J. Cell Sci. 86, 273-286 (1986) 273 Printed in Great Britain © The Company of Biologists Limited 1986

ALGAL ENDOSYMBIOSIS IN BROWN HYDRA: HOST/SYMBIONT SPECIFICITY

M. RAHAT AND V. REICH Department of Zoology, The Hebrew University of Jerusalem, Jerusalem 91904, Israel

SUMMARY Host/symbiont specificity has been investigated in non-symbiotic and aposymbiotic brown and green hydra infected with various free-living and symbiotic species and strains of Chlorella and Chlorococcum. Morphology and infrastructure of the symbioses obtained have been compared. Aposymbiotic Swiss Hydra viridis and Japanese H. magnipapillata served as controls. In two strains of H. attenuata stable hereditary symbioses were obtained with Chlorococcum isolated from H. magnipapillata. In one strain of H. vulgaris, in H, oligactis and in aposymbiotic H. viridis chlorococci persisted for more than a week. Eight species of free-living Chlorococcum, 10 symbiotic and 10 free-living strains of Chlorella disappeared from the brown hydra within 1-2 days. In H. magnipapillata there was a graded distribution of chlorococci along the polyps. In hypostomal cells there were >30 algae/cell while in endodermal cells of the mid-section or peduncle <10 algae/cell were found. In H. attenuata the algal distribution was irregular, there were up to five chlorocci/cell, and up to 20 cells/hydra hosted algae. In the dark most cells of Chlorococcum disappeared from H. magnipapillata and aposymbiotic hydra were obtained. Chlorococcum is thus an obligate phototroph, and host-dependent hetero- trophy is not required for the preservation of a symbiosis. The few chlorococci that survived in the dark seem to belong to a less-demanding physiological strain. In variance with known ChlorellaJH. viridis endosymbioses the chlorococci in H. magni- papillata and H. attenuata were tightly enveloped in the vacuolar membrane of the hosting cells with no visible perialgal space. Chlorococcum reproduced in these vacuoles and up to eight daughter cells were found within the same vacuole. We suggest that the graded or scant distribution of chlorococci in the various brown hydra, their inability to live in H. viridis and the inability of the various chlorellae to live in brown hydra are the result of differences in nutrients available to the algae in the respective hosting cells. We conclude that host/symbiont specificity and the various forms of interrelations we show in green and brown hydra with chlorococci and chlorellae are based on nutritional-ecological factors. These interrelations demonstrate successive stages in the evolution of stable obligatoric symbioses from chance encounters of preadapted potential cosymbionts.

INTRODUCTION Two major groups have been recognized in the genus Hydra. One group comprises the green 'viridissima' hydra that host endosymbiotic algae in their digestive cells, and the second group are the 'brown' non-symbiotic hydra such as H. vulgaris or H. attenuata that were known to be found in nature without algal endosymbionts (Campbell, 1983). Goetsch (1924, 1926, cited by Kanaev, 1952), claimed that he found in his laboratory H. attenuata infected with a free-living 'Chlorella magna', and that he

Key words: algae, Chlorella, Chlorococcum, Hydra, symbiosis. 274 M. Rahat and V. Reich could infect these hydra with other chlorellae. Such infected brown hydra have not been reported since, and later attempts to repeat these experiments and infect brown hydra with chlorellae have failed (Pardy & Muscatine, 1973; Muscatine et al. 1975; Jolley & Smith, 1980; Rahat, 1985a). To date, only Chlorella spp., of symbiotic or free-living origin, have been reported to form stable symbioses in the green hydra (Park et al. 1967; Muscatine et al. 1975; Jolley & Smith, 1980; Rahat & Reich, 1984, 1985a). It has been claimed for the brown hydra that they cannot phagocytose green algae or host them to form stable symbioses (Pardy & Muscatine, 1973; Jolley & Smith, 1980; Rahat, 1985a). The Japanese//, magnipapillata has been classified with the 'vulgaris' group of the non-symbiotic brown hydra (Campbell, 1983), but has recently been shown by us to host in its endodermal cells unicellular green algae of the genus Chlorococcum (Rahat & Reich, 19856). Extensive data have been acquired on the Chlorella/H. viridis endosymbioses (e.g. see Parker al 1967; Pool, 1979; Jolley & Smith, 1980; Meints & Pardy, 1980; McNeil ef al. 1981; McAuley & Smith, 1982; Rahat, 19856; Rahat & Reich, 1985a). It is thus of special interest to compare the specificity and interrelations of the 'old' and 'new' cosymbionts. We describe here the infection of green and brown aposymbiotic and non- symbiotic hydra with chlorellae and chlorococci, various degrees of prolonged persistence of algae in some aposymbiotic and non-symbiotic species of hydra and the consequent formation of stable algal endosymbioses in H. attenuata. We describe in detail some of the host/symbiont nutritional and ultrastructural spatial inter- relations.

