Marine invertebrates typically present indirect development and their generally planktonic larvae act as the dispersive stage. This is a particularly important adaptation, as it directly influences species distribution and abundance, and also enables geographical range expansion and recovery of populations (Strathmann et al., 2002). Many factors influence larval development and its dispersal, such as oceanographic conditions, predation avoidance by planktivorous organisms, energy reserves, and feeding abilities (Pechenik, 1990). As a response to such wide array of scenarios, there is a significant diversity of larval forms presented by metazoan organisms (Wray and Raff, 1991). Certain larval types are planktotrophic and required to feed, thus presenting specialized organs, while others are lecithotrophic and can survive through metamorphosis while relying solely on endogenous reserves (Thorson, 1950; Pechenik, 1999).

Despite the diversity of requirements for different marine invertebrate larvae, certain larval forms not only require to feed in the plankton, but also acquire symbionts (Fitt et al., 1986; Mieog et al., 2009; Garcia Ramos and Banaszak, 2014). Many metazoan phyla, especially scleractinian corals within the Cnidaria, engage in a mutualistic symbiosis with the photosynthetic dinoflagellates in the genus Symbiodinium (Stat et al., 2006; Venn et al., 2008). The majority of these corals are broadcast spawners and newly-formed planula larvae, while lecithotrophic, must acquire their symbionts horizontally (Harrison and Wallace, 1990; Baird et al., 2009; Harrison, 2011). The planktonic larvae acquire different free-living Symbiodinium strains from the environmental pool before selecting one or more specific strains at a later stage (Belda-Baillie et al., 1999; Weis et al., 2001; Coffroth and Santos, 2005).

Further, although numerous studies are available on animal-dinoflagellate symbioses, there is still an overall paucity of knowledge on this relationship during host larval development, especially in regard to metabolite exchange and symbiosis-related differential gene expression. In fact, this paucity of information raises the question as to whether the mutualistic nature of this relationship is already present during host larval development or becomes established at a later stage. Likewise, while there is much information on the impacts of climate change and anthropogenic influence on adult and symbiotic coral reef organisms (Hoegh-Guldberg, 1999; Hughes et al., 2003; Hoegh-Guldberg et al., 2007), comparatively little is known about the impacts of these environmental changes on Symbiodinium-associated animal larvae. Therefore, here we aim to compile, describe, and interpret the available data on metazoan larva-Symbiodinium associations, with emphasis on the ecological, molecular and biochemical relationship, evolution, and impact of climate change.

Four different metazoan phyla produce larval forms that associate with Symbiodinium: Porifera, Cnidaria, Acoelomorpha, and Mollusca (Table 1). Two protist taxa are also known to associate with Symbiodinium, Foraminifera (Pochon and Gates, 2010), and Ciliophora (Mordret et al., 2016). Invertebrate larvae acquire symbionts in two different ways: vertically (transmitted to the offspring in parental oocytes) or horizontally by oral ingestion during larval development. Both strategies have their advantages. Vertical transmission ensures that a successful symbiont strain is supplied to the larvae (Wilkinson and Sherratt, 2001), while horizontal transmission allows for the larvae to acquire a diversity of symbionts, some of which may be more compatible with different environmental conditions encountered after dispersal (Baird et al., 2007).

Sponges and acoelomorph worms are the two groups in which particularly little is known about their association with Symbiodinium, especially during larval development. Within sponges, the association with Symbiodinium seems to be largely restricted to the family Clionaidae (Hill et al., 2011) and parenchymella larvae contain parentally-seeded symbionts (Mariani et al., 2000, 2001). In the case of acoelomorph worms, the genus Waminoa harbors two different symbiotic dinoflagellates, Symbiodinium and Amphidinium, which are also transmitted vertically (Barneah et al., 2007b). Other acoelomorph genera are known to associate with Symbiodinium (Trench and Winsor, 1987), but no information regarding their larval development has been reported.

