Reproductive Biology of Geodia Species (Porifera, Tetractinellida) From Boreo-Arctic North-Atlantic Deep-Sea Sponge Grounds

Boreo-arctic sponge grounds are essential deep-sea structural habitats that provide important services for the ecosystem. These large sponge aggregations are dominated by demosponges of the genus Geodia (order Tetractinellida, family Geodiidae). However, little is known about the basic biological features of these species, such as their life cycle and dispersal capabilities. Here, we surveyed five deep-sea species of Geodia from the North-Atlantic Ocean and studied their reproductive cycle and strategy using light and electron microscopy. The five species were oviparous and gonochoristic. Synchronous development was observed at individual and population level in most of the species. Oocytes had diameters ranging from 8 μm in previtellogenic stage to 103 μm in vitellogenic stage. At vitellogenic stages, oocytes had high content of lipid yolk entirely acquired by autosynthesis, with no participation of nurse cells. Intense vertical transmission of bacterial symbionts to the oocytes by phagocytosis through pseudopodia was observed, especially in late stages of oogenesis. The density of oocytes within the sponge tissue was on average 10 oocytes/mm2 across all species, higher than that of most temperate and tropical oviparous species studied elsewhere. Spermatic cysts were widespread over the tissue during early stages, or fused in larger cysts, around the canals in later stages, and occupying between 1.5 and 12% of the tissue in males. The reproductive season spanned similar periods for all Geodia spp.: from late spring to early autumn. During the reproductive peak of each species, between 60 and 90% of the population was engaged in reproduction for most species. Given the present hazards that threaten the boreo-arctic tetractinellid sponge grounds, it becomes crucial to understand the processes behind the maintenance and regeneration of populations of keystone deep-sea species in order to predict the magnitude of human impacts and estimate their ability to recover. The information provided in this study will be useful for developing adequate conservation strategies for these vulnerable deep-sea habitats.


INTRODUCTION
Deep-sea sponge grounds are marine habitats first described by Klitgaard et al. (1997) in the Faroe Island shelf and slope areas, and later shown to be widespread across the North-Atlantic Ocean (Klitgaard and Tendal, 2004;Murillo et al., 2012;Beazley et al., 2015). These habitats are formed by three-dimensional structures created by the sponges, which provide refuge for other animals, sometimes of high economic value, and enhance biodiversity (e.g., Klitgaard, 1995;Beazley et al., 2013Beazley et al., , 2015Kenchington et al., 2013). Furthermore, because of their high pumping activity  and consumption of organic matter (Leys et al., 2018), sponges play a major role in nutrient cycling (Maldonado et al., 2012), specifically in carbon turnover (Pile and Young, 2006;Yahel et al., 2007;Cathalot et al., 2015;Kahn et al., 2015), nitrogen flow (Hoffmann et al., 2009;Rooks et al., 2020) and the silicon cycle (Maldonado et al., 2020). Indeed, recently, studies point to serious impacts on ecosystem function when sponges are removed from the deep sea due to fishing bycatch (Pham et al., 2019).
Despite their importance at functional and ecological level (Kutti et al., 2013;Maldonado et al., 2016), deep-sea sponge grounds have been highly exposed to commercial bottom trawling (e.g., Freese, 2001;Roberts, 2002;Pham et al., 2019). In addition, sediment disturbance caused by equipment deployment at the seafloor from various anthropogenic activities, can lead to the clogging of their aquiferous system and eventually to their death . For these reasons, deep-sea sponge grounds in the North Atlantic are now considered as vulnerable marine ecosystems by the OSPAR convention for the protection of the marine environment of the North-East Atlantic (OSPAR, 2008), needing protection according to United Nations General Assembly Resolution 61/105 and Food and Agriculture Organization (UNGA, 2006;FAO, 2009). Up to now, most areas with dense biomass have been protected from bottom trawling in North-West Atlantic by the North-West Atlantic Fisheries Organization (NAFO, 2012(NAFO, , 2014, but there are still areas of deep-sea sponge grounds with high levels of exposure (Thompson and Fuller, 2020). Nowadays, a huge effort is conducted to better understand sponge ground ecosystems' services, to estimate their overall importance, and evaluate the consequences of adverse impacts on these habitats (ICES, 2018;NAFO, 2018).
However, few ecophysiological studies have been conducted on deep-sea sponge species (e.g., Strand et al., 2017) and only two studies regarding their reproductive timing have been published so far. One study focused on the species Thenea abyssorum Koltun, 1959(Koltun, 1959 (order Tetractinellida and suborder Astrophorina), Tentorium semisuberites (Schmidt, 1870), and Radiella sol Schmidt, 1870(Schmidt, 1870, previously called Trichostemma sol (order Polymastiida) from the Barents Sea (Witte, 1996), which had been described among the most abundant demosponges in abyssal Greenland, Iceland, and Norwegian seas by Barthel and Tendal (1993). The other one focused on the species Geodia barretti Bowerbank, 1858 (Bowerbank, 1858;Spetland et al., 2007), which is very abundant in temperate and boreo-arctic grounds of the North-Atlantic Ocean (Klitgaard and Tendal, 2004;Cárdenas et al., 2013). In both publications, researchers found that the studied species were gonochoristic and oviparous with one or two reproductive cycles per year which were dependant on primary production blooms.
Porifera is one of the earliest diverging clades in the phylogeny of the Metazoa (Dohrmann and Wörheide, 2013;Feuda et al., 2017;Zumberge et al., 2018;Laumer et al., 2019). But despite their early origin, sexual reproduction in sponges is not only morphologically similar to that in other Metazoa, but also the molecular toolkit that regulates the gametogenesis is conserved from Porifera to Chordata (Koutsouveli et al., 2020b). Despite the conserved process of reproduction through evolution, even within the phylum Porifera, the reproductive strategy greatly varies among sponge groups in the following traits: e.g., oviparity/viviparity, gonochorism/hermaphrotidism, larval types, number of reproductive cycles per year, and seasonality. These features are mainly constrained by phylogeny (Riesgo et al., 2014), but there is also a high degree of plasticity correlated to the abiotic conditions of their environment. In this sense, sponges lack sensory organs, however, they are able to respond to a variety of abiotic stimuli. One of the main driving factors in latitudes with high seasonal variations is temperature. Many studies reported a correlation of the temperature increase with sex determination, gametogenesis, or induction of spawning, while others have found a triggering of reproductive period after a thermal drop (Chen, 1976;Fromont and Bergquist, 1994;Corriero et al., 1998;Ereskovsky, 2000;Riesgo and Maldonado, 2008). In habitats where temperature is constant all year round, different stimuli have been proposed to play a role, such as photoperiod, wave fluctuation, rainfall, and salinity variations (e.g., Elvin, 1976;Corriero et al., 1998;Calazans and Lanna, 2019). In the deep sea, the reproductive season seems to be dependent on other environmental stimuli such as the primary productivity blooms that occur once or twice over the year (Young and Eckelbarger, 1994;Witte, 1996).
A crucial biological feature for the ecological success of sponges is the larval type and its dispersal potential, which is directly related to the connectivity patterns of these populations and therefore their resilience. All sponge larvae are lecithotrophic since they do not have any digestive tract to process food. So, the major source of energy during larval dispersal is the yolk previously accumulated during gametogenesis and embryogenesis (Maldonado and Bergquist, 2002), with only few exceptions in which larvae were also observed to phagocytize some nutrients from their environment (Ivanova and Semyonov, 1997). Consequently, the dispersal capabilities of sponge propagules are tightly linked to the quality and quantity of the yolk they acquire even before reaching the larva stage. The origin and type of yolk in the female gametes depends on the reproductive strategy of the species (Riesgo et al., 2015): the yolk can (i) be produced by the oocyte itself, through the Golgi apparatus and the rough endoplasmic reticulum (Fell, 1974;Simpson, 1984), (ii) be acquired by direct phagocytosis from the mesohyl (Fell, 1974;Simpson, 1984), (iii) be provided by nurse cells in the surroundings of the oocytes (Fell, 1974;Simpson, 1984), and (iv) be the product of bacterial phagocytosis by the oocyte . The yolk in sponges is mainly composed of lipid, protein, and glycogen (Fell, 1969(Fell, , 1974Simpson, 1984). The amount and the type of yolk in a given oocyte/embryo can give clues on the duration of the gametogenesis/embryogenesis, and the dispersal time of the released propagules. This knowledge has profound ecological implications for the survival and possible recovery of the populations in case of disturbance, and to a greater extent, of the species in the studied habitat.
In this study, fresh material of five out of the six boreoarctic North-Atlantic Geodia species was collected between 2011 and 2019 in the North-East Atlantic (boreal: G. phlegraei, G. macandrewii, G. barretti, G. atlantica; arctic: G. hentscheli).
The main goal was to study the seasonality and the strategy of gametogenesis of these five species through histological and cytological observations. Knowledge on seasonality and reproductive strategies is crucial for developing appropriate conservation strategies for boreo-arctic sponge grounds; a deeper and wider understanding of the reproduction of key sponge ground species could inform us on the population structure and their dispersion pattern, and the capacity of the population to survive and regenerate in case of disturbance.

