For coral sampling, sustainable practices were followed, with small fragments of Corallium rubrum collected from larger, unharmed colonies, under permits issued by French authorities. At the coral culture facility of the Centre Scientifique de Monaco, aquariums are continuously supplied with seawater from the Mediterranean Sea, which is further filtered and maintained at controlled temperature of 15°C. C. rubrum colonies were maintained in darkness and fed daily with frozen plankton mixture (Ocean Nutrition). Two primary culturing methods for C. rubrum are employed: branches suspended on strings, and colonies growing laterally on glass slides. The preparation of C. rubrum colonies growing laterally on glass coverslips was previously described in (Le Goff et al., 2017). For time-lapse Video, a total of four laterally grown colonies, each about 2 cm², were monitored for different period of time ( Table 1 ).

Close-up photographs of laterally grown colonies were taken at 20-minute intervals using a Canon 5 DS (sensor: surface area 864 mm², 50.6 megapixels, 4.1µm pixel pitch) equipped with a Laowa 24mm probe lens (2:1 macro, f/40 aperture, waterproof, full-frame). Lighting was provided by a Flash Godox MS300 transmitted via four 0.8mm optical fibers necessary to compensate for the f/40 aperture.

The camera was set to ISO 100 and a 1/200 shutter speed, synchronized with an X2T-C radio transmitter, and controlled by a TC-80N3 remote for 20-minute intervals shooting. It was powered by an AC-E6N adapter with a daily reset at 12:01 AM to ensure continuous 24-hour operation.

A custom culture chamber for the coral colony was designed using Rhinoceros 8 and printed on an Elegoo Saturn 3D printer. Its design ensured stability during time-lapse acquisition, facilitated fine adjustments of the focal distance, and allowed removal, maintenance, and repositioning of the colony.

Photos were saved on a 512GB SDXC Extreme Pro UHS-1 card and processed in Adobe Lightroom (RAW to JPG formatting). The time-lapse videos were assembled and compressed in Adobe Premiere Pro and exported as MP4 files. See Supplementary Figure 1 for photographs of the system.

In order to investigate anatomy, a branch of C. rubrum was fixed overnight at 4°C in chilled artificial seawater/paraformaldehyde (PAF) fixation buffer [425 mM NaCl, 9 mM KCl, 9.3 mM CaCl₂, 25.5 mM MgSO₄, 23 mM MgCl₂, 2 mM NaHCO₃, 100 mM HEPES, pH 7.9, 1% Triton X-100, 4.5% PAF]. The sample was then decalcified in [100 mM HEPES, pH 7.9, 500 mM NaCl, 250 mM EDTA, pH 8.0, 0.4% PAF, 0.1% Tween 20] at 4°C until complete skeleton dissolution (5 days). After three rinses in Milli-Q water, the sample was post-fixed in ice-cold methanol for 3 hours. The sample was then gradually rehydrated with increasing dilutions of water/methanol with a final Milli-Q water bath. Next, the sample was immersed in a 2% potassium hydroxide (KOH) solution for 2 hours, then rinsed in water and buffered with [200 mM Tris, pH 7.8]. Finally, the sample was carefully incubated in 5% hydrogen peroxide (H₂O₂) for 1 hour and photographed with a Nikon SMZ1270 trinocular equipped with a DS-Fi3 camera.

In order to investigate histology using paraffin-embedded sections, a branch of C. rubrum was fixed and decalcified as above. Then, the sample was dehydrated with increasing concentrations of ethanol and embedded in paraffin. Seven- micrometer sections were cut on a Leica RM 2155 microtome and mounted on silane-coated glass slides. Paraffin was removed with baths of xylene substitute and further rehydrated in several decreasing concentrations of ethanol with a final Milli-Q water bath. Sections were post-fixed in Bouin’s solution for 1 hour at 56°C and rinsed in running tap water for 5-10 minutes to remove the yellow color. Slides were first stained in Weigert’s iron hematoxylin working solution for 10 minutes, rinsed under running water, then stained in Biebrich scarlet-acid fuchsin solution for 10-15 minutes, then rinsed and differentiated in phosphomolybdic-phosphotungstic acid solution for 10-15 minutes. Sections were transferred directly (without rinse) into aniline blue solution for 5-10 minutes, then rinsed briefly and differentiated in 1% acetic acid solution for 2-5 minutes. After a final rinsing step, slides were quickly dehydrated in ethyl alcohol, and mounted with Canada balsam resinous mounting medium. Samples were then imaged with a Leica microscope equipped with a CCD camera.