MATERIALS AND METHODS Solutions used M solution: a buffered salt solution resembling pond water (Lenhoff & Brown, 1970), used for growing hydra. We added Phenol Red to this medium to facilitate the visual monitoring of the pH in this medium (Rahat & Reich, 1983a). BBM+: Bolds Basal Medium (Bischoff & Bold, 1963), with an addition of organic nutrients (Rahat & Reich, 1985a), used to grow the algae in vitro. Antibiotics mixture: lOO^gml"1 of each of penicillin, streptomycin, neomycin and rifampicin (Rahat & Reich, 19836), used to obtain axenic hydra.

Hydra used The Japanese H. magnipapillata (Sugiyama& Fujisawa, 1977; Sugiyama, 1983; Rahat & Reich, 19856), a South African and an Australian strain of H. attenuata were used together with several strains of European hydra. The latter were H. oligactis, two strains of H. vulgaris and apo- symbiotic hydra of a Swiss strain of H. viridis (Rahat et al. 1979). The European and Australian hydra were originally obtained from Professor P. Tardent, Institute of Zoology, University of Zurich, Switzerland, and have been cultured in our laboratory for several years. Three strains of green H. viridis were used: Swiss, European and a Jerusalem strain isolated here from an aquarium. The hydra were grown in M solution at 20 (±2)°C, under continuous illumination of 2500 lux (6500lux for H. magnipapillata), and fed three times a week with freshly hatched larvae of Artemia sp. Algal endosymbiosis in brown hydra 275

Preparation of aposymbiotic H. magnipapillata One single bud was cut from a dark-grown polyp and verified by light and ultraviolet microscopy to be aposymbiotic. This bud and its descendants were fed six times a week to repletion. Within 3 weeks a clone of about 100 aposymbiotic H. magnipapillata was obtained.

Isolation of Chlorococcum for infection About 50 green H. magnipapillata were incubated for 48 h in a mixture of antibiotics to render them bacteria-free (Rahat & Reich, 19836). The hydra were then washed out of the antibiotics and placed separately into sterile test tubes containing 5 ml of BBM+ solution, one hydra in each test tube. In this medium Chlorococcum slowly 'leaked' out of the hydra to the bottom of the test tube. After a few days the algae were collected with a pipette, washed by repeated centrifugation and resuspended in M solution. These algae were used to infect the various species of hydra. Similarly isolated Chlorococcum were used to initiate in vitro cultures of these algae, to be described separately. In later experiments, in vitro cultured Chlorococcum were successfully used for infections.

Infection of hydra with Chlorococcum and Chlorella Larvae of Artemia spp. were used as a vector to infect the various hydra, respectively, with symbiotic Chlorococcum isolated from H. magnipapillata, with eight species of free-living Chlorococcum, with 16 strains of in vitro cultured symbiotic and non-symbiotic chlorellae (Rahat & Reich, 1985a), and with symbiotic chlorellae freshly isolated by homogenization of three strains of green H. viridis, i.e. Swiss, European and Jerusalem strains. Four- to five-day-old larvae were mixed in M solution with a suspension of the respective algae, and incubated for 20—30 min. When the gut of the Artemia turned green, they were fed to the hydra. The algae were thus phagocytosed together with food particles. Compared to in vitro cultured Chlorococcum, chlorococci separated from freshly homogenized H. magnipapillata gave less satisfactory infections, as nematocytes present in such preparations prevented the Artemia from ingesting many algae. The infected hydra were fed two to three times a week. Examination of the hydra for infecting algae was best done after several days of starvation.

