The ecology and phylogeny of cyanobacterial symbionts in sponges


Kayley Usher, School of Plant Biology, University of Western Australia, Stirling, Crawley, WA, Australia HWY 6009.


Cyanobacteria have flexible photosynthetic apparatus that allows them to utilise light at very low levels, making them ideal symbionts for a wide range of organisms. Sponge associations with cyanobacteria are common in all areas of the world, but little is known about them. Recent research has revealed new cyanobacterial symbionts that may be host specific and two major clades, ‘Candidatus Synechococcus spongiarum’ and Oscillatoria spongeliae, that occur in widely separated geographic locations in unrelated sponge hosts. These clades may represent a cluster of closely related symbiont species, or may be single species that are maintained by periods of horizontal transmission over large distances. Erroneous assumptions regarding the importance of cyanobacterial symbionts to the survival of individual sponges or species may arise from cyanosponges being deemed to be phototrophic or mixotrophic without studies of their photophysiology. This review brings together recent and past research on cyanobacterial associations with sponges, including their biogeography, phylogeny, host specificity, and ecology.


Photosynthetic sponges are significant components of tropical and temperate reef ecosystems around the world, providing food and habitat for a wide range of organisms. They fill a similar ecological niche to the hard corals, but are more extensive in their distribution and more varied in the range of symbioses formed. Hosts include a wide range of Demospongiae and Calcarea (Diaz 1996), and symbionts include cyanobacteria and eukaryotic rhodophytes, diatoms, dinoflagellates and chlorophytes (Rützler 1990; Taylor et al. 2007 and references therein). Unfortunately, photosynthetic sponges have not captured the public imagination in the same way that hard corals have and our understanding of these symbioses remains rudimentary.

Cyanobacteria appear to be the most important group of photosynthetic symbionts and their presence can cause significant modifications in sponge morphology (Taylor 1973; Saràet al. 1998). Sponges with cyanobacterial symbionts are referred to as ‘cyanosponges’ and, to date, 100 cyanosponge species have been reported (Diaz et al. 2007) in 26 Demospongiae and 17 Calcarea families (Diaz 1999). Cyanobacterial symbionts are usually intercellular (Rützler 1990; Wilkinson 1992), but sometimes occur in specialised vacuoles termed ‘cyanocytes’ (Wilkinson 1978). Although recent papers have significantly expanded the known phylogeny and biogeography of cyanobacterial symbionts, it is likely that more species remain to be discovered. The methods by which cyanobacterial symbionts are transmitted to new sponge generations, recognition (or evasion of the sponge immune system), and regulation of symbiont numbers are largely unknown, as are the range of contributions to the host and their importance to host success.

Typically, 30–50% of sponges on tropical reefs are cyanosponges, but the percentage is variable within regions and is sometimes as high as 80–90% (Wilkinson 1983, 1987b, 1992; Rützler 1990; Diaz 1996). Diaz (1996) examined the relative success of species of cyanosponges compared to other sponges in Belize by investigating their relative abundance and size. Seven of the nine most abundant species hosted photosynthetic symbionts, with six of these having small single-celled cyanobacteria. The latter covered an area 35–45% of that covered by the total sponge population. Cyanosponges are probably faster growing and more competitive for space than sponges without photosynthetic symbionts (Wilkinson & Cheshire 1988), and some dominate the benthos (Borowitzka & Hinde 1999) or have become aggressively competitive in disturbed areas, overgrowing and killing live coral (Vicente 1985; Rützler & Muzik 1993; Diaz et al. 2007) (Fig. 1).

Figure 1.

 The cyanosponge Chondrilla australiensis overgrowing live coral in Western Australia.

Recent studies investigating cyanosponges with molecular tools have greatly expanded our knowledge of these symbioses. This review provides an introduction to cyanobacteria and explores the current understanding of cyanosponges and how they function in reef ecosystems. It looks at the benefits of cyanobacteria to sponge hosts, modes of cyanobacterial symbiont transmission, the biogeography of different types of association, and the diversity and host specificity of symbionts.


Cyanobacteria are unique among the prokaryotes in their ability to form symbioses with a broad range of hosts (Rasmussen & Johansson 2002), serving as ‘chloroplasts’ in symbioses with a variety of non-photosynthetic partners, including marine invertebrates (sponges, ascidians, echiuroid worms) and fungi (Raven 2002). They are also found in symbiosis with photosynthetic hosts, including diatoms, mosses, liverworts, ferns and cycads, where they fix atmospheric nitrogen (N2) (Raven 2002).

Originally classified under the botanical code, cyanobacteria have been variously referred to as ‘blue-green algae’ and ‘zoocyanelles’. Today, cyanobacterial phylogeny remains confused (Honda et al. 1999). Molecular studies have demonstrated that members of many genera are not closely related and traditional taxonomy based on morphology, pigment composition and nutritional needs are not accurate (Turner 1997; Urbach et al. 1998).

Thylakoid membranes are found in all cyanobacteria and are the most obvious ultrastructural feature that distinguishes them from other eubacteria. Phycobilisomes, the light-harvesting complexes of cyanobacteria, are attached to the outside of the thylakoid membranes. These protein-pigment aggregates contain chlorophyll a and the phycobiliproteins phycocyanin, allophycocyanin and phycoerythrin. However, the Prochlorales are an exception, with most having chlorophyll b and lacking phycobiliproteins. Each phycobiliprotein absorbs light maximally from a different part of the light spectrum, and together they absorb energy that is poorly utilised by chlorophyll. Some cyanobacteria are able to change the ratios of phycobiliproteins in a process called complementary chromatic adaptation to make maximum use of the light that reaches them.