Sample Collection
One hundred and twenty eight specimens were collected from five Geodia species: G. hentscheli, G. phlegraei, G. barretti, G. macandrewii, and G. atlantica during expeditions in the boreoarctic and North-East Atlantic from the beginning of April to the end of September mainly during the years 2016 and 2017 ( Figure 1, Table 1, and Supplementary Table S1). Samples from Hordaland coast, were collected by Agassiz trawl on board the "K. Bonnevie" research vessel (RV), those from the Kosterfjord were collected using an ROV during the expedition of RV "Nereus," and samples from Korsfjord were collected by triangular dredge on board the RV "Hans Brattstrøm." In addition, samples from the Sula reef, Schulz Bank, and Tromsøflaket (western Barents Sea), were collected either by Agassiz trawl, or ROV (AEGIR 6000) on board of the RV "G.O. Sars" ( Table 1 and  Supplementary Table S1). Finally, samples from Rosemary Bank Seamount were collected by Agassiz trawl on board of MRV "Scotia." All collection details and metadata associated were archived in the PANGAEA data repository 1 (Koutsouveli et al., 2020a). In all expeditions, three cubes of sponge tissue from each specimen were collected (∼5 mm 3 each) and directly fixed in 2.5% glutaraldehyde solution in 0.4M PBS and 0.34M NaCl and stored at 4 • C, until their further processing in the laboratory. The identification of all species was performed based on spicule analyses. Regarding the specimens collected by trawl/dredge, we subsampled tissue only from entire or almost entire collected specimens and not from small pieces of sponges in order to avoid to sample tissue from the same specimen several times.

Sample Processing and Histological Sectioning
For further histological processing, samples were then desilicified by immersing the sponge tissue in 5% hydrofluoric acid overnight and then rinsed profusely with distilled water. After that, the tissue was either processed for light microscopy or Transmission Electron Microscopy (TEM). For light microscopy, tissues were dehydrated through an increasing ethanol series and then embedded in paraffin (60 • C, melting point) overnight after a brief bath in xylene. For TEM, sponge tissues were postfixed in 2% osmium tetroxide in 0.4M PBS for 1 h at 4 • C and then rinsed in distilled water. Similarly, an increasing ethanol series was used FIGURE 1 | Map of North Atlantic indicating the coordinates of the different sampling areas of the study. The Map was retrieved from the ETOPO1 database from NOAA server (Amante and Eakins, 2009) and illustrated in R.
to dehydrate the tissue, and then it was embedded in LRW resin (preparation according to guidelines of the manufacturer).
The protocol for the preparation of the samples was the same as mentioned in Koutsouveli et al. (2020b). Briefly, for light microscopy preparations, thick sections (5 µm) were obtained with an HM 325 rotary microtome (ThermoFisher Scientific) and stained with a standard Harris' Hematoxylin and Eosin (HandE) and Methylene Blue. For TEM, ultrathin sections (approximately 70 nm) were obtained with an Ultracut ultramicrotome (Reichert-Jung) and then stained with an uranyl acetate/lead citrate staining protocol (Reynolds, 1963). Light microscopy preparations were observed with an Olympus microscope (BX43) with a UC50 camera at the Natural History Museum's imaging facilities, while ultrathin sections on gold coated grids were observed with a Hitachi Transmission Electron Microscope (H-7650) at 80 kV at the imaging facilities in Kew Botanical Gardens.

Measurements on Histological Sections
Several parameters were recorded to understand the life cycle dynamics of all five species of Geodia, (i) size, density, and   In parenthesis the number of male and female specimens.
Frontiers in Marine Science | www.frontiersin.org fecundity estimations of reproductive features (oocytes and spermatic cysts), (ii) bacteria and yolk abundance within the oocytes, (iii) length of the reproductive period (considering the timing of appearance/disappearance of gametes), and (iv) the percentage of the population engaged in reproduction. In the following sections, specimens with gametes in their tissues were referred as reproductive while specimens without any gamete observed in their tissue sections were called non-reproductive.

Size and Density of Reproductive Features
The measurements of size and density of the reproductive features within the sponge tissue and for the fecundity estimations were obtained from light microscopy images with the Olympus microimaging software CellSens standard, connected to the microscope. Measurements of yolk and bacterial content in the ooplasm were acquired from TEM images with ImageJ (Schneider et al., 2012) on the TEM images. The size of oocytes and spermatic cysts and density of oocytes were calculated in three different areas (0.58 mm 2 each area) of tissue per specimen at 10x magnification. Size and density of spermatic cells could not be obtained, since that would require their individual measurement and counting, but they were so densely packed within the spermatic cysts that they could not be visually separated and counted. As thus, only the size of spermatic cysts was reported here.