Light and electron microscopy For microscopic examination of their cells, hydra were macerated according to David (1973). Photomicrography (Figs 1, 2) was done by Nomarski Differential Interference Microscopy on Technical Pan film. For electron microscopy (Figs 3, 4, 5), hydra were starved for 8 days to eliminate disturbing lipid globules from their tissue. The hydra were then fixed for 1 h in a solution containing 4% glutaraldehyde in 0-1 M-cacodylate buffer at pH 7-4, rinsed in the same buffer and postfixed for 1 h in 1 % OsO^ and 0-1 M-cacodylate buffer, rinsed again and left overnight in cacodylate buffer at 4°C. The hydra were then stained for 30 min in the dark with 2 % uranyl acetate, rinsed in distilled water, dehydrated through a graded ethanol series and embedded in epoxy resin (Spuir, 1969). Some sections were stained again with uranyl acetate.

RESULTS Infection experiments As a control for all infection experiments, aposymbiotic H. viridis and H. magni- papillata were reinfected, respectively, with Chlorella and Chlorococcum originally isolated from these hydra, and algal persistence was verified. Chlorellae freshly isolated from three strains of green H. viridis and 16 strains of chlorellae grown in vitro in our laboratory disappeared from H. magnipapillata and the other brown hydra within 1-2 days, although eight of the strains grown in vitiv form stable symbioses with H. viridis (Rahat & Reich, 1985a). 276 M. Rahat and V. Reich

Chlorococci from H. magnipapillata formed stable symbioses with two strains of H. attenuata that retained these algae for almost a year, at the time of writing this paper, and passed the algae to buds through many generations. In one strain of H. vulgaris, in H. oligactis and in the Swiss strain of aposymbiotic H. viridis chlorococci could be found for up to 6 days after infection. No persistent infections were obtained with any of the free-living species of Chlorococcum. Table 1 sum- marizes these results.

Identification of Chlorococcum Chlorococcum has been isolated from H. magnipapillata and grown in vitro (to be described in a separate publication). Comparison of microscopic morphology of the isolated Chlorococcum with eight species of free-living Chlorococcum generously supplied by the Cambridge Culture Centre for Algae and Protozoa, confirmed the endosymbiotic algae to be a Chlorococcum sp. (Starr, 1955; Bold & Parker, 1962).

Populations of Chlorococcum in H. magnipapillata and H. attenuata In given batches of hydra simultaneously infected with Chlorococcum, some hydra retained more algae than others, and in some very few or no algae were found. H. magnipapillata grown at the light intensity used (6500 lux) were green to the naked eye, and the hypostome could be seen to be dark green (Fig. 1). At lower

Table 1. Infection, persistence of algae and formation of stable symbioses in brown and green hydra with free-living and symbiotic Chlorella and Chlorococcum Infecting algae Infected Symbiotic Free-living Symbiotic Free-living hydra* Chlorella} Chlorella% Chlorococcum^ Chlorococcum^ H. magnipapillata (Jap) — - + + — H. attenuata (Afr) — - + — H. attenuata (Aus) — — + — H. vulgaris (Eur) — — +— — H. vulgaris (Eur) — — — — H. oligactis (Eur) — — +— — H. viridis (Eur) ++ — +— —

( —) All infecting algae disappear within 3 days; ( + —) some algae persist in cells of hydra for more than 6 days; (+) symbiosis, infecting algae reproducing and passed on to buds; (+ + ) symbiosis, hydra green to the naked eye. •Geographical origins of the various hydra are shown in parentheses: Jap, Japan; Afr, South Africa; Aus, Australia; Eur, Europe. f Eight free-living strains of in vitro cultured Chlorella that form stable symbiosis with Swiss aposymbiotic H. viridis (Rahat & Reich, 1985a), and symbiotic chlorellae freshly isolated from three strains of this hydra. \ Nine free-living strains of Chlorella that do not persist in aposymbiotic Swiss H. viridis (Rahat & Reich, 1985a). §/« vitro cultured Chlorococcum spp. isolated from//, magnipapillata. \ Eight species of free-living Chlorococcum obtained from the Cambridge Algal Collection (see Materials and Methods). Algal endosymbiosis in brown hydra 111