The photosynthetic apparatus of cyanobacteria is extremely flexible, adapting to both the quantity and quality of light and enabling them to grow in a wide range of irradiances, including ocean depths receiving only 1% of surface irradiance (Furnas & Crosbie 1999) and full sunlight. In addition to changing the relative ratios of phycobiliproteins, adaptation is achieved by changing the concentration and the size of light-harvesting complexes (Bennett & Bogorad 1973; Wyman & Fay 1986; Golubic et al. 1999; Moore & Chisholm 1999), and light adaptation has been observed in the symbionts of Cymbastela spp. (Seddon et al. 1993). Phycoerythrin is more efficient in low light and its predominance in these conditions is often reflected in the red/brown colour of the sponge host, while levels of the green-yellow phycocyanin are higher in high light (Wyman & Fay 1986).

Cyanobacteria avoid photoinhibition caused by strong light by moving phycobilisomes around on the thylakoid membrane. Phycobilisomes frequently detach and re-associate with reaction centres, changing their distribution between Photosystem II and Photosystem I (Joshua & Mullineaux 2004). In very intense light these state transitions allow cyanobacteria to maximise their use of available energy, and reduce photodamage to Photosystem II by absorbing excess energy. This capability may decrease the chances of cyanosponges bleaching.

Phycoerythrin has a distinct yellow/orange (563 nm) fluorescence spectrum when excited by blue light (450–490 nm), a characteristic that can be useful for rapidly visualising cyanobacteria in fresh or frozen tissue (algae fluoresce red). However, care must be taken when using fluorescence microscopy to determine the occurrence of Oscillatoria spongeliae in sponge samples. The trichomes of this symbiont are easily damaged, with cells becoming swollen and then bursting (Hinde et al. 1994). Damaged cells rapidly lose phycobiliproteins (Larkum et al. 1987) and this also occurs in samples preserved by freezing (K. Usher, personal observation). Very high background autofluorescence in sponge tissue is a sign that photosynthetic pigments have been lost from cyanobacterial symbionts, the cells of which may not be apparent.

Biogeography of Sponge/Cyanobacterial Associations

It has been stated that symbioses of invertebrates with photosynthetic organisms are significantly more prevalent in the nutrient-poor waters of tropical regions than in temperate waters (Wilkinson 1983, 1987c, 1992; Wilkinson & Cheshire 1990; Verde & McCloskey 1996; Davy et al. 1997); however, in most cases this has not been established quantitatively (Davy et al. 1997). Cyanobacterial associations with sponges may be as common in temperate areas as in tropical ones. Roberts et al. (1999) estimated that more than 65% of sponges in the temperate waters of eastern Australia contained photosynthetic symbionts, and they are also common in the nutrient-rich waters in Papua New Guinea (Diaz 1996), where high rainfall, mountainous terrain and tectonically active geography contribute large quantities of solutes and sediment to coastal areas (Brunskill 2004). The hypothesis that cyanosponges are largely restricted to nutrient-poor waters therefore does not appear to be correct, and probably results from a lack of studies in temperate waters.Cowen (1983) suggests that a continuous supply of photosynthate has a greater importance for the host than its calorific value would suggest, especially in conditions of low food supply. I hypothesise that the additional energy donated by photosynthetic symbionts allows the host to expend more energy for reproduction and growth than non-photosynthetic sponges in the same area, giving it an advantage over competitors even in nutrient-rich waters.

Indeed, Oscillatoriaspongeliae and ‘Candidatus Synechococcus spongiarum’ clades occur in many regions of the world. ‘Candidatus Synechococcus spongiarum’ has been sequenced from the Mediterranean, Red Sea, Zanzibar, the Caribbean, tropical and temperate Australia, Bahamas and Guam (Diaz 1997; Hentschel et al. 2002; Gómez et al. 2004; Usher et al. 2004a; Steindler et al. 2005; Thacker 2005; Taylor et al. 2007). Oscillatoria spongeliae has been sequenced from Palau and Guam (Micronesia), Zanzibar, the Mexican Pacific, the Caribbean and Papua New Guinea (Thacker & Starnes 2003; Flatt et al. 2005; Ridley et al. 2005a; Steindler et al. 2005; Thacker et al. 2007). Others have reported cyanobacterial symbionts that resemble these clades in additional regions (Larkum et al. 1987; Hinde et al. 1994; Diaz 1996).

Ecology of Cyanosponges

Wilkinson & Vacelet (1979) reported that some cyanosponges grew more rapidly in areas receiving sunlight than other sponges, and it has been demonstrated that photosynthates in the form of glycerol and organic phosphate are transferred from cyanobacterial symbionts to sponge hosts (Wilkinson 1979), supplying up to 50% of the sponge’s energy budget and 80% of its carbon budget (Wilkinson 1983, 1987b; Borowitzka et al. 1989; Cheshire et al. 1997). Wilkinson and colleagues (Wilkinson 1983, 1987a, 1992; Wilkinson & Trott 1985) classified two types of cyanosponges. ‘Phototrophs’ have a characteristic flattened morphology with a large surface area presented for photosynthesis to take place, and the majority of their energy requirements are met by photosynthesis (Wilkinson 1983, 1987b). Cyanosponges having a smaller surface area to volume ratio and those which rely on heterotrophic feeding for more than half of their energy requirements are referred to as ‘mixotrophic’. Phototrophic sponges include Phyllospongia lamellosa (Esper, 1794), which obtains 80% of its daily carbon requirements from its photosynthetic symbionts (Cheshire et al. 1997), and Lamellodysidea herbacea (Keller, 1989), which is unable to survive without the contribution from its cyanobacterial symbionts (Borowitzka & Hinde 1999).