Fecundity Estimations
While female fecundity was estimated based on the total number of oocytes produced per sponge, male fecundity estimations were based on the percentage of the sponge area occupied by spermatic cysts. For the female fecundity estimations, firstly the number of oocytes/mm 3 was extrapolated with the modified Elvin formula (Elvin, 1976), E = N t/d + t /v, as used by Di Camillo et al. (2012), where N is the number of the oocytes in a microscopic section; t, the thickness of the section; d, the diameter of the oocytes and v, is the volume of the histological section. Secondly, the area and the volume of each species were calculated. It was considered that almost all the studied sponge species (except for G. atlantica) had a spherical shape. The average diameter per species (Supplementary Table S2) was based on dimensions previously published (Cárdenas et al., 2013) or observed (personal observations, HTR) for these species, and then the area of the sponge was calculated according to the formula 4 × π × r 2 and the volume of the sponge according to the formula, 4 / 3 × π × r 3 . Regarding G. atlantica, its shape was considered more similar to a hollow cone, then using the dimensions (70 cm diameter × 40 cm height) mentioned in Cárdenas et al. (2013) for the maximum volume and using 3 cm as average thickness for the tissue of the sponge, the area of a hollow cone: π × r × s was calculated, where s, is the slant height, s = √ r 2 + h 2 and the volume for a hollow cone (outside volume -inside volume; where the volume for the cone is π × r 2 × h/3) (Supplementary Table S2). We should report here that although the diameters of Geodia spp. might be different in other publications (i.e., Klitgaard and Tendal, 2004;Cárdenas et al., 2013;Kutti et al., 2013) than the mean diameter used here, we mainly used specimens coming from the Arctic region for this analysis and so we extracted the information mainly based on photos of the specimens and personal observations (HTR). Only the dimensions of G. atlantica were retrieved by already published data (Cárdenas et al., 2013) as we never collected a whole specimen for this species to estimate dimensions. Consequently, the fecundity rates for the species of this study are a rough estimation and they should be used with caution.
For the fecundity estimations and the oocyte content (bacterial symbionts and yolk content), individuals with the most mature oocytes (maximum diameter of Vi oocytes) were used for G. phlegraei (Sula reef, July), G. atlantica (Korsfjord, September), and G. hentscheli (Schulz Bank, August) while individuals with mature oocytes, but not those with the most mature oocytes, were used for G. barretti (Barents sea, August) and G. macandrewii (Sula reef, July) because only those specimens had been processed for TEM, having a more complete information. Regarding the male individuals, only some of the G. phlegraei and G. atlantica individuals were in the latest developmental stages, while specimens of G. macandrewii, G. barretti, and G. hentscheli were in less advanced but similar developmental stages. For each species, the individuals with the most mature SP_II spermatic cysts (i.e., largest average size) were chosen to estimate male fecundity.

Histological Features of Geodia Species
General Description of the Mesohyl of Geodia spp.

Reproductive Features of Geodia spp.
All species of Geodia were gonochoristic; not a single specimen of Geodia spp. had both female and male reproductive features at the same time. In all species, individuals at the beginning of oogenesis with small previtellogenic (PV) oocytes, and individuals with more developed, vitellogenic (Vi) oocytes were found (Supplementary Figure S2A), although sometimes these two stages were found in the same individual (e.g., G. atlantica, Korsfjord June). Similarly, both male individuals with spermatic stage I (SP_I) cysts and more mature, spermatic stage II (SP_II) cysts were identified (Supplementary Figure S2B). In all Geodia spp., the diameter of PV oocytes measured on average 19 ± 6 µm and that of SP_I spermatic cysts was 30 ± 6 µm (Supplementary Figure S2). At a later stage, Vi oocytes had an average diameter of 54 ± 14 µm and SP_II spermatic cysts around 60 ± 25 µm (Supplementary Figure S2). G. hentscheli had the smallest size of both female and male reproductive features among the Geodia species (Supplementary Figure S2), when comparing individuals with either early or relatively late developmental stages, while G. atlantica had the largest oocytes and mature spermatic cysts found among the species studied here (Supplementary Figure S2).

Oogenesis
Oocyte development in most species was synchronous at both individual and population levels. All species showed very similar cytological features which are described together in the following sections. At the beginning of oogenesis (Figure 2), ameboid PV oocytes (8-25 µm in max. diameter) (Supplementary Figure S2A) with a well distinct nucleus and nucleolus were widespread throughout the mesohyl of the sponges (Figures 2A,B). A thin layer of collagen started to form around the PV oocytes (Figures 2C-E). PV oocytes were full of electron light vesicles, with few lipid droplets observed in the ooplasm of the oocytes (Figures 2D-G). While many bacterial symbionts were present in the mesohyl outside the oocyte ( Figure 2D), only very few were observed within the oocyte at this stage ( Figures 2E-H). Phagocytosis of bacteria was achieved with pseudopodia ( Figures 2D,H). Both the Golgi apparatus and the nuage (among mitochondrial clouds) could be observed at this stage in most species (Figures 2E,F).
In a later developmental stage, Vi oocytes were larger in general (>50-60 µm) (Figure 3 and Supplementary Figure S2B) and migrated to the vicinity of the canals (Figures 3A,C). Mature oocytes of G. barretti (50-60 µm), G. macandrewii (40-60 µm) and G. atlantica (103 µm in the largest diameter) were ready to get released (Figures 3B,C and Supplementary Figure S2A). When in the canals, mature oocytes were found in clusters, enveloped by a very thin collagen membrane ( Figure 3B). A thick collagen layer was formed around the Vi oocytes but no cellular follicle (supportive membrane surrounding the oocyte) was observed in any of the studied species (Figures 3D-F). Vi oocytes contained a high amount of yolk, mainly of lipid origin (Figures 3D-H, 4, 5), and only few protein platelets were observed in the ooplasm (Figures 3D,E, 4, 5). Nurse cells were not observed in any of the Geodia spp. and the yolk was produced by autosynthesis via the rough endoplasmic reticulum and the Golgi apparatus (Figures 4B,D). In fact, lipid content was higher than protein content in both PV and Vi stages, increasing from an average of 7.5-17% compared to protein yolk that increased from 2 to 3% (Figure 5). Both G. barretti and G. hentscheli had the highest percentage of lipid content, 20 and 19% respectively, in their small Vi oocytes (35 ± 5 µm and 30 ± 17 µm in diameter respectively) ( Figure 5B). G. macandrewii and G. atlantica had 13 and 15% of lipid yolk occupying their mature oocytes (43 ± 3 µm and 95 ± 7 µm in diameter respectively) ( Figure 5B). Although at the beginning of vitellogenesis, vesicles with heterogenous yolk in oocytes of G. phlegraei and G. barretti were observed (Figure 5A), at later stages none of the studied species had heterogenous yolk in the ooplasm (Figures 3D-H, 4, 5B). Unfortunately, no high resolution TEM images of Vi oocytes of G. phlegraei could be obtained to investigate the presence of homogeneous and heterogeneous yolk in the later maturation stages.
Extensive vertical transmission of bacterial symbionts from the mother to the descendant at Vi stage in most species was also observed (Figures 3F-H, 4A, 5). Bacteria were collected by oocyte pseudopods and each bacterium was stored in vesicle in which bacterial division was observed in some cases (Figure 3G, insert). No signs of digestion were observed in the stored bacteria in the oocytes (Figure 3G). The average bacterial content increased from 1% in PV oocytes to 4% in Vi oocytes across species ( Figure 5). The Vi oocytes of G. macandrewii (diameter ∼43 µm) and G. atlantica (with diameter ∼95 µm) had the largest area occupied by bacteria (∼7%) (Figures 3F, 4, 5B), followed by G. hentscheli with 4.4% in Vi oocytes (∼30 µm in diameter) (Figures 3E, 5B). However, Vi oocytes of G. barretti of similar size to those of G. hentscheli (diameter ∼35 µm), had a very low bacterial content (0.6-0.7%) without a significant increase in bacteria between the PV and Vi stages (Figures 3D,E, 5).

