulte & Riehl, 1976), like those depicted by Bouland et al. (1982). The apical elements of the tentacles are the first structures to contact substrates while the tentacles actively forage for food, strongly suggesting that the buds are likely to be chemosensory (Bouland et al., 1982). The cells of the epidermis of each bud are adjacent to the epineural nervous plate, which is connected to the hyponeural nerve plexus of the tentacles (Fig. 3E, F). Mesothelial muscles are present in each buccal tentacle ( Fig. 3E; Bouland et al., 1982;McKenzie, 1987). (iv) 'Mucous cells', also called 'mucocytes' were observed in the tentacles of Dendrochirotida, Holothuriida and Apodida (Table 1). As similar cells are also observed outside the tentacles, their mucus-secreting role is probably to ensure the presence of a physical barrier on the epidermis. (v) Type-1 and (vi) type-2 secretory cells are thought to play a direct role in the capture of food particles. Type-1 secretory cells, also referred to as 'granular cells', 'glandular vesicular cells' or 'papillate cells' (Table 1) are characterised by numerous dense-cored vesicles of 200-700 nm, with their diameter varying among species. These vesicles are found in many microvilli and are thought to be secreted into the cuticle. They may be homologues of the adhesive cells found in the podia of echinoderms. Type-2 secretory cells, also named 'type-2 neurosecretory cells' or 'granular cells', have been observed in Dendrochirotida and Apodida (Table 1). This cell type possesses numerous densecored vesicles of 60-130 nm, again with their diameter varying among species. These vesicles are found in the cell apex and also are thought to be secreted into the cuticle. They are thought to be homologues of de-adhesive cells found in the podia of echinoderms. Thus, the chemical capture of food particles could be performed by the secretion of type-1 secretory cells and, when the tentacles are placed in the oral cavity, adhesion could be removed by the secretion of type-2 secretory cells.
This hypothesis is currently only theoretical and future functional morphology studies are needed, together with transcriptomic data to characterise the proteins expressed at the tentacle buds.

(b) Food particle ensnarement
Authors have long debated the mechanisms of food capture by which particles are trapped within bud interstices. Roberts (1979) suggested that expansion of the peltate tentacles caused inter-bud spaces to open on the tentacles, which could then mechanically trap particles when the tentacles retract. Cameron & Fankboner (1984) reported mechanical ensnarement to be of relatively little importance for Parastichopus californicus [now Apostichopus californicus (Stimpson)] since there is no reverse process of tentacle expansion/relaxation while the tentacle is in the pharyngeal cavity. Food detection is likely to be related to the ciliated cells as described in Section II.2, as the tentacles spread onto the substratum or in the water column.
The foregut includes the pharyngeal bulb, oesophagus, and the first descending loop of the intestine. The oesophagus is divided into two parts of equivalent lengths, both ending in a sphincter (sphincters 1 and 2) (Fig. 4B). The midgut begins immediately beyond the first intestinal loop where the proximal tubules of the rete mirabile are attached and consists of the ascending anterior loop of the intestine. A third sphincter separates the midgut from the hindgut. The hindgut, which represents approximately 70% of the total length of the gut, consists of the final posterior descending loop of the intestine and the cloaca from which respiratory trees emanate (Dolmatov & Ginanova, 2009 anal end of the animal where it loops back. The dorsal haemal plexus is a complex of pulsating channels attached to the dorsal midline of the foregut. Longitudinally along the ventral side of this region runs the ventral sinus. The rete mirabile is a complex of tiny tubules, entangled within the respiratory tree, and attached to the ascending loop of the midgut. These tubules terminate into the main channel of the dorsal haemal plexus. Although the foregut, midgut, and hindgut have three different functions (accumulation, digestion, and assimilation), most species of Synallactida, Persiculida, and Holothuriida are able to assimilate food from these three parts (Massin, 1978).

(b) Cell and enzymatic activities
The epithelia of the pharynx, oesophagus, and cloaca all contain T-shaped cells that are involved in the uptake of dissolved organic matter. The gut epithelium is composed of enterocytes with a filamentous coat for intracellular digestion (Féral & Massin, 1982). A broad range of hydrolytic gut enzymes has evolved in association with the diet Table 1. Ultrastructural composition of the tentacle buds of holothuroids with the nomenclatural terms used in analyses by various authors. The first column describes the corresponding cells observed in podia of echinoderms where a duo glandular adhesive system (adhesion and de-adhesion) has been identified Authors Fankboner ( Sophistication in a seemingly simple creature (Féral, 1989). Hydrolytic enzyme activity has been reported within the digestive tract, on cell membranes (Féral, 1989), and inside cells (Lawrence, 1982). Roberts et al. (2000) reviewed the enzymes found in the gut of deposit-feeding holothurians (shallow-water species in Holothuriida, Synallactida, Apodida and Dendrochirotida, and deep-sea species of Holothuriida, Persiculida, Synallactida and Elasipodida), finding broad similarities in their hydrolytic enzymes. Gut enzymes include: (i) esterases that hydrolyse short-chain fatty acids, with strong activity throughout the gut (Fish, 1967;Féral, 1989), (ii) lipases that hydrolyse long-chain fatty acids, with lowest activity (Clifford et al., 1982;Féral, 1989;Manship, 1995), (iii) proteases, (iv) peptidases, which hydrolyse peptide bonds and may be important in the initial stages of protein digestion (Massin, 1984;Féral, 1989;Roberts et al., 2000), (v) saccharidases, and (vi) phosphatases, present at high concentrations in the gut tissue (Féral, 1989;Boetius & Felbeck, 1995). Trefz (1958) observed phagocytic activity of abundant cells in the holothurian gut, such as round phagocytic amoebocyte cells that ingest and digest bacteria (Bacillus subtilis), but not indigestible particles.