As phototrophic sponges are dependent on photosynthesis to meet the bulk of their energy requirements it follows that they are obligate symbioses, and mixotrophic sponges are often assumed to be facultative, although this may not necessarily be true. But studies of the photophysiology of cyanosponges are rare, and many sponges are classified as phototrophic based on morphology alone. An example of the confusion this may cause is the encrusting sponge Chondrilla australiensis Carter, 1873. The flattened morphology and high concentrations of cyanobacteria throughout the tissues of thin individuals may lead one to think the species is phototrophic. However, individuals of C. australiensis that largely lack cyanobacteria occur in dark areas such as caves (Usher et al. 2001) so the species, being facultative, may be classified as mixotrophic. These white C. australiensis are small, thin, and are not reproductive during periods when all other individuals in the area are, suggesting that their reproduction may be critically and adversely affected by the lack of cyanobacterial symbionts. Hence, cyanobacterial symbionts may not be necessary for individual sponge survival, but may be important for the survival of the species. Without careful studies of the photophysiology of each cyanosponge species, it is difficult to assign a category of phototrophic or mixotrophic. In addition, not all sponge species with foliose or thin encrusting morphology have photosynthetic symbionts, so these growth forms are not necessarily adaptations to promote photosynthetic capability.

Equally, it may also be misleading to categorise cyanosponges solely in terms of the relative contribution of energy from photosynthesis compared to filter feeding. Other services that cyanobacterial symbionts can provide may be as important including N2 fixation, chemical defence, provision of a sunscreen and ammonia conversion. Research is needed to understand the role of these factors in sponge survival and reproduction; however, they may partly explain the observation of sponges with low and medium densities of cyanobacterial symbionts in their tissue (Erwin & Thacker 2007). The toxic bioactive compounds that many cyanobacteria produce may deter predators, or act as antibiotics or antifouling chemicals in symbiosis (Borowitzka & Hinde 1999), and several studies have demonstrated they are produced by cyanobacterial symbionts in cyanosponges (Unson & Faulkner 1993; Unson et al. 1994), although heterotrophic bacterial symbionts are responsible in other instances (Bewley et al. 1996). Nitrogen fixation by cyanobacterial symbionts was reported by Wilkinson & Fay (1979), although others have been unable to find evidence of this (Diaz & Ward 1997; Borowitzka & Hinde 1999). More studies investigating this question are needed, and their value would be increased by the use of molecular tools to identify the symbionts in question. Marine Synechococcus, a major sponge symbiont genus, do not fix nitrogen but can utilise ammonium (Glover 1985). However, members of the genus Oscillatoria do fix nitrogen, and these are also important sponge symbionts. Cyanobacteria are effective at screening out UV radiation (Glover 1985), and cyanobacterial symbionts may protect sponges by providing a sunscreen, permitting them to grow in shallow water. There is little evidence that sponges digest cyanobacterial symbionts to utilise them as a food source (Borowitzka & Hinde 1999); however, digestion has been reported in several studies (Vacelet 1971; Berthold et al. 1982) and may be the result of poor health in one of the partners.

The benefits of partnerships with photosynthetic symbionts do not come without a cost to the host. A range of marine invertebrates, including sea anemones (Plantivaux et al. 2004) and corals (Lesser & Farrell 2004), are known to experience dangerous levels of reactive oxygen species generated by photosynthesis during periods of elevated temperature and irradiance, damaging both the host and the symbiont. This phenomenon, called bleaching, has been reported in a number of sponges (Vicente 1990; Fromont & Garson 1999; Cerrano et al. 2001), and evidence for the involvement of photosynthetic symbionts was found for Petrosia ficiformis (Poiret, 1879) (Regoli et al. 2000). This study reported the production of higher levels of antioxidants in P. ficiformis to counteract free radicals produced by Aphanocapsa feldmannii, with antioxidant concentrations increasing in the warmer summer months. However, heterotrophic bacterial symbionts may assist in the removal of reactive oxygen species both by utilising oxygen for respiration, and by producing quenchers and/or scavengers of active oxygen species. Miki et al. (1996) demonstrated that two strains of Flexibacter in the sponge Halichondria japonica (Kadota, 1922) produce the carotenoid zeanthin, a powerful quencher of singlet molecular oxygen and scavenger of organic free radicals. High oxygen levels also inhibit photosynthesis, leading to photorespiration and a subsequent fall in photosynthetic output. Thus, a combination of autotrophic and heterotrophic microbial symbionts may maximise benefits to the host by maximising photosynthetic output while limiting free radical damage.

The benefits of symbiosis to cyanobacteria are less clear, and sponges may simply provide an acceptable environment in which to grow. Protection from UV irradiation is not likely to benefit Synechococcus in symbiosis, as free-living species grow in high light/high UV environments (Furnas & Crosbie 1999). However, sponges provide a solid substrate and access to higher levels of ammonium and phosphorus than occur in the ocean. This is likely to benefit cyanobacteria, and indeed free-living Synechococcus proliferate in coastal areas (Partensky et al. 1999) characterised by higher nutrient concentrations (Furnas & Crosbie 1999). Protection from predation by flagellates and ciliates may be another benefit cyanobacteria gain from being embedded in host tissue. Grazing is a serious problem for free-living cyanobacteria, and grazing rates can be higher than growth rates (Fahnenstiel et al. 1991; Furnas & Crosbie 1999). Nonetheless, being in symbiosis does not always provide protection. Some molluscs are specialist feeders of well-defended sessile invertebrates such as sponges, sequestering and using toxins for their own defence (Debelius 1996). Some of these predators may be feeding specifically on the symbionts, and not on the host. Becerro et al. (2003) demonstrated that the opisthobranch Tylodina sp. is not interested in eating the choanosome of Aplysina aerophoba Nardo, 1843, which lack the choanosome of cyanobacterial symbionts, strongly suggesting that the symbionts are the target of predation, not the sponge tissue itself. Grazing by Tylodina sp. on A.aerophoba in which cyanobacteria had not been removed was restricted to the first 2 mm, the cyanobacteria-rich ectosome. A similar phenomenon has been observed with cowries (Mollusca, Caenogastropoda) consuming the surface tissues of the cyanosponge Chondrilla australiensis (Fig. 2).