Spermatogenesis
SP_I spermatic cysts were scattered throughout the mesohyl, measuring on average 20-40 µm in all species ( Figure 6A and Supplementary Figure S2B). Very early spermatic cysts contained only a few cells, mainly spermatogonia (Figures 6A-C). Bacteria were very abundant in the surroundings of the spermatic cysts, but they were seldomly observed within the lumen of the spermatic cyst. In the SP_I spermatic cysts of G. hentscheli, a few bacteria were phagocytosed by the spermatogonia, which were also full of vesicles and stored reserves (Figures 6C,D). The spermatogonia did not show a fully functional flagellum, but the axonemal structure could be seen in the cytoplasm of the spermatogonium, probably being resorbed ( Figure 6D). This indicates that sperm cells, clearly derived from choanocytes, lose the choanocyte flagellum during early spermatogenesis.
At a more mature stage, SP_II spermatic cysts measured on average 40-80 µm ( Figure 6E and Supplementary Figure S2B) and formed clusters around the canals ( Figure 6E). In some cases, a thin layer of mesohyl between two spermatic cysts seems to disintegrate, inducing the fusion of different spermatic cysts into one ( Figure 6F). Fused clusters of spermatic cysts reached a maximum length of ∼200 µm ( Figure 6G). No cellular or collagenous follicle layer surrounding the spermatic cyst was observed at any stage in any of the studied Geodia spp. (Figures 6A,B,H, 7A).
Inside SP_II spermatic cysts, there were more than 300 cells in the process of division to be developed into sperm cells (Figures 6G,H). Their development within the SP_II spermatic cyst was clearly asynchronous as cells in different developmental stages were observed, the ones dividing at early stages in    the outer parts and the ones in late developmental stages, including spermatids and spermatozoa, toward the lumen of the cyst (Figure 7A). Spermatogonia were around 5-10 µm in maximum diameter; they had an amoeboid shape, with a well-developed nucleus and nucleolus (Figure 7A), and an internalized flagellum in the cytoplasm (Figure 7B). Primary (G) Spermatic cysts fused into one large spermatic cyst after mesohyl layer disintegration in G. phlegraei and were gathered around the canals (ca). (H) Close-up of a SP_II spermatic cyst of G. phlegraei with several spermatic cells, smaller than spermatogonial cells in size, compared to those in Figure 6B.
spermatocytes were frequently observed dividing, they had a slightly smaller size (∼4-6 µm), no nucleolus, and showed chromatin condensations in the nucleus (Figures 7C,D), as well as synaptonemal complexes (Figures 7D,E). Secondary spermatocytes had more compacted chromatin and a smaller size (∼4 µm) than the previous developmental stage (Figures 7E,F). The internalized flagellum already disappeared at the stage of primary spermatocyte, and in the secondary spermatocyte, FIGURE 7 | Ultrastructure of developmental stage_II (SP_II) spermatic cysts. SP_II spermatic cysts of G. phlegraei from Tromsøflaket, western Barents Sea in August showing (A), asynchronous development of sperm cells within the spermatic cyst with spermatogonia (s) in the outer part of the spermatic cyst, primary spermatocytes (ps) and secondary spermatocytes (ss) toward the inner part with increasing chromatin compactions. (B) Details of the structure of a spermatogonium with a nucleated (nu) nucleus (n) and the internalized axonemal structure of the flagellum (f). (C) Dividing primary spermatocyte with chromatin compaction after nuclear division. In the periphery of the spermatic cyst, ameboid-like cells (ac) are present. (D) Dividing primary spermatocytes with a cytoplasmic bridge (cb) before cytokinesis is completed and the synaptonemal complexes (syn) in the nucleus, (E) a secondary spermatocyte within a spermatic cyst, with chromatin compaction (cc) within the nucleus (n), synaptonemal complexes (syn), mitochondria (m), and the basal body (bb) before the new flagellum formation. (F) SP_II spermatic cyst with primary spermatocytes (ps), secondary spermatocytes (ss), spermatids (sd) and a potential spermatozoon (sp). Chromatin compaction (cc) and the acrosome (ar) of the cone-shaped late spermatid/early spermatozoon. the basal body was observed indicating that a new flagellum is being formed during late spermatogenesis in Geodia spp. (Figure 7E). Then, spermatids had an even higher degree of chromatin compaction and a smaller size (∼2 µm) ( Figure 7F). Finally, although we did not manage to obtain a clear picture of a spermatozoon with a flagellum, a potential spermatozoon (or a mature spermatid) was observed with several acrosomal vesicles in the apical part of the conical cell body (Figure 7F). The nucleus at its majority was composed by compacted chromatin and the mitochondria in the cytoplasm could not be observed, probably because of the angle of the image. Putative spermatozoa had the smallest size (∼1.5 µm) within the spermatic cyst, suggesting that this might be indeed the mature sperm cell ( Figure 7F).
Finally, we did not observe bacteria within sperm cells at later developmental stages.

Density of Oocytes and Female Fecundity (Number of Oocytes Produced per Individual)
The number of oocytes per square millimeter of sponge tissue varied among individuals of the same species and among species (Figure 8). In average, Geodia spp. produced 7.2 ± 3.6 oocytes/mm 2 of sponge tissue. However, some extreme cases were also identified. For instance, in two specimens of G. hentscheli with oocytes in PV stage, one had on average 37 oocytes/mm 2 and the other 1.7 oocytes/mm 2 (Figure 8). Additionally, a specimen of G. macandrewii with Vi oocytes, had the second highest value, 27 oocytes/mm 2 on average (Figure 8).
According to the calculations based on the modified Elvin's formula and the volume of each species, G. macandrewii (which had the second largest body volume among the studied Geodia species) contained the maximum number of oocytes within its body (∼116 million oocytes), followed by G. atlantica which was estimated to contain ∼78 million oocytes per individual. G. barretti and G. phlegraei were estimated to potentially release ∼30 and ∼17 million oocytes respectively, and finally, G. hentscheli was expected to produce only ∼4 million oocytes per individual ( Table 2).

