This is the first report of the propagation of the toxic dinoflagellate Dinophysis fortii Pavill. under laboratory conditions when fed on the marine ciliate Myrionecta rubra grown with the cryptophyte Teleaulax amphioxeia (W. Conrad) D. R. A. Hill. In contrast, reduced growth of D. fortii (max. of 3–4 divisions) and formation of small cells were observed in the absence of the ciliate or when provided with T. amphioxeia only as prey, showing that D. fortii cannot utilize T. amphioxeia as prey. In the TEM observation of D. fortii cells, which had fully fed on the ciliate prey, well‐developed chloroplasts (5–12 μm in length) were seen and three thylakoids were usually arranged in most of the chloroplasts observed, but chloroplasts having two thylakoids were sometimes confirmed. In cells starved for 4 weeks, decrease of chloroplast numbers and disappearance of large chloroplasts were observed, and only a few small chloroplasts (0.5–2 μm in length) remained in the marginal regions. In the observation of the sequestration process of the chloroplasts ingested from M. rubra by D. fortii, within 15 min after D. fortii captured M. rubra, incorporation of almost all of the chloroplasts was observed, while most of the other cell contents still remained in the M. rubra cell. After that, dispersion of the ingested chloroplasts toward the marginal regions was confirmed, suggesting that chloroplasts of M. rubra are ingested and dispersed in D. fortii cells in advance of the ingestion of the other cell contents to prevent them from being digested in food vacuoles. The ingested chloroplasts can also function as kleptoplastids.

Kleptoplastidy is the retention of plastids obtained from ingested algal prey, which may remain temporarily functional and be used for photosynthesis by the predator. We showed that the marine dinoflagellate Dinophysis mitra has great kleptoplastid diversity. We obtained 308 plastid rbc L sequences by gene cloning from 14 D. mitra cells and 102 operational taxonomic units (OTUs). Most sequences were new in the genetic database and positioned within Haptophyceae (227 sequences [73.7%], 80 OTUs [78.4%]), particularly within the genus Chrysochromulina . Others were closely related to Prasinophyceae (16 sequences [5.2%], 5 OTUs [4.9%]), Dictyochophyceae (14 sequences [4.5%], 5 OTUs [4.9%]), Pelagophyceae (14 sequences [4.5%], 1 OTU [1.0%]), Bolidophyceae (3 sequences [1.0%], 1 OTU [1.0%]), and Bacillariophyceae (1 sequence [0.3%], 1 OTU [1.0%]); however, 33 sequences (10.8%) as 9 OTUs (8.8%) were not closely clustered with any particular group. Only six sequences were identical to those of Chrysochromulina simplex , Chrysochromulina hirta , Chrysochromulina sp. TKB8936, Micromonas pusilla NEPCC29, Micromonas pusilla CCMP491, and an unidentified diatom. Thus, we detected >100 different plastid sequences from 14 D. mitra cells, strongly suggesting kleptoplastidy and the need for mixotrophic prey such as Laboea , Tontonia , and Strombidium -like ciliates, which retain numerous symbiotic plastids from different origins, for propagation and plastid sequestration.