•A,

Abbreviations used for all figures: c, chloroplast; cc, Chlorococcum, cm, hydra-cell membrane; cw, algal cell wall; ect, ectoderm; end, endoderm; gl, gland cell of hydra; mil, mitochondria of hydra; n, nucleus; st, starch grains. Fig. 1. Photomicrograph of: A, thick cross-section from upper third of H. magni- papillata showing Chlorococcum spp. in the endoderm. Starch grains in algae are seen as white dots. B. Chlorococcum in cells of macerated H. magnipapillata; left, cells from hypostome of hydra packed with many chlorococci; right, cells from lower part of hydra hosting a few algae only. Scale: 10 Jim between bars. 278 M. Rahat and V. Reich light intensities the number of algae/hydra and number of 'green' polyps in the population decreased. If grown in the dark for prolonged periods most algae were lost from the hydra, but in some hydra a few algae remained. When the hydra were returned to intense light these algae reproduced and the hydra greened again. From H. magmpaptllata that lost all their symbiotic algae in the dark we obtained an aposymbiotic strain. Some polyps of the Australian strain of H. attenuata turned completely green after infection with symbiotic chlorococci, but others were pale to the naked eye although they hosted many single or clusters of Chlowcoccum for at least a year at this writing (Fig. 2). Judged from colour changes in the symbiotic H. attenuata the

Fig. 2. Photomicrograph of thick mid-section and cells of H. attenuata showing Chlorococcum spp. in endodermal cells. Starch grains are seen as white dots in algae. Scale: 10^m between bars. Algal endosymbiosis in brown hydra 279 number of chlorococci/hydra varied in time although growth conditions were constant.

Distribution of Chlorococcum in tissue and cells Like Chlorella in H. viridis, the endosymbiotic algae in H. magnipapillata and H. attenuata were hosted in the endodermal cells, but were located both distal and proximal to the cell nucleus (Figs 1,2). In H. magnipapillata there was a distinct quantitatively graded distribution of Chlorococcum along the polyp. Most algae were located in the hypostomal cells, there were less in the mid-section and very few in the peduncle and tentacles. The quantitative differential distribution of the endosymbiotic algae was reflected in the number of algae in cells from the upper and lower parts of the hydra. In hypostomal cells there were >30 algae/cell while in some endodermal cells from the peduncle there were <10 or none at all (Fig. 1). In H. magnipapillata bisected transversely, the number of algae increased in the tip regenerating the new hypostome. In the polyps and gastroderm of H. attenuata distribution of chlorococci was irregular, and algae could be found both proximally and distally to the cell's nucleus (Fig- 2). An average number of Chlorococcum /cell or hydra cannot be given because of the great variation (between 0 and 30) in number of these algae in the digestive cells of the hydra, and in different polyps in a population.

The vacuolar habitat The host/symbiont spatial interrelations in H. magnipapillata and H. attenuata were similar. Symbiotic chlorococci were closely enveloped in the vacuolar mem- brane of the host cell with no visible perialgal space (Figs 3, 4). In H. viridis similarly prepared for electron microscopy as a control, there was such a space (see also Rahat & Reich, 1984). Chlorococcum reproduced inside the vacuoles of the hydra by cell fission (Figs 3, 4), and up to eight daughter cells were sometimes found within the same vacuole. The cell wall of the symbiotic Chlorococcum was rather thick compared to that of endo- symbiotic chlorellae found in H. viridis (Rahat & Reich, 1984), showed a 'wavy' rim and sometimes a thickening of the cell wall (Fig. 5).

DISCUSSION What is it that enables some algae to form symbioses in hydra while others do not, and why do some hydra host algae while others do not? To examine these questions critically we have to define our terms and summarize known data. We should distinguish between: (1) infection of hydra by algae that are rejected or digested within 1—2 days. (2) Persistence of algae in the hydra that can last for more than a week. (3) Stable hereditary algal/hydra symbioses. In the latter case the algae reproduce in the cells of the hydra and are passed on to subsequent cell Fig. 3. Electron micrograph of Chlorococcum spp. in endoderm of: A, H. magnipapillata (X4600); B, H. attenuata (XS200). Note dividing algal cells. Algal endosymbiosis in brx/wn hydra