Male Fecundity (Area of Tissue Occupied by Spermatic Cysts per Individual)
In G. phlegraei, the SP_II spermatic cysts occupied at least 12% of the total area of the sponge, and for G. atlantica only 7% of the total area ( Table 2). For the other three species, G. macandrewii and G. hentscheli had a similar investment of spermatic cyst production (∼3%) and G. barretti had the lowest investment of all species (1.5%) ( Table 2).

Reproductive Season of Geodia Species
From May to September, several individuals from the five Geodia spp. were producing gametes ( Figure 9A, Table 1, and Supplementary Table S1). An exception was G. macandrewii: even though it was reproductive in September (in Svalbard and western Norway), none of the eight specimens collected during August (Tromsøflaket, western Barents Sea) was reproductive ( Figure 9A, Table 1, and Supplementary Table S1). For G. phlegraei, only one specimen was collected in September (in Svalbard) and it was not engaged in reproduction ( Table 1). For G. atlantica, no samples were collected in August, but otherwise, this species was reproductive throughout the rest of the sampling months and sampling points ( Figure 9A, Table 1, and Supplementary Table S1). Finally, some G. hentscheli specimens collected in late June and August in Schulz Bank were reproductive ( Figure 9A, Table 1, and Supplementary  Table S1). Based on the maturation level of the oocytes observed in specimens the specific months and studied areas, the approximate period of spawning for all the Geodia spp. was estimated from May to November, depending on the area considered ( Figure 9B).
In detail, in May and June, almost all the studied species contained reproductive features in early stages of gametogenesis and the size and maturation of gametes increased over the months (Figure 10 and Supplementary Figures S3, S4). During July and August, oocytes of most species were Vi (Figure 10 and Supplementary Figure S3). An exception was G. barretti, in which one reproductive individual had already Vi oocytes (56.1 ± 9 µm) in May (in Hordaland coast) when most other individuals had PV oocytes, and another individual had PV oocytes (23.3 ± 6.4 µm) in July (in Sula reef) when all others had Vi oocytes. Other exceptions were an individual of G. phlegraei (from Tromsøflaket, western Barents Sea) and another one of G. hentscheli (from Schulz Bank) which were also in early developmental stages in August (Figure 10 and Supplementary Figure S3), and surprisingly, one individual of G. phlegraei with Vi oocytes obtained in July in Sula reef The calculations were based on the theoretical shape and the mean diameter of each species (based on published dimensions, personal communication with (HTR) and personal observations). The calculations of the volume and area of the sponge per species and the diameter of oocytes used are shown in Supplementary Table S2. ( Figure 10 and Supplementary Figure S3). For G. macandrewii, G. barretti and G. atlantica, their largest Vi oocytes (in the final Vi stage) were observed in September (Figure 10 and Supplementary Figure S3). However, in September in addition to females with mature (Vi) oocytes, there were also females of G. macandrewii and G. atlantica with PV oocytes in Svalbard (not shown in the graph) and Rosemary Bank Seamount respectively (Figure 10 and Supplementary Figure S3). Regarding the spermatocytes, there was a general tendency of enlargement over the period from May to September (Supplementary Figures S3,  S4) in all species.

Ratio of Reproductive Individuals (Males/Females) at Population Level
During the reproductive months, a high percentage (65-90%) of G. phlegraei individuals participated in reproduction and most of them were male with a sex ratio of 3:1 ( Figure 11A and Table 1). A high percentage (65-75%) of the population of G. macandrewii was also engaged in reproduction but with either females predominating (1:2) or with equal proportions of males and females (1:1) ( Figure 11B and Table 1). Regarding G. barretti (Figure 11C and Table 1), although a low percentage of individuals were producing gametes (11.6-20%) in May, June, and August, 80% of the individuals were reproductive in September in the Rosemary Bank, Seamount, while interestingly in the Korsfjord, western Norway, none of the seven individuals were reproducing. For G. barretti, the number of male and female individuals was equal (1:1). On the same note, for G. atlantica, 30% of the individuals collected were reproducing in May -June. In September, in contrast to G. barretti, the latter species had 40% of individuals engaged in reproduction in the Rosemary Bank, Seamount (Rockall Trough, 1034 m) and almost 80% of the individuals reproducing in the Korsfjord (western Norway, continental shelf). The sex ratio for G. atlantica was 1:1 ( Figure 11D and Table 1). Finally, G. hentscheli recruited between 60 and 80% of its individuals for reproduction in June and August respectively, with more females than males, approximately in a ratio of 1:3 ( Figure 11E and Table 1).

Morphological Characteristics of Gametes
The Geodia species studied here were all gonochoristic and oviparous, as observed in previous studies on the Geodiidae family Sciscioli, 1969, 1970;Mercurio et al., 2007;Spetland et al., 2007), which belongs to the Astrophorina suborder, one of the three suborders currently recognized in the Tetractinellida order (Astrophorina, Spirophorina and Thoosina). Thenea abyssorum, another deep-sea astrophorin (family Theneidae) has also been shown to be gonochoristic and oviparous (Witte, 1996), suggesting that this could be the rule for the whole suborder. In the suborder Spirophorina, about half of the Tetillidae studied are also oviparous and gonochoristic (Liaci et al., 1976;Watanabe, 1978;Lepore et al., 2000) while another half, e.g., Tetilla polyura (Topsent, 1904) and Tetilla longipilis (Topsent, 1904), are viviparous (Topsent, 1904;P. Cárdenas, unpublished data). The suborder Thoosina is entirely hermaphroditic and viviparous (Bautista-Guerrero et al., 2010;Carballo et al., 2018). With the current knowledge of the Tetractinellida phylogenetic tree, and based on previous analyses of ancestral character reconstruction for demosponges (Riesgo et al., 2014), we hypothesize an ancestral gonochoristic and oviparous condition for Tetractinellida which later shifted to hermaphroditism/viviparism in Thoosina, although data are missing on the reproduction of one entire clade of Tetractinellida including the Stupendidae and the Scleritodermidae/Siphonidiidae/Azoricidae (Carballo et al., 2018). Besides the phylogenetic constraints, gonochorism and oviparity seem to be predominant in deep-sea invertebrates as more than 80% of deep-sea corals are also broadcast spawners (oviparous) and gonochoristic, while the most common pattern in shallow-water corals is hermaphroditism and oviparity (Brooke and Young, 2003;Burgess and Babcock, 2005;Waller, 2005;Harrison, 2011;Pires et al., 2014). Whether there is an ecological advantage for gamete release in deep-sea areas, in terms of lower predation over eggs is still unknown, but it would be necessary to investigate it to understand the ecology of deep-sea organisms. Oogenesis in the studied Geodia spp. had very similar characteristics to that described previously for several astrophorins (Geodia conchilega, G. cydonium, G. barretti, Stelletta grubii, Erylus discophorus, T. abyssorum), including the formation of a collagen layer outside the oocyte, vertical transmission of bacterial symbionts acquired by phagocytosis, absence of nurse cells and autosynthesis of nutrients by the oocyte itself via the Golgi and endoplasmic reticulum apparatus (Liaci and Sciscioli, 1969;Sciscioli et al., 1989Sciscioli et al., , 1991Sciscioli et al., , 1994Witte, 1996;Mercurio et al., 2007;Spetland et al., 2007). Oocytes in all Geodia spp. seemed to derive from archaeocytes, which at the beginning of the gametogenic cycle have a very similar shape and size (Liaci and Sciscioli, 1969;Sciscioli et al., 1994). On the other hand, spermatic cysts in all the studied species were clearly derived from choanocyte chambers, as previously suggested for other Geodiidae (Liaci and Sciscioli, 1969;Mercurio et al., 2007) but questioned for G. barretti (Spetland et al., 2007). These results clearly show a choanocyte origin of spermatic cysts, even for G. barretti. Spetland et al. (2007) suggested an alternative origin of the spermatic cysts based on the larger size of cysts than chambers, probably because they did not consider the fusion of several cysts. In this study, there is evidence that the early cysts did show a similar size and morphology to chambers (i.e., no collagen or cellular envelope), and fusion events were observed in later stages across all species.