(c) Obtaining nutrients through cloacal water retention The major functions of cloacal ventilation are commonly described as respiration, excretion, and salt balance. Holothurians with respiratory trees (absent in Apodida and Elasipodida) have the ability to pump large volumes of water into and out of the cloacal opening. A similar pumping mechanism is reported for echinoderms such as holothuroids (Newell & Courtney, 1965;Brown & Shick, 1979) and edrioasteroids (Bell, 1977), annelids (Wolcott, 1981), and crustaceans (Fox, 1952).
Innovative work, using radiography and stable isotope enrichment, demonstrated that the epithelium of the respiratory tree can assimilate dissolved organic matter such as monosaccharides and amino acids from the water column during cloacal water retention (Fontaine & Chia, 1968;Brothers, Lee & Nestler, 2015). The respiratory tree of Cucumaria lactea [now Ocnus lacteus (Forbes & Goodsir)] can assimilate dissolved 3 H-labelled glycine and 3 H-labelled glucose (Fontaine & Chia, 1968). The respiratory tree of A. californicus assimilated 15 N-labelled amino acids and peptides (Brothers et al., 2015) and 14 C-labelled unicellular algae, suggesting the transfer of nutrients from the respiratory tree into the haemal system (Jaeckle & Strathmann, 2013). Thus, is possible that holothurians can obtain nutrients through their respiratory trees as a result of cloacal pumping, but additional studies are needed to clarify its importance to these animals.
(1) Deposit-or suspension-feeding The first larval stage in holothurians, the auricularia, feeds on phytoplankton (planktotrophic) by means of cilia located on the epidermis. The second larval stage, the doliolaria, is non-feeding, and the final stage, the pentactula, develops the tentacles that are present in juveniles and adults. The first and second stages can be absent in holothurians with direct development (McEdward & Miner, 2001;Raff & Byrne, 2006). In general, juveniles and adults are particlefeeders, which usually exhibit either suspension-(in Dendrochirotida) or deposit-feeding (most other taxa) behaviour (Roberts et al., 2000;Pawson, 1970;Bakus, 1973). Suspension-feeding holothurians tend to live in high-energy environments to allow them to intercept food particles (Taghon & Jumars, 1984). Deposit-feeding behaviour can be subdivided into three groups related to feeding depth (Roberts et al., 2000). (i) Epibenthic deposit-feeders feed at the interface between the water column and sediment, or on the surface of seagrasses (Cuvillier, 2016), the sand scattered over the tegument of other specimens (e.g. H. atra; J.P., personal observations), boulders and corals (Roberts et al., 2000; J.P., personal observations), or sponges (Hammond & Wilkinson, 1985). (ii) Subsurface depositfeeders feed below the sediment surface. (iii) Funnel deposit-feeders create funnel-shaped depressions in sediments and feed on particles trapped in these funnels (Jumars, 1993).
Feeding mode may vary at the individual or population scale. According to Cadée (1984), many benthic individuals are opportunistic in their feeding mode. Some individuals of species in Dendrochirotida (Roberts et al., 2000), Synallactida, Persiculida, and Holothuriida (Da Silva, Cameron & Fankboner, 1986) can shift between deposit-feeding and suspension-feeding. This may be possible if the individual is positioned in a negative geotropic orientation (Da Silva et al., 1986). Leptopentacta elongata (Düben & Koren) and other shallow-water species (Fankboner, 1981;Levin, 1989) are known to shift opportunistically from suspension-to deposit-feeding when the concentration of suspended particulate organic matter is low. This opportunistic behaviour is driven by environmental factors such as food pulses in shallow-water (tidal or seasonal fluctuations) (Cushing, 1959) and deep-sea ecosystems (Jumars, Self & Nowell, 1982;Billett et al., 1983).
At the population scale, the feeding modes of some holothurian species can vary depending on their geographic location. Holothuria (Thymiosycia) arenicola Semper is an epibenthic feeder, feeding on surface layers of sand under coral debris in the Indo-West Pacific region and Cuba (Levin, 1989), whereas in the Bahamas, it feeds on particles in sediments (Mosher, 1980) as a subsurface deposit-feeder. The underlying reasons for these intraspecific differences at a population scale, and their drivers, are poorly studied.