To observe very slow biological processes on visible time scales, we assembled a custom-made video imaging system capable of tracking growing C. rubrum colonies over extended periods of time (weeks to months, see Supplementary Figure 1 and the materials and methods section for a detailed description). It is important to emphasize to the reader, that in order to fully appreciate the implications from this study, one must refer to the Supplementary Video Files (listed Table 1 ) which illustrate the data that is discussed throughout the manuscript.

We captured the growth and development of a primary polyp, at the approximate age of two months old ( Figure 2A ; Supplementary Video 1 ). Notably, this primary polyp clearly exhibited the dimorphism of both an autozooid and a siphonozooid. By documenting this rarely observed event of the settled primary polyp, we were able to compare a side-by-side contrast of a primary polyp expressing only one siphonozooid and autozooid to a colony expressing multiple copies of this dimorphism ( Figure 2B ; Supplementary Video 2 ). Our imaging system also enabled documentation of simultaneous (coordinated) contractions and expansions of autozooids and siphonozooids. This behavior likely helps to maintain hydrostatic pressure within the colony’s water canal system.

A striking feature of C. rubrum is its intense red coloration due to a high concentration of carotenoid pigments in its biomineralized structures (Cvejic et al., 2007). However, this coloration complicates the observation of the anatomical arrangement of the C. rubrum branches. When examining the external features of a branch of live C. rubrum, the external arrangement of the polyp dimorphism is readily apparent with interspersed siphonozooids and autozooids spread across the tissue ( Figures 3A, B ). To better visualize the internal structure, we developed a fixation-depigmentation protocol (see materials and methods). We observed the large primary canals, located deeper in the tissue, which run parallel to the longitudinal axis of the branch, while smaller secondary canals, closer to the surface, connect to the primary canals. Also, this is the first clear visualization of siphonozooids and autozooids connecting to the canal system ( Figures 3C, D ).

Histological sections of C. rubrum stained with Masson’s trichrome revealed the detailed arrangement of siphonozooids, autozooids and the canals ( Figure 4 ). Under microscopic examination, the retracted autozooids are distinguishable from siphonozooids, both of which connect to the canal network. A magnified view of the siphonozooid shows the asymmetrical ciliated siphonoglyph ( Figure 4B ) while the larger autozooid reveals a developing oocyte ( Figure 4C ) (see discussion section).

To date, work on Octocorallia has relied heavily on field observations and sampling, which provide an individual snap-shot of time with respect to the biology of these corals. We made a number of novel observations by imaging C. rubrum horizontal growth over extended periods of time in the laboratory. Firstly, we observed that damage or removal of the autozooid stalk including tentacles leads to regeneration [one of the biological hallmarks of cnidarians (Child, 1903; Henry and Hart, 2005)] of the complete autozooid within 2 weeks ( Supplementary Video 3 ). During our extensive surveying of coral colonies over time, we did not observe that damaging or removing the autozooid stalks would stimulate de novo polyp synthesis in the surrounding area. Instead repair and regeneration of the damaged autozooid would occur following injury.

By tracking the same C. rubrum colony for 11 months, we were able to observe its growth, including lateral expansion through the formation of new tissue at the margins, and the development of new polyps ( Supplementary Video 4 ). Notably, these laterally growing colonies also produce a calcified skeleton between themself and the glass (Le Goff et al., 2017), implying that lateral growth reproduces the typical growth pattern (i.e. calcification and tissue expansion) of a wild colony. Growth of the expanding C. rubrum colony over this extended period of observation can be visualized in the supplementary videos, which reveals the slow rate of expansion. During recording, we gathered novel information about connections within the polyp dimorphism. In particular, we demonstrated that a number of the siphonozooids are precursors to autozooids, as this transformation was observed multiple times over several months ( Figure 5 ; Supplementary Videos 5 , 6 ). Importantly, we also observed that new siphonozooids form in the colony and that not all siphonozooids that form will transform into autozooids ( Figure 5 ; Supplementary Videos 5 , 7 ). Furthermore, in some cases, autozooid formation can occur continuously without pausing at the siphonozooid stage ( Supplementary Video 8 ). Thus, in C. rubrum, the siphonozooid is the precursor developmental stage to the autozooid. We were not able to determine whether all siphonozooids eventually transform into autozooids. Interestingly, our data also show that, contrary to the two-week period for autozooid regeneration ( Supplementary Video 3 ), siphonozooid-to-autozooid differentiation takes approximately four weeks to complete ( Supplementary Video 6 ). If the formation of a novel siphonozooid also takes about four weeks to complete ( Supplementary Video 7 ), the formation of an autozooid without pausing at the siphonozooid stage required two months to complete ( Supplementary Video 8 ). This suggests that, in addition to potential differences in the energetic requirements for de novo synthesis, the longer time-frame for differentiation favors autozooid repair following damage, to restore the colony’s functional physiology in the shortest time possible.