Reproductive Strategy and Sex Ratio: Commonalities and Differences
Oocytes of deep-sea Geodia spp. from this study and of G. barretti from Spetland et al. (2007), have overall larger sizes (individuals with oocytes measuring more than 45 µm in diameter were detected in all species), with some species reaching 100 µm when fully grown, than mature oocytes of the shallower species, G. cydonium, that measure 30-40 µm (Sciscioli et al., 1994;Mercurio et al., 2007). Similarly, specimens of the family Tetillidae from deeper waters had 30% larger mature female gametes (∼74 µm) than shallow-water specimens (∼57 µm) (Meroz-Fine et al., 2005). The increasing size of eggs with depth had also been observed in deep-sea crustaceans and echinoderms that have larger lecithotrophic eggs than shallow water species (Mauchline, 1973;Corey, 1981;Cameron et al., 1988), indicating a possible general strategy for several marine invertebrates (Thorson, 1950). The size of mature oocytes in marine organisms with lecithotrophic embryonic and larval stages, is indicative of the pelagic larval duration (PLD) as larger oocytes are expected to accumulate a larger content of nutrients. Larger egg size can have profound ecological implications in marine invertebrates, such as longer planktonic phases and PLD and advantages in external fertilization given the larger targets for sperm (Marshall and Keough, 2007).
The planktonic phase duration in lecithotrophic propagules is dictated not only by the amount but also the type of yolk. In their oocytes, all the Geodia spp. had far more lipid than protein yolk. Even though lipid yolk in cold-water sponges is not always the rule (Koutsouveli et al., 2018;Koutsouveli et al., unpublished data), it is usually predominant over protein yolk in dispersal propagules of some sponge species from cold environments, while being 30% or less of the nutrient reserves in temperate and tropical species (Riesgo et al., 2015). For example, Stelletta grubii, an astrophorin species from temperate shallow waters, produces more protein than lipid yolk (Sciscioli et al., 1991). Furthermore, the temperate species G. cydonium also seems to have higher amounts of protein yolk accumulated in their oocytes (Sciscioli et al., 1994) than the Geodia spp. of this study. The predominance of fat provides more energy than protein and is metabolized faster, which is important for quick developmental changes in environments that are poor in nutrients, like the deep sea (Giese, 1966;Clarke, 1980). In addition, lipids also provide high buoyancy to the eggs which allows them to float in the water column (Clarke, 1980). The high content of lipids in the released oocytes of the deep-sea Geodia spp. could suggest an adaptation to a longer stay in the water column than the propagules of their temperate counterparts, until fertilization and ultimately embryogenesis take place, allowing longer dispersion distances and ensuring population connectivity over larger areas.
The vertical transmission of the bacterial symbionts in Geodia spp. was observed in oocytes of late developmental stages. This phenomenon has also been described in G. cydonium, as also in other astrophorid species (i.e., E. discophorus and S. grubii) (Sciscioli et al., 1989(Sciscioli et al., , 1991, in which bacteria were acquired by phagocytosis from the mesohyl, and were not digested but rather protected in vesicles within the oocytes, where they divided. As thus, these bacteria are true symbionts and are not consumed for nutritional needs at least during gametogenesis. Vacelet (1975) has suggested that these bacteria can serve as nutrients at a later embryonic stage. A recent study, not yet peer-reviewed, has showed that bacterial symbionts of Geodia spp. (including G. barretti, G. hentscheli, and G. atlantica), provide the host with precursor lipid building blocks, contributing to the synthesis of sponge unsaturated long-chain fatty acids (De Kluijver et al., 2020). So, the high content in lipid yolk might be related to the presence of high microbial abundance in these species, and it might be the case that by-products of bacterial symbionts within the oocytes contribute to the "autosynthetic" activity of lipid formation. On the other hand, in spermatogenesis, some bacteria were engulfed by early spermatogonia but disappeared in later spermatogenic stages. So, these sparse phagocytized bacteria are more likely to be used directly as a source of energy. This is a relatively novel observation in sponges, previously found only in the sperm of Spongia officinalis (Gaino et al., 1984) and Petrosia ficiformis , and it clearly points to choanocytes as the cells producing sperm.
Although the major reproductive patterns (i.e., oviparity and gonochorism) of the deep-sea Geodia spp. are clearly constrained by phylogeny (Riesgo et al., 2014), being the strategy at the genus, and possibly at suborder levels (e.g., Sciscioli, 1969, 1970;Witte, 1996;Mercurio et al., 2007;Spetland et al., 2007), there is plasticity in some of the reproductive traits adopted by each species in order to cope with the energetic demands required for gametogenesis in these relatively oligotrophic habitats (Witte, 1996). First of all, slight variations were observed in seasonality among the species. For instance, individuals of G. barretti with mature female gametes and those of G. atlantica with premature female gametes were found at the same location and year, e.g., in Rosemary Bank Seamount in September 2016. This phenomenon has also been observed in Mediterranean species sharing the same habitat, which have to cope with the same nutrient regimes, staggering in time their resource allocation Pérez-Porro et al., 2012;De Caralt et al., 2018). The different timing in gametogenesis and spawning among species could also be a strategy for avoiding cross-fertilization between different species and therefore hybridization. Furthermore, differences in terms of reproductive investment at individual and population levels were observed. Specifically, G. barretti seems to be the species with least investment in reproduction as only a small fraction of the population produces gametes and in relatively low numbers compared to the other studied species. This potentially could lead to a fragile situation of G. barretti if the environmental conditions or the populations themselves are perturbed. Interestingly, the gametes of G. barretti and G. hentscheli have smaller sizes compared to G. phlegraei or G. atlantica (Figure 10 and Supplementary Figures S2, S3) but they contain a higher amount of yolk (even though not the most mature vitellogenic oocytes were measured for G. barretti) (Figure 5), which can give them all the necessary reserves and buoyancy for longer drifting and thus ensure a higher fertilization success, expecting even higher yolk content in more mature oocytes of G. barretti. On the other hand, G. atlantica, with relatively lower amounts of yolk ( Figure 5) will have shorter survival rate and lower buoyancy, so shorter drifting of its oocytes. However, G. atlantica produced more oocytes than the rest of species (three times more oocytes per individual, compared to G. barretti) (Figure 8 and Table 2), and a high percentage of the population is reproducing each year (Figure 11D), which increases the chances for successful reproduction. These species-specific reproductive traits can be correlated with observations of a most continuous presence of G. barretti compared to the patchier distribution of G. atlantica for instance, in Northern Norwegian continental shelf, while both showed a high abundance (Kutti et al., 2013). Although phylogenetically distantly related to G. atlantica (Cárdenas et al., 2010), G. macandrewii seems to follow a fairly similar strategy. Finally, in G. phlegraei, a comparatively higher percentage of the population participate in reproduction (Figure 11A), producing fewer oocytes (Figure 8 and Table 2) but highly enriched with nutrients ( Figure 5).
Overall, oocyte densities in Geodia spp. (N of oocytes/mm 2 ) were comparable or even higher to those reported on other oviparous sponges from temperate and tropical habitats. Riesgo and Maldonado (2008) found that the three oviparous and gonochoristic Mediterranean sponge species, Cymbaxinella damicornis (called Axinella damicornis in the publication), Raspaciona aculeata, and Chondrosia reniformis reached a maximum of 3-7 oocytes/mm 2 , while the coral reef sponge species Cliona vermifera had a maximum density of 3.5 oocytes/mm 2 and 60-85% of its population participated in reproduction (Bautista-Guerrero et al., 2014). Two other Mediterranean oviparous, hermaphrodite Cliona spp., C. viridis and C. celata, showed a density of ∼10 oocytes/mm 2 and a 70-90% of the population to participate in reproduction (Piscitelli et al., 2011). An exception was specimens of the family Tetillidae (most possibly the species Levantiniella levantinensis) which produced up to 155 oocytes/mm 2 in some Mediterranean deeper habitats (Meroz-Fine et al., 2005). In the Geodia spp. of this study, a mean of 10 oocytes/mm 2 was observed across species (Figure 8), with some specimens having a much higher number than the average (e.g., 38 ± 4 oocytes/mm 2 in PV individual of G. hentscheli and 27 ± 6 oocytes/mm 2 in Vi individual of G. macandrewii). In addition, a high percentage of their population participated in reproduction, as also mentioned for G. cydonium (Mercurio et al., 2007). So, despite the scarcity of nutrients in these habitats throughout the year, these species seemed to have a high reproductive effort, that together with their interplay with microbial symbionts (Radax et al., 2012;De Kluijver et al., 2020) could explain the dense aggregations of these species in the sponge grounds. This pattern contrasts to the theory that deep-sea benthos has usually a low reproductive effort due to unfavorable abiotic conditions, such as lack of nutrients (e.g., Thorson, 1950).
Even though, a relatively low number of specimens for each species was available, a dominance of females over male individuals (or equal ratios) was still observed in most species of the study, thus confirming previous studies on Geodia spp. (Liaci and Sciscioli, 1969;Mercurio et al., 2007;Spetland et al., 2007), as well as in most demosponges (e.g., Wapstra and van Soest, 1987;Tanaka-Ichihara and Watanabe, 1990;Witte et al., 1994Witte et al., , 1996Corriero et al., 1996Corriero et al., , 1998Lepore et al., 2000;Meroz-Fine et al., 2005;Maldonado and Riesgo, 2008). G. phlegraei was the only species where males were three times the females. Also, there was a lower number of oocytes produced by female individuals compared to sperm generation by male individuals (e.g., in G. phlegraei ∼16 million oocytes/sponge and ∼30 billion spermatozoa/sponge respectively), which is possibly related to the higher energetic demands of oogenesis. Therefore, more female individuals should be engaged in each reproductive cycle in order to increase the percentage of oocytes released in the water column and a successful fertilization. Male dominance is a rare characteristic for sponges (Simpson, 1984;Sarà, 1993) but the energy requirements and allocation might be behind such a pattern: as the production of yolky female gametes requires higher energetic demands, this implies that the more individuals are involved in female gamete production, the more individuals of a population will be vulnerable at the end of the reproductive cycle due to lack of reserves, and so the whole habitat can be more vulnerable in case of disturbance. An indication that males require less energy to produce their gametes was shown in Oscarella lobularis, in which the male to female ratio increased massively (10 male:1 female) during food limitation (Ereskovsky et al., 2013). However, the feature of G. phlegraei cannot be directly correlated with abiotic conditions, and further sampling is needed to derive any conclusion.