(2) Daily burrowing cycle A study on the burrowing behaviour of shallow-water holothurians in Palao Island classified holothurians into two groups according to their feeding habits (Yamanouchi, 1939): (i) species that do not burrow into sediments or other substrates and feed continuously, such as Holothuria (Halodeima) edulis Lesson, H. (Semperothuria) flavomaculata Semper or H. atra (Yamanouchi, 1939;Trefz, 1958;Uthicke, 1994) and (ii) species that show a daily cycle of burrowing and feeding (Yamanouchi, 1939(Yamanouchi, , 1956Mercier et al., 1999;Lavitra et al., 2009). The factors regulating burrowing and feeding cycles appear to be complex (Yamanouchi, 1939(Yamanouchi, , 1956, with differences among species increasing the complexity of cues that drive such behaviour and impeding generalisation. According to Yamanouchi (1956), most adults prefer to feed during the day, particularly in the early afternoon when the water temperature is highest (Mercier et al., 1999) and productivity of the marine ecosystems is maximal (Heil et al., 2004). Some species, such as Stichopus chloronotus Brandt, move under corals from 00:00 to 10:00 and then return to sediments to feed during the day (Yamanouchi, 1956). Similarly, Stichopus variegatus (now Stichopus herrmanni Semper) hides under seagrasses between 20:00 and 10:00 (Yamanouchi, 1956). Coulon & Jangoux (1993) argued that it may be more energetically advantageous for juveniles of H. tubulosa to stop feeding during the coldest hours of the day. Individuals exposed to abnormal water temperatures exhibit unusual behaviour and feeding activity (Kato & Hirata, 1990). The temperate species A. japonicus aestivates during the warmest months when water temperatures are between 20 and 24.5 C (Choe, 1963), especially large and mature individuals . Aestivating individuals hide under structures and enter a state of dormancy Yang et al., 2005) in which feeding activity ceases (Yuan et al., 2007) until water temperature becomes more favourable. During aestivation, A. japonicus individuals undergo a series of physiological and morphological changes (Wang et al., 2008), losing 30-50% of their body mass . The digestive tract degenerates to half its preaestivation size Liu et al., 1996). Physiological responses associated with digestion are reduced during aestivation, with lower activity of digestive enzymes in the principal portions of the gut (Cui, Dong & Lu, 2000). Similarly, under high temperatures (austral summer), Holothuria (Metriatyla) scabra Jaeger, changes its burrowing behaviour, remaining on the sediment surface and feeding frequently, whereas at low water temperatures (austral winter) it remains beneath the sediment surface for most of the day (Mercier, Battaglene & Hamel, 2000). Thus, seasonal patterns will play a major role in the regulation of the daily burrowing cycle of this species. The effects of salinity on the foraging activities of holothurians are unclear. Mercier et al. (1999Mercier et al. ( , 2000 found that H. scabra cease feeding when water salinity is lowest, and burrow into the substrate. James & James (1994) reported conflicting results, with juveniles emerging during low tide. Skewes et al. (2006) reported that only a third of H. scabra adults emerge during high tide.
Some studies show conflicting results concerning daily burrowing cycles of juveniles and adults. On one hand, daily burrowing cycles can be similar for juveniles and adults of the same species. In Palao Island, H. scabra adults were observed burrowing in the sediment between 03:30 and 15:00 and feeding between 15:00 and 03:30 (Yamanouchi, 1956). In the Philippines, H. scabra juveniles burrow in sediments between 03:00 and 09:00 and remain hidden until they emerge to feed between 15:00 to 03:00 (Altamirano, Recente & Rodriguez Jr., 2017). Despite their different life stages and different geographical regions, both species therefore feed during the same time periods. On the other hand, holothurian size has been reported to affect burrowing and daily cycles. Mercier et al. (1999) found the burrowing cycle of H. scabra to be driven by light for small juveniles (10-40 mm), which burrow at sunrise and emerge close to sunset, whereas temperature was more influential for intermediatesized juveniles (40-140 mm), which emerge earlier in the afternoon. Like many echinoderms, the nocturnal feeding and movement habits of smaller holothurians might be an adaptation to avoid predation (Nelson & Vance, 1979;Hammond, 1982).
To determine particle selectivity in deposit-feeding holothuroids, most studies compare the biochemical composition and concentration of compounds present within sediments around and beneath foraging individuals and in their foregut (Moriarty, 1982;Hammond, 1983;Amaro et al., 2010). Examining particle selectivity is subject to the difficulty of investigating sediment ingestion at the appropriate scale (Lopez & Levinton, 1987). Particle selection is generally based on size, surface texture, specific gravity, and the presence of an organic coating (Taghon, 1989), but it can also be explained by mechanical processes (Jumars et al., 1982). Discussion of particle selectivity by depositfeeding holothurians usually refers to two different modes: the selection of particles with a specific grain size or selection of those with higher organic content. We found 29 studies describing both aspects of particle selectivity for shallowwater (23 studies) and deep-sea (six studies) holothurian species (Table 2). A total of 50 species have been studied, with only one study focused on suspension-feeding species (Hamel & Mercier, 1998).

(a) Size selection of particles
Whether deposit-feeding holothurians exhibit a preference for substrates with a specific grain size is still debated (Mercier et al., 2000), with conflicting results available (Table 2) even within a single species. For example, Mezali & Soualili (2013) stated that H. tubulosa and Holothuria (Roweothuria) poli Dell Chiaje select a grain size ranging between 200 and 600 μm, whereas Massin & Jangoux (1976) found no size preference for these same species. Some authors conclude that holothurians are unselective regarding the size of ingested particles because ingested sediment particles tend to be very similar in size to those of the bottom sediment (Yingst, 1976;Sloan & von Bodungen, 1980;Hammond, 1982). Trefz (1958) noted that diverse holothurian species forage and subsist on different substrates, suggesting that they may focus their feeding on specific substrates. Of the 29 shallow-water species evaluated to date, 20 show grain size selectivity (Table 2); among the four deepsea species studied, only Molpadia blakei (Théel) is unselective (Table 2; Khripounoff & Sibuet, 1980). Particle-size selection seems to be species dependent: preferred grain size is 200-600 μm for H. tubulosa (Mezali & Soualili, 2013), 125-250 μm for H. scabra (Baskar, 1994), and 2000-3500 μm for H. nobilis (Roberts & Bryce, 1982). Deep-sea species seem to prefer finer particles than shallowwater species, with a median of 7-14 μm for Benthogone rosea Koehler and 8-54 μm for Paroriza pallens (Koehler) (Khripounoff & Sibuet, 1980). Particle-size selection may change within a species depending on the season: Holothuria (Stauropora) hawaiiensis Fisher and Bohadschia vitiensis (Semper) prefer finer particles during the spawning period and gravel and coarse particles during the rest of the year (Dar & Ahmad, 2006).
More information is required to assess the benefits of feeding on small-grain sediments for holothurians and the reasons why only some species are selective. Although the mechanisms by which a preferred substrate is selected are not understood, Roberts (1979) proposed that this could be associated with the morphology of the oral tentacles.