Fig. 4. Electron micrograph of: A, Chlorococcum spp. dividing in cell of//, attenuata (X8000); B, detail of A (x 17 000). 212 M. Rahat and V. Reich and polyp generations. Host/symbiont coevolution may then lead from a facultative to an obligate endosymbiosis. Our data from present and former studies show that: (1) all hydra we examined phagocytosed algae together with their prey. (2) Symbiotic and some free-living

cm

Fig. 5. Electron micrograph of: A, Chlorococcum spp. in cell of H. magnipapillata (X13 500); B, detail of algal cell wall and hydra cell membrane (X53 000). Algal endosymbiosis in brown hydra 283 chlorellae can form symbioses with various strains of H. viridis (Rahat & Reich, 1985a), but they do not persist in brown symbiotic or non-symbiotic hydra. (3) The Japanese brown H. magnipapillata hosts symbiotic Chlorococcum (Rahat & Reich, 19856). (4) Symbiotic Chlorococcum, but not free-living chlorococci, persist for prolonged periods in cells of green and brown hydra and some may form symbioses in 'non-symbiotic' brown hydra (Table 1). (5). In contrast to symbiotic chlorellae the symbiotic Chlorococcum is an obligate phototroph. It might survive in the dark but will not reproduce there. In the light it will also grow in vitro. Whatever alleged identity algae might have in order to be recognized as endo- symbionts (Muscatine et al. 1975; Muscatine, 1982), it is certainly contained in algae freshly isolated from H. viridis or H. magnipapillata. These algae however do not form reciprocal symbioses. In H. viridis symbiotic Chlorococcum could survive for several days, while chlorellae disappeared from H. magnipapillata within 1-2 days. There must thus be some factors other than an algal 'identity' recognized by hydra that determine their ability to form symbioses. Execution of the claimed barriers to colonization by algae of the hydra cell, i.e. refusal to take up certain algae and rejection of algae taken up, or their digestion, are accomplished by the hydra cells within 1-2 days. Phagocytosed algae that persist in the cells of hydra for more than a few days have passed the above barriers. Such successful algae, however, may later disappear from their host due to their inability to reproduce in the intravacuolar environment or to compete with the reproduction rate of the host cell (Rahat, 1985a,b). We have recently shown correlations between algal growth requirements and their ability to form symbioses in hydra (Rahat & Reich, 1985a). We interpret the present respective data (Table 1), to show that the vacuolar environments in H. viridis and H. magnipapillata cannot satisfy the nutritional requirements of the algae that cannot form reciprocal symbioses. A histochemical study of nutrients present in the respective cells of hydra might explain the different numbers of chlorococci they host (Figs 1,2). We conclude that host/symbiont specificity in hydra is based on ecological—nutritional factors (Rahat & Reich, 1986). In H. viridis chlorellae are hosted in spacious vacuoles (Muscatine et al. 1975; Rahat & Reich, 1984, 1985a). The close contact of Chlorococcum with the host-cell membrane and cytoplasm in H. magnipapillata and H. attenuata (Figs 3, 4, 5) indicates different host/symbiont inter-relations. Any interchange of nutrients would be direct, without the mediation of a perialgal vacuolar medium. The conspicuous thick cell wall found in the symbiotic Chlorococcum (Figs 4, 5) might be correlated with its ability to live as an endosymbiont. However, some other species of this genus that we have examined have a thick cell wall (Bold & Parker, 1962), but did not form symbioses in our hydra. We tried to investigate the intravacuolar environment in hydra by correlating