Seasonality and Adaptations of the Reproductive Strategy to the Deep Sea
Organisms in most deep-sea environments normally face relatively constant abiotic conditions with some very predictable changes along the year, such as nutrient blooms after a seasonal primary production (Southwood, 1977;Clarke, 1980). In these habitats, the phenomenon of seasonality in reproduction is strong, mostly happening during the summer months, after the spring bloom, when the energy reserves have increased due to increased nutrient uptake (George and Menzies, 1967;Clarke, 1980). In general, astrophorin demosponges from the northern hemisphere, both deep-sea (Witte, 1996;Spetland et al., 2007) and shallow ones (Mercurio et al., 2007), reproduce annually, during summer months. In this study, the five boreo-arctic Geodia spp. were also reproducing during the summer months (May-September). However, this pattern presented some deviations in the different locations of the study. For example, reproductive specimens of G. barretti at Korsfjord (western Norway, 100-400 m) were barely found in June and September, an observation that comes in accordance with Spetland et al. (2007), who detected an annual cycle of G. barretti shifted to earlier months of the year (February -June) at this same location. However, reproductive features of G. barretti were found at Rosemary Bank Seamount (Rockall Through, 1304 m) in September. On the other hand, Spetland et al. (2007) also found that populations of G. barretti in Kosterfjord (Sweden, 80 -200 m) had two annual cycles, one in February -June (estimated spawning time, May/June) and the other in July -October (estimated spawning time, October). Interestingly, no reproductive activity in specimens collected either in April or May in the same location was observed.
Differences in the reproductive cycle of a species between two locations or even between years in the same location can be explained by variations in environmental factors (Witte et al., 1994;Meroz-Fine et al., 2005). Similar observations have been reported for the deep-sea corals Primnoa resedaeformis (Gunnerus, 1763) and Paramuricea placomus (Linnaeus, 1758) in the Gulf of Maine (Fountain et al., 2019), in which populations from different areas showed variations in timing and even in the reproductive potential (e.g., max. size of the gametes, reproductive individuals/population, fecundity ranges). Likewise, populations of the cold-water coral Lophelia pertusa (Linnaeus, 1758) in the NE Atlantic and Trondheimfjord populations had variations in timing, maximum size of oocytes, and duration of their reproductive cycle (Waller and Tyler, 2005;Brooke and Järnegren, 2013). Brooke and Järnegren (2013) also attributed those differences likely to regional variations in the food supply. Mercier et al. (2011) found an indirect connection between the increased activity in the spawning of the gametes in six deepsea species of echinoderms and the new/full moons, suggesting that lunar cycles affect the cyclic currents and the deposition of particular matter to the seafloor, which is then the main factor influencing the reproduction in the deep sea. Nutrient supply determines many aspects of reproduction in deep-sea invertebrates: from initiation of gametogenesis to vitellogenesis and time of spawning, depending on the reproductive strategy of the species (Tyler et al., 1992(Tyler et al., , 1993Gooday, 2002;Young, 2003). For instance, deep-sea species tend to generate large eggs and invest a lot of energy on yolk formation of their eggs (even more if they have lecithotrophic larvae), and the stage of maturation/vitellogenesis is synchronized with periods of high food delivery (Eckelbarger and Watling, 1995).
The reproductive cycle of deep-sea sponges, including astrophorins and more specifically Geodia spp., was also suggested to be triggered by primary production (Witte, 1996;Spetland et al., 2007). In this analysis, specimens from locations at different depths and geophysical conditions were studied. In the Schulz Bank, only one plankton bloom is observed per year with the peak of carbon flux (input of organic matter) during June (U. Hanz, personal communication). Similarly in the Barents Sea, the peak of particulate organic carbon flux in the seafloor occurs during June -July (von Bodungen et al., 1995), which coincides with the maturation phase of gametes both in species of this study and in another deep-sea astrophorins (Witte, 1996). In Norwegian fjords though, there is an important spring bloom (March -April) and a smaller autumn bloom (August -September) (Wassmann, 1991;Wassmann et al., 1996), and the latter also seems to be synchronized with oocyte maturation for at least some of the studied species here. Deep-sea corals also synchronize their gametogenesis with the influx of nutrients to the seafloor as they take advantage of the spring/summer blooms to initiate their gametogenesis and then they spawn at the beginning of the following year (Maier et al., 2020). In the northern part of the Kosterfjord, very close to the sampling area of this study, spawning of the deep-sea coral L. pertusa begins in late January to February (Larsson et al., 2014), though other populations in Trondheimfjord (closer to the Sula reef) spawn from mid-February to late March (Brooke and Järnegren, 2013;Larsson et al., 2014) coinciding with a later summer bloom in this area. As the reproductive activity of many deepsea invertebrates, which are seasonal reproducers, seems to be dependent on nutrient fluxes coming from the surface, it means that fluctuations in annual primary blooms could slightly displace the reproductive cycle of these species.