In the case of cnidarians, while broad knowledge on their association with Symbiodinium is available, there are still relatively few studies on the larva-Symbiodinium association (Rodriguez-Lanetty et al., 2006; Voolstra et al., 2009b; Schnitzler and Weis, 2010; Wolfowicz et al., 2016). Approximately 80-90% of reef-building coral species (Scleractinia) are broadcast spawners and acquire symbionts horizontally (Harrison and Wallace, 1990; Baird et al., 2009; Harrison, 2011). On the other hand, many of the less common brooding species transmit symbionts vertically, most notably in the family Pocilloporidae (Baird et al., 2009). In the case of soft coral species (Alcyonacea), the incidence of vertical transmission seems to be higher (Benayahu et al., 1988; Achituv et al., 1992) than for scleractinians. Coral larvae are typically lecithotrophic (Morse et al., 1996) and while symbiont acquisition may improve host fitness, it does not seem to be a requirement for metamorphosis (Schwarz et al., 1999; Mieog et al., 2009). In the case of other cnidarians such as anemones and jellyfish, there is still very little information on the role of Symbiodinium in their larval ecology.

Among mollusks, giant clams (subfamily Tridacninae) are possibly the most conspicuous species associated with Symbiodinium. In this group symbiont acquisition is horizontal (Fitt and Trench, 1981) and a requirement for successful metamorphosis (Mies et al., 2017a). Differently from other metazoan hosts, symbionts are only acquired during the second larval stage (veliger), as the first larval stage is a non-feeding trochophore (Fitt et al., 1984; Heslinga et al., 1984). To date, giant clams are also the only known Symbiodinium hosts that during larval development do not move symbionts from the digestive tract or equivalent to endodermal tissues (Schwarz et al., 2002; Marlow and Martindale, 2007; Huang et al., 2008), as they remain in the digestive tract throughout the larval development (Fitt et al., 1986; Hirose et al., 2006). The other mollusk group associated with Symbiodinium is Gastropoda. The Queen Conch Strombus gigas is the largest among them to do so and symbiont acquisition is also horizontal and apparently required for metamorphosis (Garcia Ramos and Banaszak, 2014). While a large diversity of nudibranchs (mainly aeolids) are known to associate with Symbiodinium, there is a scarcity of larval experiments, possibly due to their typically restricted diet (Dionísio et al., 2013). The only exception is Berghia stephanieae, a species that maintains Symbiodinium in dorsal extensions of the digestive diverticulum. Its lecithotrophic veliger larvae undergo metamorphosis in the absence of Symbiodinium (Carroll and Kempf, 1990; Kempf, 1991), but recent studies suggest that this species does not engage in a mutualistic relationship (Mies et al., 2017c). While nudibranchs usually acquire symbionts horizontally (Loh et al., 2006; Burghardt and Wägele, 2014; Ahmadian et al., 2016), a case of vertical transmission has been recently reported (Wecker et al., 2015).

Symbiodinium dinoflagellates may be found in two different life stages, a planktonic and free-living zoospore or a symbiotic and non-motile coccoid cyst (Freudenthal, 1962; Schoenberg and Trench, 1980; Stat et al., 2006). Thus, by presenting a free-living stage, the mutualistic association is not obligate for Symbiodinium (Stat et al., 2006). Species within this genus are morphologically cryptic and identification is performed through the use of molecular biology techniques (LaJeunesse et al., 2012; Arif et al., 2014). Based on ribosomal, plastid and mitochondrial phylogenies, Symbiodinium are divided in nine clades (A–I, Figure 1) and many subclades (Rowan and Powers, 1991; LaJeunesse, 2001; Pochon et al., 2014). Species-level diversity is still being assessed, mainly through sequencing of the internal transcribed spacer 2 (ITS2, see LaJeunesse and Trench, 2000), either by denaturing gradient gel electrophoresis (DGGE) based separation of ITS2 genomic copies (LaJeunesse and Trench, 2000; LaJeunesse, 2002; Thornhill et al., 2006) or by next-generation sequencing based elucidation of ITS2 diversity (Arif et al., 2014; Batovska et al., 2016; Hume et al., 2016; Ziegler et al., 2017). As a result, many new taxa and/or strains have been described recently (LaJeunesse et al., 2012; Hume et al., 2015; Lee et al., 2015; Ziegler et al., 2017), with probably more than 100 extant species being present in the genus Symbiodinium (LaJeunesse, 2001). Next-generation sequencing methods also allow to approach cryptic diversity in Symbiodinium-host relationships, i.e., the presence of rare, background Symbiodinium taxa that might be missed by genotyping methods with a lower resolution. This approach holds a dual premise: firstly, intragenomic allelic diversity of the ITS2 marker allows to resolve differences within and between individuals and species previously not possible (Batovska et al., 2016); secondly, in the case of Symbiodinium, deep sequencing of ITS2 diversity may uncover symbiont associations previously not detected (Arif et al., 2014; Boulotte et al., 2016). However, the contribution of low abundant Symbiodinium taxa to host resilience remains to be determined (Lee et al., 2016). Regardless of these considerations, the highest species diversity is found within clade C, that dominates Indo-Pacific coral reefs and is also one of the most abundant clades in the Atlantic (LaJeunesse, 2005). Host species tend to associate with a single Symbiodinium strain (Goulet, 2006), but multiple symbiont taxa (of different clades) may be found within a single host specimen (Rowan and Powers, 1991; Carlos et al., 2000; Baker, 2003). The type of Symbiodinium is a relevant factor for holobiont fitness, in both adult and larval stages (Schwarz et al., 1999; Mieog et al., 2009). While it has been reported that there is host life stage-specificity for symbiont strains, as juvenile and adult acroporid corals actively associate with different clades (Abrego et al., 2009), there is no life-stage specificity reported for larval stages yet.