(b) Selection of organically rich particles
A wide range of holothurian species from tropical, shallowwater regions (Moriarty, 1982;Hammond, 1983;Uthicke, 1999;Uthicke & Karez, 1999), temperate seas (Hauksson, 1979;Amon & Herndl, 1991), and deep seas (Miller et al., 2000;Wigham et al., 2003Wigham et al., , 2008 are known to select organically rich particles from the sediment (Table 2). These species seem to be attracted by sediments with high organic content (Yingst, 1982) and can differentiate and capture the preferred particles (Massin & Jangoux, 1976;Moriarty, 1982). Out of the 37 species studied, 29 prefer particles enriched in organic matter (   water and deep-sea species seem to be highly selective, with 17 out of 20 (85%) and 12 out of 17 species (70%), respectively, showing selectivity for organic matter (Table 2). Generally, the organic matter content of the ingested sediment is much higher than in the sediment surrounding the animal (Hauksson, 1979). Khripounoff & Sibuet (1980), studying the selective feeding of four abyssal species (Psychropotes longicauda Théel, P. pallens, B. rosea and M. blakei), found that with concentrations of organic carbon and nitrogen in the foregut were four and six times greater, respectively, than concentrations in the local sediment. Moriarty (1982) suggested that particle size selection in holothurian deposit-feeders might be explained by a non-uniform distribution of organic matter in sediments. Uthicke (1999) supported this hypothesis, arguing that a higher organic content is usually associated with smaller sediment grains with low specific gravity, which a richer host microflora (bacteria, microalgae, etc.), due to their higher surface area to volume ratio (Johnstone, Koop & Larkum, 1990). All studies, with one exception [for Holothuria (Holothuria) stellati Delle Chiaje (Mezali & Soualili, 2013)], that showed a preference for a particular grain size also found a preference for organically rich particles where this was investigated (Table 2). A relationship between the organic load and particle size could therefore explain the confusion between these two selective strategies of deposit-feeders.

(c) Patch selectivity
Patch selectivity describes the preference of a mobile organism to feed on patches in a heterogeneous environment (Uthicke & Karez, 1999). A patchy distribution of shallow-water holothurians, thought to be related to their feeding habits (Uthicke & Karez, 1999), has been described in ecosystems including Milne Bay Province and Torres Strait islands (Skewes et al., 2002(Skewes et al., , 2006, the Gulf of Mannar in India (Asha et al., 2015), New Caledonia (Purcell et al., 2009), and most parts of the western Indian Ocean (Conand, 2008). For holothurians, the energetic cost of processing poor sediment is higher than the cost of moving to a more suitable feeding substrate (Mercier et al., 1999). Mercier et al. (1999) found that the locomotive speed of H. scabra juveniles was higher on substrates with low organic matter content (71-331 cm day −1 ) than on substrates of better quality (150-215 cm day −1 ), reflecting an active search for organically rich sediments. Similar results were reported for H. atra and S. variegatus (now S. herrmanni), which move 0-52 m per day and cover greater distances on poorer feeding areas (Yamanouchi, 1939). Other studies on Cucumaria frondosa (Gunnerus) (Hamel & Mercier, 1998), P. californicus (now A. californicus) (Cameron & Fankboner, 1984), Parastichopus chitonoides (Young & Chia, 1982), Actinopyga echinites (Jaeger) (Wiedemeyer, 1994), and H. scabra (James & James, 1994;Altamirano et al., 2017) generally describe the quality of the substrate where holothurians were observed but do not detail optimal substrates in terms of nutritional or ecological features (e.g. optimal grain size, organic matter content, origin of sediments, presence of conspecifics or optimal light regime).
A patchy distribution also has been observed in deep-sea ecosystems for several species (Billett, Llewellyn & Watson, 1988;Ruhl & Smith, 2004) that feed preferentially on nutritionally rich food patches (Hauksson, 1979;Hudson et al., 2005;Jamieson et al., 2011). Particulate organic carbon deposited on the seabed is considered an important factor controlling the local abundance and composition of macrofauna and megafauna (Sibuet, 1985;De Leo et al., 2010). Significant correlations have been found between food availability and megafaunal abundance, particularly for holothurians Ruhl & Smith, 2004). Preferences for selective feeding amongst holothurian species may therefore function in niche partitioning (Roberts, 1979;Sloan & von Bodungen, 1980;Massin & Doumen, 1986), with reduced intra-and inter-specific competition as a result (Jamieson et al., 2011).

IV. THE SEDIMENT INGESTED BY HOLOTHURIANS: A COMPLEX FOOD SOURCE
Despite much published work on holothurians, very few studies have focused on the quality of their food. Table 3 provides Sophistication in a seemingly simple creature a list of known food sources for holothurians, of both organic and inorganic origin. We divide these food sources into two categories: living and non-living fractions. Living fractions include organisms associated with sediments such as bacteria, photosynthetic organisms or meiofauna. Non-living fractions include organically derived detrital matter such as phytodetritus, dead and decaying animals, and faecal pellets and inorganic compounds such as coral scraps, shell remains, coralline algae, foraminiferal tests, and silicates. Sadly, holothurians have not escaped the increasing pollution of the ocean, with ingestion of plastic particles or microplastics recorded (Graham & Thompson, 2009;Renzi et al., 2018).
(1) The living fractions The total biomass of bacteria is relatively high in shallowwater and deep-sea sediments (Zobell & Morita, 1959;Danovaro, Fabiano & Della Croce, 1993;Danovaro et al., 1998;Rex et al., 2006). Most benthic bacteria are not suspended in interstitial water but are attached to mineral or organic sediment particles (Dale, 1974) in aggregates and colonies (Taddéi, 2006). Thus, bacteria could represent a major food source for deposit-feeders (Sorokin, 1972;Massin, 1982) such as holothurians. We found only 10 articles that studied bacteria in the diet of deposit-feeding holothurians, of which four focused on the diversity of ingested bacteria, three on their abundance, and three on both. To our knowledge, the abundance and diversity of bacteria in the gut of suspensionfeeding holothurians has not been assessed.