Marine Conservation in Boreo-Arctic North-Atlantic Deep-Sea Sponge Grounds
In marine conservation, reproductive timing, larval dispersal, and connectivity are key factors to take into account for developing adequate strategies to protect a variety of marine organisms (e.g., corals, mussels, fish) and habitats (Carson et al., 2010;Treml et al., 2012;Schmiing et al., 2016). In deep-sea sponge grounds, due to lack of essential biological information and to discontinuous monitoring, only prediction models based on generalized approaches are used to assess the potential dispersal of sponge populations and their capacity for regeneration in case of disturbance (e.g., Kenchington et al., 2019). The reproductive timing is a crucial parameter to consider when designing a conservation strategy, not only because it is the period of recruitment, but also because reproductive individuals are more vulnerable to disturbances due to their high energetic investment in gamete production and spawning. In addition, in the case of mechanical disturbance during the reproductive season, organisms may release premature propagules that cannot survive (Henry et al., 2003). For instance, samples of the deep-sea coral L. pertusa, from intensively trawled areas on Darwin Mounds, did not produce any gametes, also showing high levels of clonality that indicate a pause of sexual reproduction (Le Goff-Vitry et al., 2004;Waller and Tyler, 2005). In turn, samples of the same species in areas with less exploitation produced larger amounts of gametes during the exact same time (Waller and Tyler, 2005). Shallowwater corals (e.g., Pocillopora damicornis, Linnaeus, 1758) and sponges (e.g., Luffariella variabilis, Poléjaeff, 1884) reduce their reproductive activity during stress and invest more of their energy in survival (e.g., metabolic maintenance, tissue repair) (Brown and Howard, 1985;Ward, 1995;Ettinger-Epstein et al., 2007). Similar behaviors are then expected for their deepsea counterparts.
In Geodia grounds, the fragmentary information related to lifespan, reproductive activity, reproductive strategy, and population connectivity patterns of the dominant species, has made the development of long-term conservation strategies on these habitats very difficult and in some cases unsupported (Thompson and Fuller, 2020). The species of the genus Geodia have one or two annual reproductive cycles. The spawning period for Geodia spp. in the studied areas has a broad range (May to November) depending on the area considered. Indeed, the reproductive period of Geodia spp. in sponge grounds should be taken into account to implement special protective measures related to fisheries during such times. Estimated spawning periods for Geodia spp. are completely different than that observed for deep-sea corals, which in the boreo-arctic areas of the North Atlantic spans from January to March (Brooke and Järnegren, 2013;Larsson et al., 2014). This therefore calls for specific measures of protection for the different habitats in the deep-sea. Such information combined with population genetic studies on these species and oceanographic circulation patterns would give more accurate predictions of dispersal and connectivity patterns of these populations in order to develop adequate conservation strategies specific to each species and habitat.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material and archived in the PANGAEA data repository (https://doi.pangaea.de/10.1594/ PANGAEA.921626). Further inquiries can be directed to the corresponding author/s.

AUTHOR CONTRIBUTIONS
VK, AR, PC, and HR collected the samples. AR and VK contributed to conception and design of the study. VK, AR, and MC processed the samples. VK analyzed the data and drafted the manuscript. AR and PC did essential corrections to the manuscript. All the authors contributed to manuscript revision, read, and approved the submitted version.

FUNDING
This work was supported by the H2020 EU Framework Program for Research and Innovation Project SponGES (deep-sea Sponge Grounds Ecosystems of the North Atlantic: an integrated approach toward their preservation and sustainable exploitation) (Grant Agreement no. 679849). This document reflects only the authors' view and the Executive Agency for Small and Mediumsized Enterprises (EASME) is not responsible for any use that may be made of the information it contains. Funding was also provided by a grant of the Spanish Ministry of Science and Innovation (PID2019-105769GB-I00) to AR.