Animal larvae acquire symbionts from different sources, but mainly from the water column and sediment (Adams et al., 2009; Cumbo et al., 2012; Nitschke et al., 2016). The amount of Symbiodinium cells acquired is highly variable among species and may vary according to environmental conditions. For acroporid corals Cumbo et al. (2012) report less than 10 symbiont cells acquired by larvae, while Harii et al. (2009) registered an average of 50-60 symbiont cells, but with high variability. Mies et al. (2017c) report about 30 cells for both giant clam and nudibranch larvae and more than 100 for coral larvae.

Animal hosts are able to select Symbiodinium strains through post-phagocytic recognition mechanisms involving lectin/glycan interactions (Koike et al., 2004; Wood-Charlson et al., 2006; Fransolet et al., 2012). These mechanisms are also active during larval development and reports also show that coral (Bay et al., 2011) and gastropod (Garcia Ramos and Banaszak, 2014) larvae actively select symbionts. Desired Symbiodinium species are kept while undesired ones suffer apoptosis (Dunn and Weis, 2009). However, there are reports that coral larvae may have no preference for different symbiont strains (Cumbo et al., 2012). In a study testing the offer of a homologous and two heterologous strains to Fungia scutaria larvae, there was no statistical difference in acquisition among strains, except for one of the heterologous strains that had significantly lower acquisition (Rodriguez-Lanetty et al., 2004). Interestingly, homologous strains have been shown to distribute differently in larval tissues than heterologous strains (Rodriguez-Lanetty et al., 2006), and it has also been reported that homologous symbionts establish a more benefitting symbiosis (Trench et al., 1981; Schwarz et al., 1999; Dunn and Weis, 2009). However, Mies et al. (2017c) found no differences in survival in coral, nudibranch, and giant clam larvae associated with both homologous and heterologous Symbiodinium clades. Further, a recent study by Biquand et al. (2017) suggests that at least part of the specificity in the host-symbiont relationship may be determined by symbiont cell size, where Symbiodinium strains with a larger cell size are less likely to establish a symbiotic relationship than taxa with a smaller cell size using the laboratory model Aiptasia (Baumgarten et al., 2015) and corals. Nevertheless, the presence of symbionts in larval tissues (see Figure 2) may significantly contribute to its development. Giant clam larvae infected with Symbiodinium grow significantly larger than aposymbiotic larvae (Fitt et al., 1986; Mies et al., 2012) and F. scutaria larvae settle earlier in the presence of symbionts (Schwarz et al., 1999). However, higher growth not always suggests the establishment of symbiosis as some coral larvae have been shown to digest the symbionts (Titlyanov et al., 1998).