Figure 2.

 A cowrie (order Caenogastropoda) consuming the surface tissues of the cyanosponge Chondrilla australiensis, leaving the white sponge tissue exposed.

The ability of cyanobacteria to photosynthesise in very low light enables these symbionts to grow successfully throughout the matrix of some sponges, and allows sponge hosts to utilise a wide range of different light environments. The maximum depth at which cyanosponges occur has received little attention; however, Wilkinson & Vacelet (1979) report the cyanosponge Chondrilla nucula Schmidt, 1862 at 55 m, and we have observed filamentous cyanobacteria in a Geodia sp. collected at 82 m at Ningaloo, Western Australia. Clearly, the clarity of the water will influence the depth at which cyanosponges will grow. Wilkinson & Trott (1985) reported that phototrophic sponges were more prevalent above 40 m on Davies Reef (Great Barrier Reef), where there is more than 5% of surface irradiance, with the greatest abundance occurring at 20 m. A similar study on soft corals containing zooxanthellae, also at Davies Reef, reported similar results with the greatest abundance occurring at 20 m (Fabricius & Klumpp 1995). These corals are generally mixotrophs, but are not able to tolerate high light levels in the way that cyanosponges can, and are thus more restricted in their distribution.

Several studies have demonstrated deleterious effects from shading and siltation on sponges with photosynthetic symbionts. Thacker (2005) found that Lamellodysidea chlorea (de Laubenfels, 1954), host to Oscillatoria spongeliae, lost a significant amount of weight after 2 weeks of shading, whereas Neopetrosia exigua (Kirkpatrick, 1900), which contains ‘Candidatus Synechococcus spongiarum’, did not. This suggests that some cyanosponges are more heavily impacted by shading than others. Roberts et al. (2006) found that shading and silt had a serious impact on the growth and reproduction of Cymbastela concentrica (Lendenfeld, 1887), and numbers of its symbiotic diatoms were significantly reduced. However, added nutrients did not affect the numbers of diatoms. These studies suggest that pollution containing suspended particulate matter such as sewage plumes adversely affects photosynthetic sponges by altering the intensity and spectral quality of light (Roberts et al. 2006).

While it is tempting to speculate that some species of cyanobacterial symbiont may contribute more to host energy requirements than others, no clear patterns emerge from data on the photophysiology of cyanosponges. Oscillatoria spongeliae is found in some phototrophic sponges, such as Lamellodysidea herbacea (Borowitzka & Hinde 1999), Haliclona walentinae (Diaz et al., 2007), Xestospongia bocatorensis (Diaz et al., 2007) (Thacker et al. 2007) and Lamellodysidea chlorea (Thacker 2005). However, ‘Candidatus Synechococcus spongiarum’ is also found in phototrophic sponges such as Aplysina fulva and Neopetrosia subtriangularis (Duchassaing, 1850) (Thacker et al. 2007), and similar symbionts are found in the phototrophic Neofibularia irata Wikinson, 1970, Carteriospongia foliacens (Pallas, 1766) and Phyllospongia papyracea (Esper, 1974) (Wilkinson 1983). Unfortunately, the symbionts involved in older studies were not sequenced, so their phylogenies are uncertain. Thacker et al. (2007) observed significant differences in gross productivity to respiration ratios between cyanosponge species hosting the same symbiont, suggesting that symbiont type may not in itself determine the phototrophic or mixotrophic nature of a sponge. It is likely that the numbers of cyanobacterial symbionts vary both within different areas of individual sponges and within sponge species. The species of host and the environment that it grows in, including light levels, predation, sediment load, and other stressors, are likely to affect the nature of the symbiosis.

Transmission of Cyanobacterial Symbionts

Symbionts may be acquired vertically (symbiont transmitted directly from parent to offspring) or horizontally (offspring acquire symbiont from the environment). The latter strategy carries the risk of not obtaining a suitable symbiont from the environment, but also has the benefit that the host may be able to acquire novel symbionts that are adapted to local conditions. Most hard corals acquire their symbionts horizontally (Schwarz et al. 1999). Only two cyanosponges, the oviparous C. australiensis (Usher et al. 2001, 2005) and the viviparous Diacarnus erythraeanus Kelly-Borges & Vacelet, 1995 (Oren et al. 2005), have been investigated in depth for the mode of cyanobacterial transmission. Both studies found that vertical transmission occurred, and it is interesting to note that both cyanosponges contained the same symbiont clade, ‘Candidatus Synechococcus spongiarum. No members of this clade have yet been found in the water column. Vertical transmission of cyanobacteria in the eggs of C. nucula, P. ficiformis and Halichondria semitubulosa Lieberkühn, 1859 was also indicated in two studies using fluorescence microscopy (Scalera Liaci et al. 1971, 1973).

Vertical transmission has the potential to confer significant competitive advantages to the planktonic gametes and larvae of cyanosponges, increasing longevity in the water column via the provision of photosynthetic energy (Oren et al. 2005; Usher et al. 2005) as has been suggested for chloroplasts in the male gametes of some lower plants (Sears 1980). For similar reasons, symbiont-containing larvae could be expected to gain a competitive advantage during establishment (Wilkinson 1992). Wilkinson & Cheshire (1988) demonstrated that young sponges that have cyanobacterial symbionts grow faster than those that do not, possibly allowing them to compete more effectively.