Our observations demonstrate that polyps were dimorphic in C. rubrum, with both autozooids and siphonozooids, as observed in other Coralliidae (McFadden et al., 2022), including Pacific precious corals (Kishinouye, 1904; Nonaka et al., 2012). Time-lapse observations have allowed us, for the first time, to witness the transformation of certain siphonozooids into autozooids in C. rubrum. Further studies are needed to determine whether the same processes occur in other dimorphic octocorals. However, based on molecular phylogeny, amongst the closest species to Corallium rubrum is Corallium japonicum (Ardila et al., 2012; Uda et al., 2013). It is therefore highly likely that similar processes occurs in closely related species, including the Japanese species C. japonicum, Pleurocorallium elatius, and Pleurocorallium konojoi. In C. rubrum, autozooids and siphonozooids show clear differences in their internal and external morphology, as well as in their roles in reproduction, feeding and fluid circulation. The fact that siphonozooids remain in this developmental stage for extended periods and that not all siphonozooids transform into autozooids supports the idea that siphonozooids represent an arrested stage of polyp development. However, we were unable to determine the exact proportion of siphonozooids that eventually transform into autozooids. Based on our data, and those from other precious coral species (Nonaka et al., 2015, 2012), it appears that the number of siphonozooids generally exceeds the number of autozooids. Given that individual autozooids have a lifespan of about a decade (Benedetti et al., 2020; Vielzeuf et al., 2008), it would be valuable to estimate the ratio of siphonozooids to autozooids in different parts of a colony, such as intensive and slow growth zones (such as branch apices versus branch columns), to better understand the dynamics of old autozooid replacement by transforming siphonozooids. By replacing individual modules (i.e., polyps), precious corals can delay senescence and achieve lifespans of over 1000 years (Roark et al., 2009).

While blastogenesis appears as the primary mechanism for new polyp formation in C. rubrum, the cellular and molecular underpinnings of this process are unknown across octocorallians. In contrast, studies on other cnidarians, such as hexacorallians and hydrozoans, have pointed to the recruitment and differentiation of Adult Stem Cells in tissue regeneration (Gierer et al., 1972; Gold and Jacobs, 2013). Furthermore, recent research on the scleractinian coral Stylophora pistillata suggested that similar cellular mechanisms may underlie polyps formation (Levanoni et al., 2024). Our ability to monitor polyp dimorphism differentiation in C. rubrum offers a new perspective on the studying of the role of blastogenesis in zooid formation, perpetuating the longevity of these coral colonies.

The classic evolutionary paradigm held that anthozoans exhibited radial symmetry. However, detailed analyses of their gastrovascular cavity have revealed evidence for bilateral symmetry (Berking, 2007; Boero et al., 2005; Finnerty, 2005; Malakhov, 2016; Martindale et al., 2002). In the gastrovascular cavity of octocorals, the orientation of the mesenteries (internal septa or walls) relative to their retractor muscle, along with the relative positioning of the siphonoglyph, defines sulcal and asulcal chambers (analogous to ventral and dorsal chambers), which establish the bilateral symmetry axis (Bayer, 1974; Beklemishev, 1969; Hyman, 1923; Sekida et al., 2016). In Hickson’s original description (1883) of various octocorallians, the siphonoglyph was identified as a ciliated groove on the ventral side of the stomodaeum (gastrovascular cavity), that extends long, strong cilia. He observed that while it is evenly present in all polyps of monomorphic species, the siphonoglyph was more strongly marked in siphonozooids than in autozooids of dimorphic species and may even be absent in some genera like Paragorgia, a non-precious Coralliidae (Hickson, 1883).