The metabolite exchange between Symbiodinium and its adult host has been extensively documented. The host supplies the symbiont with CO₂ and other compounds for cellular synthesis such as nitrogen and phosphorus (Trench, 1979; Allemand et al., 1998; Leggat et al., 2003; Weis et al., 2008). In turn, the symbiont supplies the host with more than 90% of its metabolic requirement, in the form of organic compounds including glucose, glycerol, fatty acids, and amino acids (Muscatine, 1990; Grant et al., 1997; Papina et al., 2003; Burriesci et al., 2012). Still, very little is known about the metabolite exchange between Symbiodinium and metazoan larvae. To our knowledge. Kopp et al. (2016) is the only study to date confirming active metabolite exchange by showing that labeled ¹³C and ¹⁵N are translocated from symbiont to Pocillopora damicornis larvae. However, that same study found that these compounds contribute rather insignificantly to the larval nutrition. Other reports show that Symbiodinium types within clades A and C associated with coral and giant clam larvae produce a higher amount of the fatty acid DHA (docosahexaenoic acid, 22:6ω3–Mies et al., 2017c) and that Symbiodinium associated with giant clam larvae express a glycerol-synthesizing gene (Mies et al., 2017b), but in neither of those cases translocation was effectively confirmed.

One of the main issues hindering the better understanding of this relationship is the lack of molecular studies and information for Symbiodinium, also addressed in Leggat et al. (2011). The genomes for types within clades A, B, and F (Shoguchi et al., 2013; Lin et al., 2015; Aranda et al., 2016), transcriptomes for types within clades A-D (Bayer et al., 2012; Ladner et al., 2012; Parkinson et al., 2016) and the plastid genome for type B1 (Mungpakdee et al., 2014) and type C3 (Barbrook et al., 2014) have been sequenced in recent years, but further information, especially on metabolic pathways, is still missing. However, a few experiments investigated differential gene expression in coral larvae with and without symbionts, all of which reported the absence of or rather small gene expression differences in the transcriptomes of symbiotic hosts when compared to the aposymbiotic state (deBoer et al., 2007; Voolstra et al., 2009b; Schnitzler and Weis, 2010). Only recently, a study by Mohamed et al. (2016) detected a transient period of differential expression involving a limited number of genes (3% of assayed transcriptome) 4 h after the exposure of Acropora digitifera planula larvae to a competent strain of Symbiodinium. While yet to be applied for larval studies, symbiosis-specific genes have been identified recently in Aiptasia anemones (Bucher et al., 2016; Wolfowicz et al., 2016). From the symbiont perspective, apart from the identification of protein kinases that may be involved in the establishment of symbiosis (Rosic et al., 2014), a symbiosis-specific gene was identified in Symbiodinium clade A, an H⁺-ATPase (Bertucci et al., 2010). This gene is the only symbiosis-specific gene that had its expression verified in Symbiodinium associated with invertebrate larvae (Mies et al., 2017b,c). The expression of this gene was confirmed in the veliger larvae of the giant clams Tridacna crocea and T. maxima and in the planula larvae of the hermatypic coral Mussismilia hispida, suggesting that symbiotic interactions are present.

Symbiodinium is found in association with a diversity of polyphyletic hosts, which differ regarding aspects of symbiont uptake and symbiosis maintenance. Hence, the symbiosis between animal hosts and Symbiodinium likely formed independently on multiple occasions. The ability of Symbiodinium as well as other dinoflagellates to establish symbioses with such a diversity of hosts may derive from their evolutionary background as recently suggested by Aranda et al. (2016). Parasitism is common in many extant dinoflagellates (Kuperman and Matey, 1999) as well as their closest relatives, the apicomplexans (Shoguchi et al., 2013; Lin et al., 2015). It is thus possible that a putative parasitic evolutionary background of Symbiodinium facilitated the infection of different hosts and their subsequent proliferation as endosymbionts. Hence, while the initial symbiosis of metazoan hosts and Symbiodinium may have been of parasitic nature, these symbioses evolved toward mutualistic interactions over time. Indeed, Aranda et al. (2016) recently confirmed adaptations in Symbiodinium, which may have facilitated the nutrient exchange with their hosts, such as an enrichment in carbon and nitrogen transporter domains. Yet, or as evidence to the derived mutualism notion, even extant metazoan-Symbiodinium associations are not exclusively mutualistic interactions (Stat et al., 2008; Lesser et al., 2013). Rather these interactions may exist along a continuum ranging from parasitism to mutualism depending on the identity of the host and symbiont (Sachs and Wilcox, 2006; Stat et al., 2008). Hence, the degree of mutualism between these partners may ultimately dependent on the specificity of this association and the ability of the host to select for beneficial symbionts. Further, as the performance of Symbiodinium types depends on their host as well as environmental conditions, different hosts may select for different symbionts (Baker, 2003; Mieog et al., 2009). Thereby, the large diversity of metazoan hosts found in association with Symbiodinium likely contributed to the rapid divergence found in this genus (Pochon et al., 2006; Hume et al., 2016).