Usher et al. (2005) demonstrated that cyanobacterial symbionts are actively incorporated into the gametes of C. australiensis, suggesting that host recognition occurs. Positive recognition and incorporation of photosynthetic symbionts has been suggested for other symbioses, e.g. corals (Rodriguez-Lanetty et al. 2004), and anemones (Belda-Baillie et al. 2002). This hypothesis is in contrast to that by Wilkinson (1984) who suggested that the presence of a mucous coating masks microbial symbionts from sponge cells, enabling them to evade sponge immune systems.

Although C. australiensis symbionts are vertically transmitted, not all eggs or sperm receive their full complement, leaving open the possibility that horizontal transmission may occur. A mechanism for ensuring that offspring have access to horizontally transmitted symbionts occurs in corals, whereby mucus containing high levels of dinoflagellates accompanies spawning (Schwarz et al. 1999), and a similar process is possible in sponges. Cyanobacterial symbionts may be expelled along with eggs and sperm during spawning, if not at other times. Taylor et al. (2007) suggests that the evidence to date indicates a combination of vertical transmission of symbionts with occasional horizontal transmission. However, it is not known whether sponges are able to take up their cyanobacterial symbionts if they occur in the water column.

Host Specificity

It is likely that sponge-associated cyanobacteria are derived from multiple symbiotic events over evolutionary time (Steindler et al. 2005). Changes in sea-level, trophic regime, climate, geography and ocean currents have probably contributed to multiple acquisitions and extinctions of autotrophic symbionts, with reef faunas being particularly affected (Cowen 1983). Indeed, the fossil record suggests that Foraminifera evolved symbioses with algae at least four times during the Mesozoic and Cenozoic (Cowen 1983). However, this does not preclude the possibility that some cyanobacteria have remained in symbiosis with sponges continuously.

There is now evidence that particular sponge species host particular cyanobacterial symbionts despite geographic separation or proximity to other sponges containing different symbionts (although this may not always be the case; see Taylor et al. 2005). Ridley et al. (2005a) and Thacker & Starnes (2003) demonstrated that different sponge species always hosted the same distinct strain of Oscillatoriaspongeliae. Usher et al. (2004a) found that the cyanobacterial symbionts of Chondrilla australiensis from widely separated tropical and temperate regions were the same clade, and these associations were stable over time. Sometimes sponge species host more than one species of cyanobacterial symbiont within an individual (Usher et al. 2004a; Ridley et al. 2005b; Steindler et al. 2005), but the reasons for this are not known. Maldonado & Young (1998) demonstrated that different ‘Aphanocapsa feldmannii-like’ symbionts may have different light requirements, and temperature tolerances may also vary. The hosting of two or three different cyanobacterial species may therefore protect the host sponge under stressful conditions, as suggested for corals during bleaching episodes (Rowan et al. 1997), or maximise the photosynthetic potential of cyanobacterial symbionts by their strategic positioning in host tissue. Douglas (1998) hypothesised that vertical transmission allows a host to specialise in the most efficient symbiont by ‘capturing’ the symbiont within the host lineage via transmission to offspring. Vertical transmission thus avoids both the possibility of offspring not coming into contact with an appropriate symbiont, and the acquisition of a less effective symbiont. Unfortunately, without further study it is impossible to know if sponges with more than one species of cyanobacterial symbiont acquire their symbionts horizontally, whereas those with only one type of cyanobacterial symbiont transmit them vertically.

The two major clades of cyanobacterial symbionts are not specific to particular sponge species. ‘Candidatus Synechococcus spongiarum’ and Oscillatoriaspongeliae inhabit a wide range of sponge hosts around the world. These clades comprise closely related symbionts, and within clades 16S rDNA sequences typically differ by less than 3%. Less than 97% identity between 16S rDNA sequences is widely used as a cut-off for bacterial species (Stackebrandt & Goebel 1994). In sponge symbiosis research, 16S rDNA sequencing has been the primary tool for elucidating the phylogeny of bacterial symbionts, resulting in quantum leaps in our understanding of this field. The 16S rRNA gene is well conserved and rarely transferred between species (Ward 2005); however, it is important to note that bacteria with very similar sequences may have different metabolic capabilities due to horizontal gene transfer and mutation (Ward 2005).

In systems where symbionts are vertically transmitted, such as ‘Candidatus Synechococcus spongiarum’ in Australian Chondrilla australiensis (Usher et al. 2005) and Red Sea Diacarnuserythraenus (Oren et al. 2005) it is difficult to understand how the symbionts maintain species identity, given the geographic and reproductive isolation and different habitat. Yet these symbionts share 99% sequence identity and therefore readily qualify as the same species. There are at least three possible explanations for these observations. The first is that dispersal of symbionts occurs continuously over large geographic distances. That is, ‘Candidatus Synechococcus spongiarum’ is present in the water column and distant sponges exchange symbionts on a regular basis. This cyanobacterium has not yet been sequenced from the water column, but it is possible that it occurs in very low concentrations. The second is that 16S rDNA is not reliable in resolving closely related species within this clade (see ecotype argument below). The 93% sequence identity of the two least similar members (Taylor et al. 2007) provides evidence for some speciation within this clade. Finally, the cyanobacterium may only recently have become a symbiont and is slow to mutate, such that the same species occurs in different sponges, i.e. the mutation rate is slower than the rate of periodic selection events.