growth in vivo with that in vitro (Rahat & Reich, 1985a). We assumed that whatever is required for growth in vitro is available to the algae growing symbiotically in vivo. When aposymbiotic H. viridis are used for such a study, there is always the possibility that the hydra might be infected by an unwanted contaminating strain of 284 M. Rahat and V. Reich chlorellae (Rahat & Reich, 1984). Furthermore, the morphology of many strains or species of Chlorella that grow both in vivo and in vitro is almost identical and some cannot be told apart when inside a hosting cell (Rahat, 1985a). For such a study chlorococci should be far superior to chlorellae. Of the nine species examined by us, only Chlorococcum from H. magnipapillata has been found to infect hydra, and its morphology clearly differs from other chlorococci or chlorellae. A detailed study of in vitro growth requirements of Chlorococcum will thus unmistakenly inform us about the nutritional environment in the hydra cell vacuole. Green Chlorococcum-hostingH. magnipapillata were collected by Dr T. Sugiyama at two sites in Japan (personal communication). The symbiotic Chlorococcum grows in vitro and is probably also free-living at the collection sites. This Chlorococcum, formed in our laboratory, symbioses with Australian and African strains of H. attenuata, but hydra infected with this alga have not been reported outside Japan. The symbiotic Chlorococcum might thus be endemic to Japan and the 'failure of opportunity' should be considered a factor in the distribution of some algae/hydra symbioses. From the inability of symbiotic chlorococci to grow in the dark we may conclude that they do not obtain from their host cells all the required nutrients and that the supply by the host of such nutrients is not essential for the formation or maintenance of a stable symbiosis. As in any other population, variation and mutations also occur in algal populations (Hochberg et al. 1972). The few algae that are retained in H. magnipapillata even after prolonged growth in the dark must be a less-demanding variant of the Chlorococcum populations. The presence of such variants among chlorococci from H. magnipapillata is shown by the different numbers of algae in infected hydra and of infected hydra in a population. Similarly, during our attempts to culture in vitro Chlorococcum freshly isolated from H. magnipapillata, better growth was obtained on subsequent transfers, showing a selection of variants more adapted to growth in vitro. The latter variant could still reinfect H. magnipapillata. The 'new' symbioses we describe in the present study may not survive competition in nature (Rahat, 1985a), but they show that it is not 'recognition' that enables algae to be hosted in hydra, but rather a preadapted ability to live in the intravacuolar environment. Persistence in the hydra of a preadapted variant from a given infecting population of algae, and the selection of algae suitable for coevolution towards the formation of a stable symbiosis, are then the means by which chance encounters evolve into stable symbioses. Evolutionary steps that might lead from a 'free' life to an obligatory endosymbiotic existence are demonstrated by the following: (1) some free-living strains and species of chlorellae and chlorococci cannot grow in hydra. (2) Some chlorellae, and Chlorococcum from H. magnipapillata, grow both in vitro and in vivo. (3) Chlorococcum in H. magnipapillata being an obligate phototroph, does not get all the required nutrients from its host. (4) Some native chlorellae from H. viridis also persist in the dark, thus getting all the required nutrients from their host, and cannot be grown in vitro. Algal endosymbiosis in brown hydra 285