Climate change is having a significant impact on coral reef ecosystems, particularly because of ocean acidification and the rise in seawater temperature (Hoegh-Guldberg, 1999; Hughes et al., 2003; Pandolfi et al., 2003; Hoegh-Guldberg et al., 2007). While the effects of ocean acidification have been investigated independently in marine invertebrate larvae and Symbiodinium (Brading et al., 2011; Nakamura et al., 2011; Waldbusser et al., 2013; Rivest and Hofmann, 2014), they yet need to be assessed on invertebrate larva-Symbiodinium associations. Nevertheless, it is known that the effects of ocean acidification on cultured (free-living) Symbiodinium is phylotype-specific, with some strains (such as A13 and A2) in fact benefitting from it and increasing their photosynthetic capacity (Brading et al., 2011). As it has also been shown that symbiont presence may mitigate the combined effects of acidification and hypoxia in non-calcifying hosts (Klein et al., 2017), it would be interesting to verify whether the same would happen during host larval development.

In contrast, the bleaching phenomenon, defined by the loss of Symbiodinium cells from host tissue due to rising seawater surface temperatures has been well documented (Brown, 1997; Hoegh-Guldberg, 1999; Fitt et al., 2001). For adult and larval hosts alike, the presence of symbionts is generally regarded as positive for coral reef larvae development and fitness, but under thermal stress it may have the opposite effect. High temperatures cause the symbionts to become a source of reactive oxygen species (ROS), which leads to significantly higher larval mortality (Weis, 2008; Baird et al., 2009; Yakovleva et al., 2009; Schnitzler et al., 2012). Some of the effects of ROS include DNA damage and higher rates of antioxidant activity (Yakovleva et al., 2009; Nesa et al., 2012).

The bleaching threshold for corals and coral larvae alike is usually only 1.0–1.5°C above the prevailing maximum temperature (Edmunds et al., 2005; Mieog et al., 2009). At this time, the expression of heat-shock proteins and oxidative-stress genes becomes detectable (DeSalvo et al., 2008, 2010; Rodriguez-Lanetty et al., 2009; Meyer et al., 2011). Data by Mies et al. (2017c) show that coral and giant clam larvae from Vietnam bleach at 29°C and hardly survive past 32°C. However, the highly-tolerant larvae of Pocillopora damicornis (from Okinawa, Japan) have been shown to withstand a temperature of 32°C, despite losing the majority of its symbiont cells (Haryanti et al., 2015). Nevertheless, the bleaching threshold may be greatly influenced by host and symbiont physiology and genetics, and also by the oceanographic conditions in situ (Roik et al., 2016). The symbiont type associated with the larvae may also influence bleaching intensity, as coral and giant clam larvae associated with Symbiodinium types within A and C bleach at lower rates than types within clades B, D, E, and F (Mies et al., 2017c). However, further investigation is required, especially for strains in clade D, which increase the adult host temperature tolerance threshold and become dominant during bleaching events (Berkelmans and van Oppen, 2006; Jones et al., 2008; Mieog et al., 2009; Ladner et al., 2012). Taken together, the effects of climate change on larval-symbiont associations are not clear at present, but increasing seawater temperatures may lead to increased loads of ROS, which in turn may limit the association of certain symbiont types with host larvae.

Symbiosis is usually defined as the long-term interaction between two different organisms and mutualism is defined as a beneficial relationship for both parties. Considering the information presented in the previous sections, it remains to be determined whether a mutualistic symbiosis is in place for most metazoan larva-Symbiodinium associations. According to Davy et al. (2012), the establishment of symbiosis between Symbiodinium and its host is a complex process that involves several steps: symbiont uptake, phagocytosis by host cells (in the case of intracellular relationship), function integration, symbiont reproduction, and long-term persistence. Therefore, for hosts in their larval development, the establishment of symbiosis would have to follow the steps in Figure 3, which include symbiont acquisition, transfer of symbionts from gut to endodermal tissue, metabolite exchange and symbiont reproduction and persistence.

