The ontogeny of the shells in modern and ancient radiolarian species is poorly understood, although we are gaining insight into the dynamic role that the cytoplasm plays in depositing silica and determining the elaborate geometry of the product (e.g., Anderson 1983, Anderson and Swanberg 1981). A more complete analysis of shell ontogeny including observations of the kind presented here may help to clarify phylogenetic relationships based on developmental stages in addition to current evidence that is largely based on morphology of mature shells observed in different geological strata. The data presented here suggest that in collosphaerid radiolaria, at least, shells may develop either by simultaneous silicification of central capsules after all of them have divided (simultaneous deposition, Anderson and Swanberg 1981), or by delayed budding and formation of central capsules surrounded by siliceous shells after some of the central capsules have fully matured (successive deposition). The evidence for simultaneous division of the central capsules and skeletal deposition is well documented for some colonial radiolaria based on light microscopy (e.g., Anderson 1983, Anderson and Swanberg 1981). Moreover, nuclei of some colonial species divide by binary fission in advance of reproductive stages (e.g., Anderson 1983, p. 160) thus providing numerous nuclei in the intracapsular cytoplasm that can serve as reproductive nuclei. These may become the nuclei of swarmers or could serve as nuclei for bud formation.
Binary fission in some species of radiolaria has been reported in living individuals beginning in the nineteenth century (e.g., Hertwig 1879, Borgert 1909, 1922, Hollande et al. 1953) and discussed in the general context of interpreting radiolarian reproductive strategies. Haeckel (1887, p. xcix) noted that while the general course of individual development begins with swarmer production, there exisits in some groups a different form of ontogeny, introduced by simple division of the unicellular organism, and coming under the term "regeneration" in its wider sense. "This spontaneous division occurs quite commonly in the Polycyttaria and produces their colonies." "In all these cases the increase by division is nothing else than an ordinary case of cell-division, in which bisection of the nucleus precedes that of the central capsule." During culturing studies in our laboratories, colonies of Collozoum inerme (a species lacking skeletal elements) collected alive frequently contain central capsules that occur as binary pairs or appear elongated with two central deposits of reserve material suggesting a late stage of binary division. Central capsules of shell-bearing species are known to undergo binary fission during early stages of colony development (e.g., Figures 1 and 2).
We have only indirect evidence, however, for budding of mature central capsules in shelled species. The clearest evidence is from fossil and modern skeletons showing the bud-like attachment of a smaller shell to a larger shell (i.e., Figures 6 and 7). Although bud-like bodies have been observed in living colonial radiolaria (e.g, Figures 3, 4, and 5), we have not found the earliest stages where the bud is just protruding from the central capsule. It would be very important to perform fine structural analyses of these bud like stages to determine if they are indeed products of binary fission, to complete the suite of data supporting budding of mature central capsules, and to document the cytological events accompanying budding.
Examples of binary shells have been reported for a wide range of shelled colonial and solitary radiolaria. Kling (1971) presented photographs of paired shells with varying degrees of separation and an interconnected skeletal framework that appeared to be "stretched" suggesting deposition during binary fission, but the species were not identified. Takahashi(1981) presented scanning electron microscopic evidence of paired shells in living radiolaria, collected in sediment traps, many with narrow connecting skeletal frameworks. These included colonial radiolaria: Acrosphaera murrayana (plate 1), Disolenia collina (plate 3), Disolenia zanguebarica (plate 3), and Siphonosphaera socialis (plate 4), and quite interestingly phaeodaria: Castanidium abundiplanatum (plate 58). Takahashi (1981, plate 58) notes in the figure legend "Two specimens splitting apart" suggesting that the phaeodarian was undergoing binary fission. Haeckel (1887, p. xcix) commented that "Among the Phaeodaria division is commonly observed in the order Phaeocystina, and also in the Phaeoconchia." "In all these cases the increase by division is nothing else than an ordinary case of cell-division, in which bisection of the nucleus precedes that of the central capsule." Examples of apparent division in fossil skeletons of solitary spumellaria have been reported by deWever (1985) who examined Paleozoic radiolaria, and Sugiyama et al. (1992) who reported Pliocene fossil skeletons of Spongaster tetras with a bud-like smaller skeleton clearly attached to the surface (plate 7). During several years of culturing solitary radiolaria in the laboratory we have observed only a few instances where additional individuals of the same species were observed in the culture vessels that originally contained one mature radiolarian. We cannot say if these came from binary fission of the mature individual or were adult stages of immature individuals originally included with the culture water. We also suspect that laboratory culture conditions are not always optimal for some species, though many individuals grow and mature to release swarmers. However, it is not uncommon to observe closely-joined binary pairs of shells (usually one shell slightly smaller than the other) connected by strands of cytoplasm in living colonies (as in Figure 5) .