Most studies found a similar distribution pattern of bacteria in the digestive tract for coastal (Taddéi, 2006;Plotieau et al., 2013) and deep-sea holothurian species (Deming & Colwell, 1982;Roberts et al., 2001;Amaro et al., 2012): an increase in bacterial abundance between the sediment and the foregut contents, then a decrease between the foregut and the hindgut. For example, Taddéi (2006), studying two coastal species H. atra and Holothuria (Mertensiothuria) leucospilota (Brandt), found an abundance of 1.50 × 10 7 bacteria g −1 in coral reef shallow-water sediments of La Réunion. This increased to 3.66 × 10 7 bacteria g −1 in the foregut of H. atra and 4.46 × 10 7 bacteria g −1 in the foregut of Holothuria leucospilota, which decreased significantly to 1.10 × 10 7 and 1.01 × 10 7 bacteria g −1 , respectively, in the hindgut. Bacterial abundance was similar in the faeces (1.91 × 10 7 and 0.87 × 10 7 bacteria g −1 ). Taddéi (2006) reported that about 53% of bacteria are digested by H. atra. Plotieau et al. (2013) reported a higher abundance of 11 × 10 9 bacteria g −1 in the foregut of the coastal species H. scabra, which again decreased in the midgut (4 × 10 9 bacteria g −1 ) and remained stable in the faeces. They estimated that H. scabra digested up to 59% of the ingested bacteria. Amaro et al. (2012) reported a value of 80% for Molpadia musculus Risso. Ward-Rainey et al. (1996) found the opposite pattern for the two specimens of H. atra that they studied. The abundance of bacteria decreased from 3 × 10 4 and 3 × 10 6 colony forming units (cfu) to 3.4 × 10 3 and 6.2 × 10 4 cfu, respectively, between the sediment and the foregut, and then increased to 3 × 10 4 and 1.8 × 10 6 cfu in the hindgut. However, the low number of replicates means that these results should be considered with caution.
All deposit-feeding holothurian species studied to date regarding the abundance of bacteria have peltate tentacles, except M. musculus, which has digitate tentacles (Amaro et al., 2012). No quantitative difference in the ingestion of bacteria has been observed between species with these two tentacle types (Fig. 5). No studies to date have focused on species in Apodida and Dendrochirotida (Table 3). Apodida with pinnate (highly branched) tentacles and suspension-feeding Dendrochirotida with dendritic (ultra-branched) tentacles are likely to show different ingestion rates of bacteria due to their higher tentacle surface area or different feeding behaviour.
Our understanding of the profiles of bacterial abundance in holothurian guts is based on only nine species, and 131 individuals, and additional data on the ingestion of bacteria by holothurians are needed. Deming & Colwell (1982) suggested that holothurians can enhance the value of ingested bacteria by cultivating them in the foregut. Some bacteria may inhabit internal pouches or be attached to the gut epithelium (Harris, 1993). These enteric bacterial strains could help holothurians to digest large molecules through their secretion of hydrolytic ectoenzymes (Roberts et al., 1991). Amaro et al. (2012) studied bacterial diversity through the digestive tract of M. musculus. They found that bacterial diversity showed the same pattern as bacterial biomass: there was a higher number of operational taxonomic units (OTUs) in the foregut (< 100 OTUs) than in sediment (28-71 OTUs), and bacterial diversity decreased towards the hindgut (< 70 OTUs). These observations may highlight the ability of holothurians to cultivate particular strains of bacteria in their foregut. As some endosymbiotic bacterial production could be involved, this requires further investigation.
Only a few studies have explored bacterial diversity in the holothurian gut. Five species have been investigated to date, including one deep-sea species. All these studies found Proteobacteria to be the most abundant taxon (Fig. 5), representing 43% of bacteria in the diet of H. leucospilota (Zhang et al., 2012), 86% in H. scabra (Plotieau et al., 2013), and 55% (Sha et al., 2016) and 89.6% (Gao et al., 2017) in A. japonicus. The forms ingested most were γ-proteobacteria and α-proteobacteria for all holothurian species studied. β-proteobacteria, δ-proteobacteria and ε-proteobacteria were also present in some species.
In coastal species, 15 taxa of bacteria were identified, whereas only seven taxa were found in deep-sea species. Five of these taxa (γ-proteobacteria, α-proteobacteria, CFB lineage, Bacteroidetes and Spirochaetes) were shared between coastal and deep-sea species, perhaps because the diversity of deep-sea bacteria is lower than the diversity of coastal bacteria.
Knowledge on the abundance and diversity of bacteria is based on the study of less than 500 specimens from only Biological Reviews 97 (2022)   Sophistication in a seemingly simple creature 10 species: four from coastal waters and six from the deep sea belonging to Elasipodida, Persiculida, Holothuriida, Molpadida and Synallactida (Table 3). To the extent that these limited findings allow, it can be concluded that holothurians ingest mainly proteobacteria and show a similar distribution of bacteria through the gut across holothurian species, habitat and tentacle type (Fig. 5).
The most abundant photosynthetic taxa in the diet of holothurians seems to be diatoms (Fig. 5). Hamel et al. (1993) analysed the gut contents of Psolus fabricii (Düben & Koren), a suspension-feeding species, and found fewer than 6000 pelagic diatom cells in the first centimeter of the foregut. Diatoms were recorded at much higher concentrations in the gut of Holothuria (Selenkothuria) theeli Deichmann (up to 16,500 organisms ml −1 ; Sonnenholzner, 2003). For the deep-sea species investigated, the sighting frequency of diatoms in P. longicauda, P. pallens, B. rosea, and M. blakei gut ranges between 51 and 84% (Khripounoff & Sibuet, 1980). Tyler et al. (1992) also found a high abundance of diatoms in the gut of P. pallens. However, these studies provide no information on whether the diatoms were alive or whether only the siliceous outer skeleton was present.