The vast majority of experiments performed on invertebrate larva-Symbiodinium associations only visually observed the acquisition of symbionts and no metabolite exchange was detected, which fails to meet the criteria and events described above and shown in Figure 3. Especially considering reports that some larvae may digest the symbionts instead of engaging in a symbiotic relationship (Titlyanov et al., 1998; Mies et al., 2017c). The available gene expression studies are inconclusive in regard to whether the observed gene expression changes in coral larvae are specific to the symbiont exposure (deBoer et al., 2007; Voolstra et al., 2009a; Schnitzler and Weis, 2010; Mohamed et al., 2016). For instance, it is unclear if these changes are related to a mutualistic relationship or to the mere presence of a foreign algal body. Yet, the reports of higher production of fatty acids by Symbiodinium associated with coral and giant clam larvae (Mies et al., 2017c) and increases in larval caloric content and competence due to symbiont presence (Ben-David-Zaslow and Benayahu, 1998) suggest metabolite exchange. To date, perhaps the most compelling confirmation of a symbiosis between metazoan larvae and Symbiodinium may come from the expression of symbiosis-specific marker genes and the translocation of organic compounds. In this regard, the studies by Mies et al. (2017b,c) and Kopp et al. (2016) may provide critical support. These studies show the expression of a putative symbiosis-specific gene by Symbiodinium associated with coral (Mussismilia hispida) and giant clam (Tridacna crocea and T. maxima) larvae and the transfer of labeled carbon and nitrogen from symbiont to coral planulae (Pocillopora damicornis), respectively. However, these authors also point out that the energetic intake from this is insignificant for larvae nutrition, supporting that the initial symbiotic relationship might not be mutualistic in nature. Hence, the search for marker genes that could potentially disentangle different states in the symbiotic relationship (i.e., from an initial parasitic or commensal to a mutualistic interaction) is warranted.

Taken together, hosting Symbiodinium may be of no or little direct benefit to early larval animal host life stages. Yet, a mutualistic relationship is present in adult invertebrate–Symbiodinium associations. Thus, a gradual transition of the nature of the symbiotic association toward mutualism has to occur at one point during host development. In this context, the selection of symbionts within the host may increase the relative abundance of beneficial symbionts (Wolfowicz et al., 2016). Further, the rate of carbon translocation per symbiont may increase as the density of symbionts increases in relation to the host cells. The translocation of carbon by Symbiodinium requires the availability of excess carbon (Ezzat et al., 2015). In mature symbioses, it is suggested that this is achieved by the constant nitrogen limitation of symbiont growth (Falkowski et al., 1993; Pogoreutz et al., 2017). Yet, it is questionable whether algal symbionts are nitrogen-limited during early larval host stages. Nitrogen from host metabolism is an important source for algal nutrition (Rädecker et al., 2015). However, the ratio of symbiont to host biomass is significantly higher in early larval stages than in adult hosts (Muscatine et al., 1998). Thus, significantly more host-derived nitrogen may be available to Symbiodinium during early larval development. While this would enable a rapid proliferation of symbionts, it would inhibit the availability of excess carbon available for translocation to the host. Hence, the onset of nitrogen limitation and subsequent initiation of carbon translocation may only occur once the density of symbionts in association with the host has passed a certain threshold. This scenario, although speculative at this point, would explain the transition toward mutualistic interactions during the maturation of the symbiosis between Symbiodinium and marine animal invertebrates.

Our review provides an analysis on the relationship between Symbiodinium and marine invertebrate larvae. The information compiled and discussed shows that there is significant heterogeneity in larval response to initial infection by the symbiont, even within the same host taxon. Some species may digest symbionts while others actively engage in metabolite exchange and certain larval forms apparently select symbiont types while others do not. Symbiont presence may be imperative for the larval development and recruitment of certain host groups, while symbiont identity and homology may improve larval fitness and contribute to thermal tolerance. However, for the majority of cases investigated, the mutualistic relationship cannot be unequivocally confirmed. Rather, from the current available data, we suggest a gradual transition from parasitism/commensalism to mutualism, both on evolutionary and larval development time scales.

All authors listed, have made substantial, direct and intellectual contribution to the work, and approved it for publication.