Our histological examination of C. rubrum also clearly demonstrates a non-radial symmetry with respect to the siphonoglyph’s location in the siphonozooid (see Figure 4 ). In C. japonicum, Sekida and colleagues found that the mesenteries orientation relative to the siphonoglyph was reversed in siphonozooids compared to what was described in autozooids, suggesting this as a distinguishing marker between the two polyp types (Sekida et al., 2016). On the other hand, there is no clear evidence for a siphonoglyph in either the autozooids of C. rubrum or other Corallium species suggesting a substantial reduction of the ciliate groove during the transition from siphonozooids to autozooids in Coralliidae. While we could not examine the finer details of its orientation in the gastrovascular cavity, the distinct asymmetry of the siphonoglyph’s localization observed in this study, as well as in other representative red corals such as Primnoa pacifica (Waller et al., 2014), provides compelling evidence that the siphonoglyph serves as a hallmark of bilateral symmetry in the gastrovascular cavity of Coralliidae polyps. This is true even though the siphonoglyph, marked during early development in siphonozooids, may be substantially reduced or entirely absent later in autozooids.

The strong development of the ciliated siphonoglyph in siphonozooids compared to autozooids has led to the proposal that siphonozooids are specialized for water propulsion, while autozooids are primarily involved in feeding and reproduction (Bayer, 1974; Beklemishev, 1969; Claus, 1884; Hickson, 1883; Hyman, 1923; Moseley, 1876). Our observations ( Supplementary Video 1 ) reveal that even in young C. rubrum primary polyps, a siphonozooid is positioned near an autozooid, suggesting that coordination between these zooid types is essential for coral physiology. Furthermore, the synchronized opening and closing of both siphonozooids and autozooids in colonial forms ( Supplementary Video 2 ) indicates coordination between these zooid types, suggesting that both respond to common environmental cues. However, the extent to which the relative proportion of each zooid type and their respective roles in fluid control and feeding influence the regulation of fluid and food dissemination throughout the colony remains to be determined.

Understanding the role of polyp dimorphism is also crucial because it directly influences reproductive strategies. In most gonochoristic octocorals with polyp dimorphism, gametes develop either in the autozooids or, more rarely, in siphonozooids (Hyman, 1923; McFadden et al., 2022). For the Mediterranean red coral, the development of gonads has been confirmed to occur in the feeding polyp (i.e. autozooid, Figure 4 ) (Santangelo et al., 2003; Tsounis et al., 2006; Torrents and Garrabou, 2011), contrary to Pacific precious corals, where siphonozooids host the gonads (Nonaka et al., 2012, 2015; Sekida et al., 2016). In addition, it was observed that in rare cases, some gametes were also observed in the autozooids of Japanese precious coral species (Nonaka et al., 2015). In light of our findings that siphonozooids can transform into autozooids in C. rubrum and potentially other precious corals, the observed gamete development in the autozooid of Japanese corals can now be attributed to a limited number of siphonozooids transforming into autozooids while still developing gonads (gamete viability was not assessed in Nonaka et al. (2015). Interestingly, this difference is associated with distinct reproductive strategies: the Mediterranean red coral is larval brooder, while Pacific species are broadcast spawners (Nonaka et al., 2015). This suggest that Pacific precious corals may have developed an adaptive trait involving siphonozooid-based reproduction ( Figure 6 ). This division aligns with the idea that brooding, being more energetically demanding, benefits from the autozooid’s specialization in prey capture, which provides essential nutrients for developing larvae. Conversely, because siphonozooids outnumber autozooids, the total number of gametes produced by siphonozooids may be higher broadcast spawners, assuming an equivalent number of gametes per zooid. A critical question that arises from these observations is whether the mesenteries of siphonozooids that transform into autozooids are capable of germline maturation, or if this function is restricted solely to direct autozooid development. It remains undocumented whether siphonozooids in C. rubrum can develop gametes, a potential phenomenon that could have been overlooked in past research.

Our current finding that siphonozooids are precursors to autozooids raises new questions about the tissue characteristics of each polyp types and the physiological and/or environmental factors that drive the divergent reproductive strategies between Pacific and Mediterranean precious corals. Given that one polyp derives from the other, it remains unclear at which stage -siphonozooid or autozooid- the germline is first established. If siphonozooids represent developmentally arrested polyps awaiting the resumption of their development to autozooids, we must ask to what extent this somatic developmental control influences germline progression (formation/migration/maturation). Investigating these developmental processes through gene expression analysis and cellular tracking would help clarify this issue. Moreover, future studies combined with feeding manipulations could shed light on how nutrient availability impacts polyp dimorphism and reproductive success. In other words, understanding the intricate relationship between zooid development and gonad maturation remains a major focus for future coral research.