In their excellent review of microbial symbionts of sponges, Taylor et al. (2007) suggested that molecular evidence leans towards horizontal transmission occurring to some degree. In an intriguingly similar example of this problem, the vertically transmitted cyanobacterial symbionts of ascidians, Prochloron spp., exist as a closely related clade with 16S rDNA sequence similarities of 97–99% despite different host genera and widely separated geographic locations (Münchhoff et al. 2007). Highly iterative palindrome 1 (HIP1), a DNA fingerprinting technique that has been successful at distinguishing within-species differences in cyanobacteria, was also applied with similar results (Münchhoff et al. 2007). The authors concluded that the likeliest explanation was that Prochloron spp. are horizontally transmitted across considerable distances; however, they are yet to be found free-living.

The determination of the ‘true’ number of cyanobacterial symbiont species that occur in sponges remains as elusive as an agreed method for determining the number of bacterial species in environmental samples. Yet there is general agreement that the species concept of independent evolutionary units is useful in bacteriology (Cohen 2002). Two species may be distinguished by having less than 70% DNA–DNA hybridization (Wayne et al. 1987; Murray et al. 1990; Ward et al. 1998); however, this test cannot be applied in mixed bacterial populations such as those in sponges. 16S rDNA sequencing overcomes this problem and is widely used in bacteriology; however, it is thought to have low resolution at the species level (Lan & Reeves 2001). This poses the possibility that closely related cyanobacterial symbionts such as Synechococcus species may operate differently within their sponge hosts. Indeed, different strains of Oscillatoriaspongeliae produce different secondary metabolites and may have different roles in the symbiosis (Ridley et al. 2005a).

To address the problem of defining bacterial species Cohen (2002) proposed the ‘ecotype’, defined as a set of strains using the same or similar ecological niche and which are bound together by environment-specific periodic selection. Different ecotypes are evolutionary lineages that are free to diverge from each other by mutation or acquisition of new genes from other species. Each ecotype is identifiable as a monophyletic sequence cluster; however, geographically isolated ecotypes can diverge into separate sequence clusters. This would be particularly true for sponge symbionts, which are presumed to have low mobility. Unfortunately, the methods proposed to distinguish ecotypes, such as multilocus sequence typing and rep-PCR, rely on the availability of pure cultures, so this approach is unlikely to be useful for sponge research. Nonetheless, based on the ecotype concept it could be argued each sponge species represents a separate ecological niche, as symbionts are subject to different forms of selection in their unique environment and are free to diverge from symbionts in other sponge species. This would be particularly true for symbionts that are vertically transmitted. Perhaps even a species of sponge growing in a temperate area could be regarded as a different ecological niche to the same species growing in a tropical region.

The question as to the true number of species represented by cyanobacterial symbiont clades is important for understanding both the ecology of the symbioses and their evolutionary history. Although different ecotypes or species may have identical or nearly identical 16S rDNA sequences, this method (preferably in combination with sequencing of other housekeeping genes) remains the best approach to understanding sponge cyanobacterial symbiont phylogeny at this time. It is convenient to refer groups with >97% sequence identity as, for example, the ‘Candidatus Synechococcus spongiarum clade’. However, members of this clade may have different ecological functions within their hosts.

Cyanobacterial Species Involved in Sponge Symbioses


The Chroococcales are 0.5–30 μm unicellular cocci or rods that reproduce by binary fission or budding (Waterbury & Rippka 1989). Members of this order are particularly difficult to identify using classical techniques due to their paucity of morphological and cytological features.

Aphanocapsa feldmannii

Aphanocapsa feldmannii is a small (generally about 2 × 1.5 μm) coccoid cyanobacterium with a spiral thylakoid membrane. It was described by Feldmann (1933) from light microscopy observations of P. ficiformis and Ircinia variabilis (Schmidt, 1862), and was considered the most common and wide-spread cyanobacterial symbiont of sponges (Larkum et al. 1988; Wilkinson 1992) until recently. Aphanocapsa feldmannii-type symbionts were reported to occur in the surface tissues of more than 60 sponge species in 13 orders around the world (Diaz 1996), and in 45% of sponge species in Belize and Bermuda (Rützler 1990). However, Usher et al. (2004a) demonstrated that cyanobacterial symbionts of P. ficiformis and I. variabilis belong to the Synechococcus/Prochlorococcus group, and that symbionts resembling A. feldmannii consisted of a number of closely related symbionts (see below). To date A. feldmannii sequences have only been found from the two original sponges.


Synechococcus spp. are the most common cyanobacterial symbionts found in sponges, and although many older studies have not sequenced symbionts, they often found a predominance of small ‘A. feldmannii-like’ cyanobacteria in sponges using microscopy (Vicente 1990; Diaz 1996). Many groups have now sequenced Synechococcus in sponges using 16S rDNA. Diaz (1997) demonstrated that A. feldmannii-like symbionts in the sponge Aplysina archeri (Higgin, 1875) (Caribbean) were Synechococcus species, and Hentschel et al. (2002) reported similar results for the symbionts in Theonella swinhoei Gray, 1868 (Palau) and A. aerophoba (Mediterranean). Gómez et al. (2004) likewise reported that the symbionts of Xestospongia muta (Schmidt, 1870) belong to the Synechococcus group.