On the basis of our data and to answer the questions we posed in this discussion we propose the 'Test Tube Hypothesis': algae ingested into the vacuole of a hydra cell are subjected to a selection similar to that occurring in any abiotic habitat in nature or in medium in a test tube. Algae that are preadapted to live and reproduce in the given environment or medium will survive. If their rate of reproduction equals or surpasses that of the hosting cell they will remain as successful colonizers, exposed to mutual selective coevolution towards the formation of a stable endosymbiosis.

We thank Mr E. Hatab for his assistance with electron microscopy.

REFERENCES BISCHOFF, H. W. & BOLD, H. C. (1963). Phycological Studies, vol. 4. University of Texas publication no. 6318. BOLD, H. C. & PARKER, B. C. (1962). Some supplementary attributes in the classification of Chlorococcum species. Arch. Mikrobiol. 42, 267-288. CAMPBELL, R. D. (1983). Identifying hydra species. In Hydra: Research Methods (ed. H. M. Lehoff), pp.19-28. New York: Plenum Press. DAVID, C. N. (1973). A quantitative method for the maceration of hydra tissue. Wilhelm Roux Arch. EntwMech. Org. 171, 259-268. HOCHBERG, A., PIMSTEIN, R. & RAHAT, M. (1972)'. Properties of an ethionine-resistant (ER) mutant of Ochromonas danica.J. Protozool. 19, 66-69. JOLLEY, E. & SMITH, D. C. (1980). The green hydra symbiosis. II. The biology of the establishment of the association. Proc. R. Soc. Land. B 207, 311-333. KANAEV, I. I. (1952). Hydra: Essays on the Biology of Fresh Water Polyps (translated by E. J. Burrows & H. M. Lenhoff). Edited and published by H. M. Lenhoff, 1969, University of California, Irvine CA 92717. LENHOFF, H. M. & BROWN, R. D. (1970). Mass culture of hydra: An improved method and its application to other aquatic invertebrates. Lab. Anim. 4, 139-154. MCAULEY, P. J. & SMITH, D. C. (1982). The green hydra symbiosis. VII. Conservation of the host cell habitat by the symbiotic algae. Proc. R. Soc. Land. B 216, 415-426. MCNEIL, P. L., HOHMAN, T. & MUSCATINE, L. (1981). Mechanisms of nutritive endocytosis. II. The effect of charged agents on phagocytic recognition of digestive cells, jf. Cell Set. 52, 243-269. MEINTS, R. H. & PARDY, R. L. (1980). Quantitative demonstration of cell surface involvement in a plant-animal symbiosis: Lectin inhibition of reassociation. J. Cell Sci. 43, 239-251. MUSCATINE, L. (1982). Establishment of photosynthetic eukaryotes as endosymbionts in animal cells. In On the Origin of Chloroplasts (ed. J. A. Schiff), pp. 77-92. Amsterdam: Elsevier-North Holland. MUSCATINE, L., COOK, C. B., PARDY, R. L. & POOL, R. R. (1975). Uptake recognition and maintenance of symbiotic chlorellae by Hydra viridis. Symp. Soc. exp. Biol. 29, 175-203. PARDY, R. L. & MUSCATINE, L. (1973). Recognition of symbiotic algae by Hydra virdis. A quantitative study of the uptake of living algae by aposymbiotic H. viridis. Biol. Bull. mar. Lab., Woods Hole 145, 565-579. PARK, H. D., GREENBLATT, C. L., MATTERN, C. F. T. & MERRIL, C. R. (1967). Some relationships between Chlorohydra, its endosymbionts and some other chlorophyllous forms. y. exp. Zool. 164, 141-162. POOL, R. R. (1979). The role of antigenic determinants in the recognition of potential algal symbionts by cells of Chlorohydra. J. Cell Sci. 35, 367-379. RAHAT, M. (1985a). Competition between chlorellae in chimeric infections of Hydra viridis: the evolution of a stable symbiosis. J. Cell Sci. 77, 87-92. RAHAT, M. (19856). Origin and evolution of symbiosis in green hydra. Archs Sci., Geneve 38, 385-399. RAHAT, M. & REICH, V. (1983a). Visual monitoring of pH with phenol-red. In Hydra: Research Methods (ed. H. M. Lenhoff), pp. 35-37. New York: Plenum Press. 286 M. Rahat and V. Reich

RAHAT, M. & REICH, V. (19836). Preparing axenic hydra. In Hydra: Research Methods (ed. H. M. Lenhoff), pp. 79-83. New York: Plenum Press. RAHAT, M. & REICH, V. (1984). Intracellular infection of aposymbiotic Hydra viridis by a free- living Chlorella sp.: Initiation of a stable symbiosis. J. Cell Set. 65, 265-277. RAHAT, M. & REICH, V. (1985a). Correlations between characteristics of some free-living Chlorella sp. and their ability to form stable symbioses with Hydra viridis. J. Cell Sa. 74, 257-266. RAHAT, M. & REICH, V. (19856). A new alga/hydra symbiosis: Hydra magnipapillata of the 'nonsymbiotic' Vulgaris group hosts a Chlorococcum-Yikt alga. Symbiosis 1, 177-184. RAHAT, M. & REICH, V. (1986). Algal/hydra symbiosis: a topic in ecology. (1986). In Environmental Quality and Ecosystem Stability, vol. Ill (ed. Z. Dubinsky). Israel: Bar Ilan University Press. RAHAT, M., ZELDES, D. & REICH, V. (1979). Photoproducts of chloramphenicol: their cytotoxic effects on Hydra viridis and its symbiotic algae. Comp. Biochem. Physiol. 63c, 27-30. SPURR, A. R. (1969). A low viscosity epoxy resin embedding medium for electron microscopy. J.Ultrastruct.Res. 26, 31-43. STARR, R. C. (1955). A comparative study of Chlorococcum meneghini and other spherical zoospore producing genera of the Chlorococcales. Indiana Univ. Publ. Sci. Ser. 20. SUGIYAMA, T. (1983). Isolating hydra mutants by sexual inbreeding. inHydra: Research Methods (ed. H. M. Lenhoff), pp. 211-221. New York: Plenum Press. SUGIYAMA, T. & FUJISAWA, T. (1977). Genetic analysis of developmental mechanisms in hydra. I. Sexual reproduction of Hydra magnipapillata and isolation of mutants. Dev. Growth Differ. 19, 187-200.

(Received 1 May 1986 -Accepted, in revised form, 27 August 1986)