All of this points to a need for better documentation of evidence for binary division in radiolaria from fossil and living material. It would be particularly advantageous if micropaleontologists and biostratigraphers could document each time they observe apparent budding in fossil shells of radiolaria including the geological time markers for the strata where they were found. Likewise, as additional researchers examine plankton samples and living radiolaria in culture, any evidence of binary fission would be especially helpful to further document its occurrence in extant species. Until we have more substantial evidence about the number of species that undergo binary fission, and the frequency of occurrence in a population, it will be difficult to make inferences about its significance for radiolarian ecology and population dynamics.
Moreover, if we have a well-documented record of occurrences such as binary shell formation from fossil and living material we can be more confident about making inferences from biological observations from the present to the past. There are few variables in the fossil record that correlate well with biological data from living individuals since only the hard parts typically are fossilized. The occurrence of fossil binary shells, and evidence of skeletal morphogenesis in shells that are incompletely formed, provide one line of evidence linking observations of living individuals with those in the past (e.g., Anderson and Swanberg 1981, Thurow and Anderson 1986, Anderson et al., 1988, Amon et al. 1990). Micropaleontolgical observations of variations in skeletal geometry may provide some interesting clues to skeletal morphogenesis linking modern and ancient species.
The abnormal, asymmetrical pair of shells (Figures 6 and 7) from the fossil record and the sequence of views for a living Acrosphaera colony (Figures 3-5) support the possibility of successive deposition at least in some collosphaerid colonial radiolaria. It should be noted, however, that the rather convincing evidence of binary fission in Figures 6 and 7 occurred in Acrosphaera cyrtodon. The living colony is a different species of Acrosphaera. We do not know to what extent successive division may occur in different species of Acrosphaera. Paired shells observed in other colonoial radiolaria vary in geometry from two hemispherical shells fused at a common mid point to pairs of shells connected only by spines or an isthmus (e.g., Takahashi 1981, Paverd 1995). These observations conform to either incomplete binary fission (paired hemispheres attached due to incomplete division of the central capsule before skeletogenesis) or incomplete successive deposition of shells by delayed budding (united pairs of complete shells, e.g., Figure 6). The information presented here for fossil and living preparations is the first time these data have been synthesized into an explanation of binary fission during successive deposition of shells around central capsules. Moreover, if successive deposition occurs with daughter cells produced after maturation of the central capsule and deposition of the surrounding shell, this would alter our current understanding of reproductive cycles in radiolaria. According to our current understanding (e.g., Anderson 1983), the life cycle of radiolaria includes a single mature stage with intervening production of swarmers. The swarmers are released by the mature radiolarian and disperse in the environment. Subsequent maturation of the swarmers (either as asexual propagules or after fusion, if they are sexual gametes) yields a central capsule that becomes surrounded by a skeleton. The mature individual lives and feeds for a while and then releases swarmers, fully dissipating the cells, to produce the next generation, thus completing the life cycle. However, if there are further stages of binary fission within each generation in some radiolaria (as suggested previously in the literature and reported further here), this would permit additional stages of asexual proliferation of the radiolaria without swarmer production.
The occurrence of asexual reproductive stages (in addition to swarmer production) within the life cycle of some radiolaria may provide greater plasticity in responding to changing environmental variables including food sources, water chemistry, physical structure of the water mass, temperature, etc. If proliferation can occur by binary fission, without swarmer production, these radiolaria can respond rapidly to favorable environments and increase population densities substantially without going through a more perilous intervening stage of reproduction by flagellated swarmers. Most species have a central capsular wall composed of numerous individual organic plates that fit against each other much like pieces of a jigsaw puzzle. Since the wall is not solid, there is the possibility of expansion and protrusion of central capsular cytoplasm to produce a bud by disarticulation of the plates at the site of budding.
Further work is needed with additional living species to determine if there is similar evidence of binary fission of mature stages of individuals as reported here. Moreover, if it is limited to colonial radiolaria, it at least helps to explain how colonies can become so very prolific (up to several hundred per cubic meter) in modern oceans (e.g., Swanberg 1979, Anderson 1983). Under favorable conditions, mature colonies of shelled species may be able to generate additional central capsules, thus increasing the number of colonies and their total biomass in oceanic planktonic communities. The fossil evidence and data from extant species presented here add additional evidence that central capsules can divide by binary fission and that this may be a long established reproductive mode especially among collosphaerid radiolaria. The data support previously published findings (Anderson and Swanberg 1981), and suggest that a successive mode of division, not reported previously, may be more common, in colonial radiolaria at least, than previously realized.