Candidatus Synechococcus spongiarum’ was described by Usher et al. (2004b) in C. nucula, and many GenBank sequences (Diaz 1997; Hentschel et al. 2002; Usher et al. 2004a; Steindler et al. 2005; Thacker 2005; Erwin & Thacker 2007; Taylor et al. 2007) belong to this group, making it the largest sponge-specific clade of bacterial symbionts known (Taylor et al. 2007). Erwin & Thacker (2007) found ‘Candidatus Synechococcus spongiarum’ in 85% of cyanosponges in the Caribbean using molecular techniques. Sequences currently available are from the sponges Aplysinaarcheri, A. aerophoba, Chondrilla nucula, C. australiensis, Ircinia variabilis, I. felix, Diacarnus erythraeanus, Stelletta kallitetilla, S. pudica, Cribrochalina vasculum, Svenzea zeai, Spheciospongia florida, Pseudoaxinella tubulosa, Petrosia sp., Theonella swinhoei, T. conica, Discodermia dissoluta, Carteriospongia foliascens, Xestospongiaexigua, X. muta, Xestospongia proxima, Antho chartacea, Neopetrosia exigua, Aplysina cauliformis, Pseudoceratina arabica, A. fulva, Smenospongia aurea, Verongula gigantea, V. rigida, and Aplysina fistularis. Further sponge species known to host these cyanobacteria will come to light as research is conducted, and we have recently sequenced it from another five sponges from tropical and temperate Western Australia (M.L. Lemloh, J. Fromont, F. Brümer, K. Usher, unpublished data).

Candidatus Synechococcus spongiarum’ are small coccoid cyanobacteria with a spiral thylakoid membrane around their perimeter (Fig. 3A) and are generally 0.6–2 μm in diameter. Synechococcus species have very similar morphology to each other, however, A. feldmannii and ‘Candidatus Synechococcus spongiarum’ can be distinguished by size and subtle ultrastructural differences. The size of both A. feldmannii and ‘Candidatus Synechococcus spongiarum’ increases with the number of turns of the thylakoid (Usher et al. 2006), and the latter increases with decreasing light exposure.

Figure 3.

 A: ‘Candidatus Synechococcus spongiarum’ in Chondrilla australiensis, Western Australia. Scale bar = 500 nm. B: Aphanocapsa raspaigellae from Ircinia variabilis, France. Scale bar = 1 μm.

Aphanocapsa raspaigellae

Aphanocapsa raspaigellae was named by Feldmann (1933) as a symbiont of I. variabilis using light microscopy observations. TEM later revealed a spherical cyanobacterium 5–12 μm in diameter containing parallel thylakoids (non-spiral) around the periphery of the cell and large vacuoles (Fig. 3B). It has been reported in several sponges including Terpios hoshinotaRützler & Muzik, 1993 where it constitutes 50% of the sponge tissue, and possibly assists its sponge host to overgrow and kill corals (Rützler & Muzik 1993).

The phylogeny of A. raspaigellae is uncertain. Lewin (1975) noted that A. raspaigellae cells are similar in size and shape to Synechocystis trididemni Lafargue & Duclaux, 1979 and Prochloron spp., and Rützler (1981) make the observation that A. raspaigellae and Prochloron spp. are very similar in ultrastructure, despite their different photosynthetic pigments. Wilkinson (1992) also noted that this symbiont closely resembles Synechocystis from ascidians. The phylogeny of A. raspaigellae has not been determined using molecular analyses to date. It is now known that Prochloron spp. and S. trididemni are closely related (see below), and that photosynthetic pigments can not be used for classification. It seems likely that A. raspaigellae belongs to the Synechocystis group as it contains phycobiliproteins and does not have chlorophyll b.


Molecular studies recently demonstrated that the Prochlorales are cyanobacteria (Lewin 2002), and the three prochlorophyte genera, Prochlorococcus, Prochloron and Prochlorothrix, are only distantly related (Urbach et al. 1992; Turner 1997; Litvaitis 2002). All are unicellular, and most have chlorophyll a and b, and little or no phycobiliproteins. The free-living Prochlorococcus are the smallest cyanobacteria (0.5 μm) while the symbiotic Prochloron may be as large as 12 μm.


Prochloron spp. are symbionts of many colonial ascidians in the family Didemnidae, providing UV protection and photosynthates, but varying in size and morphology (Münchhoff et al. 2007). Rützler (1981) reported large coccoid cyanobacteria resembling Prochloron spp. and A. raspaigellae in Dictyonella (Ulosa) funicularis and Dictyonella arenosa (Caribbean Sea), and molecular studies of the symbionts of Theonella swinhoei (Japan) and Lendenfeldia dendyi (Zanzibar) indicate that these sponge symbionts are closely related to Prochloron (Steindler et al. 2005). Prochloron spp. have also been reported on the surface of two sponges on the Great Barrier Reef (Parry 1986). No free-living Prochloron spp. have been found, and they have not been successfully grown in culture (Münchhoff et al. 2007).

Synechocystis trididemni

Synechocystis trididemni has phycobiliproteins and no chlorophyll b, and is also known from ascidian symbioses. Cox et al. (1985) reported S. trididemni in the sponges Prianos aff. melanos de Laubenfels, 1954, Spirastrella aff.decumbens Ridley, 1884 and an unidentified sponge, on the Great Barrier Reef, Australia. There were consistent morphological differences between these symbionts, suggesting they were different species. They contained phycobiliproteins and Cox noted that they had a ‘striking resemblance’ to Prochloron spp. Molecular studies have indicated that ascidian symbionts Prochloron spp. and S. trididemni are closely related and arose from a common cyanobacterial ancestor, despite major differences in their photosynthetic pigments (Shimada et al. 2003). Ridley et al. (2005b) used 16S rDNA sequencing to demonstrate the presence of S. trididemni in Fasciospongia chondrodes (de Laubenfels, 1954) in Palau, and we have recently sequenced it from a sponge in temperate Australia (M.L. Lemloh, J. Fromont, F. Brümer, K. Usher, unpublished data).


The order Oscillatoriales includes all undifferentiated filamentous cyanobacteria.