Photosynthetic organisms may be significant in the diet of holothurians. Fresh organic matter provided by cyanophytes, diatoms, macroalgae and live seagrass leaves represents more than 50% of the diet for Holothuria poli, H. tubulosa, and H. forskali in Posidonia oceanica meadows in Algeria (Belbachir & Mezali, 2018). In shallow waters off Ecuador, 35% of the gut content of H. theeli consisted of microphytobenthos, with the most dominant taxa being diatoms (91%), followed by cyanophytes and dinoflagellates (Sonnenholzner, 2003). However, these studies did not assess all the possible dietary fractions for holothurians, such as bacteria or minerals, therefore these high proportions may be significant overestimates. Hamel et al. (1993) estimated that phytoplanktonic cells represent between 20 and 50% of the gut content of P. fabricii in autumn and winter, increasing during spring to reach 100% in summer in the St. Lawrence Estuary (Canada). They conclude that the ingestion of phytoplanktonic cells, especially pelagic diatoms ingested by this dendrochirotidan holothurian (suspension feeder), depends on seasonal blooms. Similar results were observed for C. frondosa, with the proportion of phytoplanktonic cells ingested reaching a maximum in summer (Hamel & Mercier, 1998). Sonnenholzner (2003) investigated the ingestion of photosynthetic organisms in a coastal deposit-feeding species from Holothuriida, finding that the relative abundance of microphytobenthic species in the H. theeli gut was significantly higher during the dry season (45.5%) than in the rainy season (20.2%) in the Gulf of Guayaquil. It would be interesting to extend these findings to all fractions present in the holothurian diet to understand which components replace photosynthetic organisms during the cold seasons. Taddéi (2006) used the concentration of chlorophyll a as proxy for photosynthetic organisms in the gut for two coastal deposit-feeders, H. atra and H. leucospilota. The profile observed was the same as for bacterial abundance: chlorophyll a concentration increased from the local sediment (2.87 ± 1.17 μg g −1 ) to the foregut (18.74 ± 7.65 μg g −1 ), decreased in the hindgut (11.28 ± 4.61 μg g −1 ), and remained at this level in the faeces (11.34 ± 4.63 μg g −1 ). These observations may indicate that photosynthetic organisms are actively selected for ingestion by these depositfeeding holothurians.
These studies combined used 378 individuals from six orders of holothurians: Dendrochirotida, Holothuriida, Synallactida, Molpadida, Elasipodida and Persiculida (Table 3). Benthic and pelagic diatoms are a significant dietary component for deposit-feeding and suspension-feeding holothurians respectively, independent of habitat, taxon and tentacle type (Fig. 5). However, the importance of photosynthetic organisms remains unclear without further studies because their biomass is subject to seasonal variations.
Only seven studies have considered the ingestion of meiofauna by holothurians (Table 3). Nematoda, Copepoda and Foraminifera seem to be the taxa ingested most frequently by holothurians although there are interspecific differences. Holothuria atra ingests up to 79% of the meiofauna present in the sediment, with polychaetes the dominant group (Kang et al., 2008). Dissection of H. theeli individuals showed that 65% of the gut contents consisted of meiofauna from eight taxa, with a dominance of crustaceans (46%) and foraminifera (35.5%) (Sonnenholzner, 2003). Belbachir & Mezali (2018) demonstrated that the percentage contribution of meiofauna in the holothurian diet can differ among locations and species. They studied the diet of four species H. poli, H. tubulosa, H. forskali, and Holothuria (Platyperona) sanctori Dell Chiaje, at two sites (Stidia and Salamandre in Algeria; separated by less than 20 km). At Salamandre, 20% of the diet of Holothuria sanctori consisted of crustaceans, whereas in Stidia crustaceans comprised less than 3%. For the three other species, crustaceans represented only 6% of the diet at Salamandre. The contribution of foraminifera ranged between 3.33% for H. sanctori and 15% for H. forskali at the same site.
For the deep-sea species P. longicauda, P. pallens, and B. rosea, the sighting frequency of benthic foraminifera reached 100% of the gut content and was 89% for M. blakei (Khripounoff & Sibuet, 1980). The sighting frequency of Nematoda and Copepoda was also high for these four species (96-100% and 0-87%, respectively). Tyler et al. (1992) recorded abundant foraminifera in the gut of P. pallens but no Nematoda nor Copepoda. Note that the sighting frequency is a poor indicator of the actual contribution of meiofauna to the diet. 290 Uthicke (1999) compared the abundances of Nematoda, Polychaeta and Harpacticoida between the sediment and the midgut content of S. chloronotus and H. atra. Nematoda were the dominant taxon in the sediments (31.33 ± 10.90 individuals 2 ml −1 ), but they were sparsely represented in the midgut of these two species (between 0.50 ± 0.84 and 0.66 ± 1.03 individuals 2 ml −1 ). Polychaeta were more uncommon in the sediment (9.66 ± 5.31 individuals 2 ml −1 ) and in the midgut of S. chloronotus and H. atra (0.17 ± 0.41 and 0.01 ± 0.00 individuals 2 ml −1 respectively). Uthicke (1999) concluded that meiofauna play a negligible role in the nutrition of these two species due to the extremely low abundance recorded in the midgut. The ingestion of these two meiofaunal taxa appears to be in proportion to their presence in the sediment, suggesting that meiofauna are probably ingested accidentally with detrital matter (Khripounoff & Sibuet, 1980;Billett et al., 1988).