Oscillatoria spongeliae

Oscillatoria spongeliae (previously known as Phormidium spongeliae) has been reported over a wide geographic range (Wilkinson 1992), with the best known example occurring in the sponge Lamellodysidea herbacea. At least 10 common sponges host this symbiont in the Indo-Pacific region, one in the Mediterranean and four in the tropical western Atlantic (Diaz et al. 2007). Molecular studies have demonstrated that it is a species complex (Thacker & Starnes 2003), with members distinguishable by their ultrastructure (Larkum et al. 1988). This filamentous symbiont varies in length from 7 to 50 cells, and cells are 7–10 μm in diameter and 4 μm long (Larkum et al. 1987). It contains phycobiliproteins (Larkum et al. 1987) and produces powerful polychlorinated compounds which are likely to act as a defence against predators (Unson & Faulkner 1993; Flowers et al. 1998; Ridley et al. 2005a). However, different strains of O. spongeliae produce different secondary metabolites (Ridley et al. 2005a). While it is not known if O. spongeliae is vertically transmitted, Thacker et al. (2007) report that each species of host harbours a unique 16S rDNA ribotype of the symbiont, and this, together with evidence for coevolution between symbionts and hosts (Thacker & Starnes 2003), suggests that vertical transmission is likely, although some degree of lineage jumping also occurs (Ridley et al. 2005a).

16S rDNA sequences currently available on GenBank show that O. spongeliae occurs in the sponges Dysidea cfr. herbacea 1B smooth form, Dysidea cfr. herbacea 1A ridged form, Dysidea cfr. granulosa, Lamellodysidea (Dysidea) herbacea, Lamellodysidea chlorea, Phyllospongia papyracea, Fasciospongiachondrodes, Lendenfeldia dendyi, Aplysina gerardogreeni, Haliclonawalentinae, Xestospongiabocatorensis, and Hyrtios violaceus.

Unknown species

A number of un-named coccoid cyanobacterial symbionts of sponges have been sequenced using 16S rDNA. The 4-μm symbiont of Mycale hentscheli (Webb & Maas 2002) is closely related to that from Cymbastelaconcentrica (Taylor et al. 2005), with Cyanobacterium stanieri the nearest relative. A 1 μm symbiont of Cymbastela marshae Hooper & Bergquist, 1992 had a non-spiral thylakoid and was covered with dense fimbriae (Fig. 4) (Usher et al. 2004a). This symbiont and a Synechococcus sp. from Haliclona sp. (Australia) occur in high densities throughout their hosts and the associations are stable over time. They are the only examples to date of these particular symbioses, and the symbionts may be host-specific. A cyanobacterium related to Leptolyngbya was sequenced from Rhopaloeides odorabile Thompson et al., 1997 in the Great Barrier Reef (Webster & Hill 2001) but apparently was not observed by TEM, so may not be a symbiont.

Figure 4.

 An unknown species of cyanobacterial symbiont in Cymbastela marshae, Western Australia, showing dense fimbriae surrounding the cell. Scale bar = 200 nm.


A plethora of diverse cyanosponges exists and any researcher who chooses to study them can head to the nearest ocean and will very likely find examples. These sponges serve important roles in the ecology of reefs as primary producers and in nutrient cycling. Cyanosponges are often among the most abundant sponge species on coastal reefs (Diaz 1996), and provide food and habitat for a wide range of organisms. As symbionts, cyanobacteria give their hosts important advantages that zooxanthellae do not provide. The ability of cyanobacteria to synthesise a sunscreen, and to photosynthesise in very low light by capturing a much larger range of the light spectrum, enables cyanosponges to grow in a wide range of environments, from intertidal zones where they are exposed to full sun, to relatively dark overhangs and caves. Cyanobacteria also have a wider temperature tolerance than zooxanthellae, and unlike hard corals, cyanosponges are common in both tropical and temperate regions. These abilities make cyanobacteria ideal symbionts, permitting cyanosponges to be highly adaptable and widely spread both globally and within regions.

Many cyanosponges await investigation, and it is likely that more unknown species of cyanobacteria will be discovered. It appears that there is variation in the abundance and diversity of cyanosponges at the local scale; however, it is not yet clear why this is so, and whether different regions of the world also vary in these respects. The difficulty in culturing both sponge hosts and cyanobacterial symbionts has slowed progress in understanding the symbioses, and research into all aspects of cyanosponges including biochemical and physiological aspects is needed. Unfortunately, the confusion that exists around the nomenclature of cyanobacteria in general hampers the efforts of those trying to gain an understanding of phylogeny and ecology of sponge symbionts. There is a real need for a complete overhaul of the old divisions based on classical taxonomy within the phylum Cyanobacteria.

Research into the mode of transmission of different associations and the potential for sponges to acquire symbionts from the water column would be particularly useful for understanding cyanosponge ecology and to determine if episodes of horizontal transmission are possible. Equally, it is important to conduct controlled experiments to determine the susceptibility of cyanosponges to global warming. Some cyanobacterial/sponge associations may be more resistant than others to increased temperature, and it is important to understand how symbiont type, host species and environment affect susceptibility to bleaching. Cyanosponges may become increasingly dominant on tropical reefs where coral populations suffer from bleaching and higher ocean acidity. Perhaps the ancient competition between hard corals and cyanosponges will once again swing in favour of sponges, with cyanosponges increasingly assisting the demise of corals by overgrowing and killing them.

The great evolutionary age of cyanosponges and the ecologically significant advantages gained by these symbioses has likely driven the evolution of cyanosponges in unpredictable ways. They have already challenged several widely held hypotheses in the world of symbiosis research, demonstrating that uniparental transmission of symbionts and the possession of only one type of symbiont with a given metabolic pathway are not always the rule. One thing seems certain; these ancient symbioses will continue to surprise and intrigue us.