One study focused on the ingestion of larval planktonic stages of meiofauna by a suspension-feeding holothurian. Meiofaunal eggs and embryos accounted for less than 15% of the intestinal content of C. frondosa in autumn and winter (Hamel & Mercier, 1998). Maximum zooplankton presence peaked just after breeding of broadcast-spawning species. The actual contribution of meiofauna to the diet of suspension-feeding holothurian thus remains largely unknown. As for photosynthetic organisms, the availability of larval planktonic stages of meiofauna varies seasonally and suspension-feeding holothurians that ingest larval planktonic stages of meiofauna in proportion to their presence in the water column must utilise other resources in the autumn and winter in cold or temperate waters. This implies that meiofauna are not an essential dietary component for holothurians.
The ingestion of meiofauna has been investigated in all holothurian orders except Apodida (Table 3). The meiofauna ingested differs between deposit-feeding and suspension-feeding holothurians (Dendrochirotida), because the latter ingest larval planktonic stages of meiofauna. However, for the other orders Foraminifera, Nematoda, Polychaeta and Crustacea are all ingested at a similarly low rate, independent of tentacle or habitat type (Fig. 5).
In the deep sea, only a small fraction of macroaggregates originating from the euphotic zone reaches the seabed to form detrital matter, with a low increment of 100-150 m day −1 (Gooday & Turley, 1990), limiting their accessibility for holothurians (Thurston et al., 1994;Thurston, Rice & Bett, 1998). Most of this fraction is faecal matter, which represents 95% of vertical particle flow (Wiebe, Boyd & Winget, 1976;Honjo, 1978) and constitutes a key component of the abyssal food web (Frankenberg & Smith, 1967). Holothurians are known to be coprophagous (Bakus, 1973;Hauksson, 1979). However, it remains unclear whether coprophagous holothurians feed on faecal pellets intentionally or simply take advantage of a proximate source of concentrated nutrients. Faecal pellets were found in the gut of all specimens of P. longicauda, P. pallens, B. rosea, and M. blakei examined (Khripounoff & Sibuet, 1980). These faecal pellets were from bivalves, pelagic crustaceans, and unidentified sources. Faecal pellets have not been recorded in the diet of coastal species.
Several studies have suggested that detrital matter is a major dietary component for both shallow-water and deep-sea holothurian species (Bordovskiy et al., 1974;Yingst, 1976;Massin, 1982;Moriarty, 1982;Jeffreys et al., 2011). Plotieau (2012) focused on the assimilation of organic compounds from seagrass phytodetritus by H. scabra. Experiments showed that H. scabra assimilated organic compounds from seagrass leaves, however, this assimilation appeared insufficient to support juvenile growth. Plotieau (2012) therefore suggested that the ingestion of seagrass detritus could be related to the heterotrophic bacteria and microautotrophs attached to them. Indeed, detrital matter, such as phytodetritus, is mainly degraded by bacteria, which colonise it rapidly in both coastal and deep-sea ecosystems (Fenchel & Jorgensen, 1977;Lochte & Turley, 1988;Thiel et al., 1989;Kaiser & Benner, 2008). If correct, this suggests that ingestion of detrital material by holothurians may represent an opportunistic behaviour, to access the bacteria attached to it, rather than detritivorous feeding per se.
The ingestion of detrital matter has been relatively poorly studied, with fewer than 100 individuals analysed (Table 3). No studies have focused on Dendrochirotida (suspension feeders) nor Synallactida. Deep-sea species ingest faecal pellets while coastal species do not. Overall, detrital matter may not be a significant fraction in the diet of holothurians, with the small quantities ingested probably more linked to the presence of bacteria. Sophistication in a seemingly simple creature The non-organic fractions ingested by holothurians are receiving increasing attention, with innovative work suggesting an important role in feeding efficiency and nutritional benefits (Plotieau, 2012).
The sediments ingested by deposit-feeding holothurians contain insoluble clastic products originating from physical and biological breakdown. An analysis of minerals in the sediment ingested by tropical shallow-water holothurians identified both primary and secondary minerals (Plotieau, 2012). The primary minerals were (i) quartz made up of a continuous framework of SiO 4 (identified as the main mineral holothurian gut component in most studies), (ii) calcite, the most stable polymorph of calcium carbonate (CaCO 3 ), (iii) aragonite, a thermodynamically unstable form of CaCO 3 at standard temperature and pressure, and (iv) magnesian calcite (Ca, Mg)CO 3 , a variety of calcite containing randomly substituted magnesium carbonate in a disordered calcite lattice that is present in echinoderm skeletons. The secondary minerals were (i) bioclasts, skeletal fragments of marine or land organisms found in sedimentary rocks (mainly composed of aragonite but also magnesian calcite and calcite), (ii) feldspars (KAlSi 3 O 8 , NaAlSi 3 O 8 , CaAl 2 Si 2 O 8 ), a group of rock-forming tectosilicate minerals, (iii) hornblende or dark amphibole, an isomorphous mixture of calcium-ironmagnesium silicate, aluminium-iron-magnesium silicate, and iron-magnesium silicate), and (iv) other, trace minerals, mainly zircon.

(c) Plastic particles
Recent decades have revealed the impact of increasing plastic input into the ocean on marine taxa, such as fishes, seabirds, turtles, and cetaceans (Cole et al., 2011), and how plastic particles penetrate the marine trophic web (Ivar do Sul & Costa, 2014). Species belonging to lower trophic levels can ingest high levels of plastic particles because they do not differentiate between these and their preferred food (Renzi et al., 2018).
Several studies have recorded the ingestion of plastic particles (Table 3) in shallow-water deposit-feeding species (Graham & Thompson, 2009;Renzi et al., 2018;Mohsen et al., 2019) and suspension-feeding species (Graham & Thompson, 2009;Iwalaye et al., 2020). C. frondosa, Holothuria (Semperothuria) cinerascens (Brandt), Holothuria grisea Selenka, Holothuria floridana (Pourtalès) and H. tubulosa were all shown to ingest plastic particles. Mohsen et al. (2019) found a lower abundance of plastic in the gut of farmed A. japonicus (0-30 particles per individual) than in the local sediments (20-1040 particles kg −1 ). It is likely that most shallow-water species will ingest plastics if they are present in the sediment (Graham & Thompson, 2009). Recent studies show that, due to the vertical transport of particles from the sea surface to the sea floor, plastic particles are sequestered in deep-sea sediments from the Atlantic, Pacific, and Indian Oceans and the Mediterranean Sea (Van Cauwenberghe et al., 2013;Woodall et al., 2014;Fischer et al., 2015). Plastic particles have been found in the gut of other deep-sea benthic invertebrates including Cnidaria, Echinodermata, Arthropoda (Taylor et al., 2016), and Mollusca (Courtene-Jones et al., 2017). Thus, deep-sea holothurian species may also be exposed to and ingest plastics. Mohsen et al. (2019) also found plastic particles in the coelomic fluid, ranging from 0 to 19 particles per individual. Iwalaye et al. (2020) found a similar result in tank experiments with H. cinerascens (32-227 microfibres per individual). In more than half of the individuals studied (57.8%), plastic particles were present in the respiratory tree (0-12 microfibres per individual). The biological impacts of the transfer of microplastic particles from the holothurian gut to the rest of the body were not analysed further. Grossmann (2014) and Assidqi (2015) assessed the impact of the ingestion of plastic particles on H. sanctori and H. leucospilota, respectively. They exposed these holothurians to plastic particles followed by hypoxic conditions to investigate the effects of plastic particle ingestion on resistance to environmental stress. Ingestion of plastic particles did not affect faeces production, evisceration, respiration rate, survival, or behavioural responses in these two species. However, Assidqi (2015) reported a higher susceptibility of H. leucospilota to oxygen depletion 60 days after plastic particle ingestion.
Experiments in mesocosms revealed that holothurians appear preferentially to select plastic particles from the sediment (Graham & Thompson, 2009). The authors hypothesised that the larger surface area of the plastic particles reduces the need to shovel or rake for other particles leading to their selection in preference to sand grains. Similar results were found in a natural environment: analyses of sediment sieved through 63-4,000 μm sieves revealed that H. tubulosa selects 100-2,000 μm plastic fragments for ingestion (Renzi et al., 2018).
When they reach shallow-water sediments, plastic particles can rapidly be colonised by bacteria (Harrison et al., 2014). Dussud et al. (2018) demonstrated that bacterial abundance and diversity is higher on plastic particles than on organic particles. Bacterial communities found on the surface of plastic particles include Proteobacteria, Bacteroidetes (Oberbeckmann et al., 2014;Curren & Leong, 2019), and Biological Reviews 97 (2022)  Cyanobacteria (Oberbeckmann et al., 2014;Dussud et al., 2018). Other organisms, such as diatoms (Carson et al., 2013;Eich et al., 2015) and dinoflagellates (Mas o et al., 2003) can also be attached to plastic particles. All of these organisms are found in the holothurian diet. Finally, because holothurians select more bacteria-rich sediments, they could easily preferentially select plastic particles with bacteria attached, although this requires further investigation.

V. CONCLUSIONS
(1) Cells on the holothurian tentacle bud epidermis are important for food capture. Ciliated cells are likely to be sensory cells that recognise food particles. Mucous cells may be responsible for the production of a physical barrier on the epidermis. The adhesion of food particles is suggested to be performed by the secretion of type-1 secretory cells; when the tentacles are placed in the oral cavity, adhesion may be removed by the secretion of type-2 secretory cells. Further studies are needed to confirm these hypothesised roles in holothurian taxa.
(2) Holothurian nutrition appears to involve two main feeding modes: some species are suspension-feeders whereas others are deposit-feeders, with a few species able to shift between suspension-and deposit-feeding. A third feeding pathway may exist, which relies on nutrient transfer from cloacal water retained by the animal to the haemal system. (3) Synallactida, Molpadida, Persiculida, Holothuriida and Elasipodida, despite their different tentacle types and the different habitats they colonise, ingest similar proportions of different food types. Their diet seems to be composed mainly of proteobacteria. Diatom ingestion may be high, but its overall contribution remains uncertain as its availability varies seasonally.
Other smaller fractions, such as detrital matter, minerals and plastic particles, may be ingested only because they support a high bacterial load. (4) Dendrochirotida is the most recent order of holothurians. The species belonging to this order have dendritic tentacles that allow them to act as suspension-feeders. Most species in Dendrochirotida are coastal (more than 98%). Their trophic specialisation presumably reduces interspecific competition for food, and they ingest large quantities of pelagic diatoms. The ingestion of other food fractions has not yet been studied for Dendrochirotida. (5) There is very little information on the ingestion of food by Apodida. This is a considerable knowledge gap because Apodida represent 17% of holothurian species. More than half of Apodida species have pinnate tentacles that are not present in other orders.

VI ACKNOWLEDGEMENTS
The authors would like to thank A. Modi for drawings and P. Cuet for advice about the contribution of photosynthetic organisms to the holothurian diet. This study was supported by a research program Ecosystèmes marins "Holomicro" funded by Région Réunion and Fonds européen de développement régional (FEDER) PO 2014-2020. J.P. is funded by the European Union FSE programme. The authors thank the three anonymous reviewers who greatly contributed to the improvement of this review.