Optimizing in vitro fertilization in four Caribbean coral species

Gamete collections

We conducted this study on the island of Curaçao in the Southern Caribbean where we collected gametes on predicted spawning nights (Marhaver, Chamberland & Vermeij, 2024) in July 2019 (D. labyrinthiformis), September 2019 (P. strigosa), October 2019 (C. natans), and October 2020 (O. faveolata). All spawn collections were performed under research and collection permit #2019/021824 granted to the CARMABI Research Station by the Ministry of Health, Environment and Nature of Curaçao. Spawn collection dates and times for each species are available in Table S1. Spawning colonies were tented with cone-shaped nets only once gamete bundles were clearly observed in the polyp mouths and release was imminent. Positively buoyant egg-sperm bundles accumulated in inverted removable 50-mL polypropylene conical centrifuge tube (Falcon; Corning Life Sciences, Tewksbury, MA, USA) secured at the top of these nets. In most cases gamete bundles took more than an hour to break up and release their contents (Table S1). During this time, we were generally able to (1) reduce the volume of seawater inside the collection tubes to a 1:1 bundle:seawater ratio to maintain high sperm concentrations and (2) return to the laboratory. All fertilization assays began within an hour after gamete bundles had broken up. Temperature in the laboratory was maintained between 28 and 29 °C, similar to daily average sea surface temperatures (SSTs) in July, September, and October 2019, and in October 2020 (NOAA Coral Reef Watch, 2020).

Experimental design

A schematic representation of the experimental design is shown in Fig. 2. For all three assays, we selected three parents as sperm-donors and a separate parent as egg-donor (Fig. 2). Hence, in each fertilization assay, eggs from a single dam had the opportunity to be fertilized by sperm from three other, non-self sires (Fig. 2). Individual parent collections with similarly high gamete densities and no visible signs of contamination or oocyte malformation were prioritized for the fertilization assays. We opted for one, single-dam-multi-sire cross (1♀ × 3♂) per species, rather than conducting multiple pairwise, single-dam-single-sire crosses (1♀ × 1♂). This design aimed to minimize risks of individual incompatibilities in pairwise crosses which have been observed in several scleractinian species (Willis et al., 1997; Baums et al., 2013; Miller et al., 2018). We ideally would have incorporated additional biological replicates (i.e., several crosses with unique dams) to capture inter-genet variability among dams of each species. This was however not feasible due to logistical limitations on the nights of spawn collections, namely completing the numerous manipulations required to conduct each fertilization assay within the ~5 h window before gamete quality diminishes consistently across all four species. For each experimental treatment, we did however include technical replicates (i.e., the same cross repeated independently for each experimental treatment) (n = 3 for C. natans, P. strigosa, and O. faveolata and n = 5 or 6 for D. labyrinthiformis).

Gamete preparation

We separated sperm and eggs upon arrival at the laboratory (Fig. 2). We first transferred each gamete collection into a modified 50-mL conical tube (Falcon; Corning Life Sciences, Tewksbury, MA, USA) fitted with a control valve at its base. Once all egg-sperm bundles had broken up, we drained sperm via the control valve, through a 100-µm mesh cell-strainer to remove impurities, and into a clean 50-mL conical tube. Positively buoyant eggs remained in the modified tubes where we rinsed them by repeatedly adding and draining 0.5-µm-filtered seawater (FSW) until clean, after which they were transferred into larger 150-mL plastic beakers filled with 100 mL of FSW. Eggs were stirred occasionally with a clean pipette until they were used in fertilization assays.

We diluted the two densest sperm collections to a concentration approximately equal to that of the most dilute collection by adding FSW until reaching a similar opacity. We then pooled 30 mL of sperm from each of these three sperm donors to ensure each sire was approximately equally represented in the 90-mL sperm pool (Fig. 2). Gamete samples were then assigned to one of three assays aimed at determining fertilization success as a function of (1) sperm concentration, (2) gamete age, and (3) gamete co-incubation time.

In the sperm concentration experiment (Assay 1), 30 mL of this dense sperm mixture were used to perform a series of dilutions as described below (Fig. 2). For use in Assay 2 and Assay 3, we further diluted a 30 mL aliquot of this mixture to target (nominal) 10⁶ cell mL⁻¹ using FSW, a concentration previously resulting in high fertilization success in several other coral species (Nozawa, Isomura & Fukami, 2015) (Fig. 2). Due to time limitations, it was not possible to quantify sperm concentration before starting the assays. This dilution was therefore achieved based on a visual assessment of the mixture’s opacity. Samples of the concentrated and the diluted sperm mixtures were fixed in an 8% solution of glutaraldehyde in FSW to allow post hoc quantification of concentration using a haemocytometer chamber (Bright-Line; Hausser Scientific, Horsham, PA, USA) under a phase contrast optical microscope (Leica, Heerbrugg, Switzerland). These actual concentrations are available in Table S2.

Fertilization assays

Assay 1: sperm concentration—This assay aimed to determine the optimal sperm concentration for successful IVF for each of the four dams representing our four study species. We did so by comparing the success of fertilization carried out at five different sperm concentrations (Fig. 2). From the concentrated mixture described above, we produced a series of eight 1/10 dilutions in FSW and selected four of these based on their opacity to target a range of ~10² to 10⁹ cell mL⁻¹ sensu Nozawa, Isomura & Fukami (2015) (Table S2a). For example, in C. natans the initial pooled sperm sample contained 1.06 × 10⁹ cell mL⁻¹, resulting in five tested conditions of 1.06 × 10⁹, 10⁸, 10⁶, 10⁴, and 10² sperm mL⁻¹ (Table S2a). To quantify the possible occurrence of parthenogenesis, self-fertilization, and/or sample contamination with non-self sperm before the experiments started, we further included an additional set of replicates as controls for which we added washed eggs to FSW with no added sperm (Fig. 2). Thus, these ‘no-sperm’ controls ensured that the fertilization rates obtained from our assays reflected true fertilization success resulting from outcrossing under the different conditions tested, though they had obviously been exposed to self-sperm in the short time prior to separation/washing. Gametes were mixed 30 to 45 min after bundles had broken up and co-incubation time was standardized to 60 min across all treatments.

Assay 2: gamete age—To determine the maximum gamete age at which high fertilization success can be achieved for each species, we mixed sperm and eggs after they had been kept separate for a controlled amount of time (Fig. 2). The first time point was initiated once egg-sperm bundles had broken up and we had separated and washed the eggs and sperm. This generally coincided with ~2 h after spawning (AS) (Table S1). The following treatments were started ~30, 60, 90, 120, 180, and 300 min later, corresponding to ~2.5, 3, 3.5, 4, 5, and 6 h gamete age (Fig. 2). Gamete co-incubation period was standardized to 60 min and sperm concentration was the same for all age treatments (~10⁶ cell mL⁻¹, Table S2b). Note that the 4 h gamete age treatment was omitted in C. natans due to a manipulation error.

Assay 3: gamete co-incubation period—To identify the optimal gamete co-incubation period for successful fertilization, we controlled the amount of time eggs and sperm were exposed to one another (Fig. 2). To do so, we placed all eggs in contact with sperm at the same time, 30 to 45 min after egg-sperm bundles had broken up, but rinsed them after 15, 30, 45, 60, 90, and 120 min. Sperm concentration was the same for all co-incubation time treatments (~10⁶ cell mL⁻¹, Table S2c).

In all three assays, 10 mL of sperm solution were added to individual wells in six-well cell culture plates (CellMAX; Biopioneer Inc., San Diego, CA, USA) fitted with 100-µm mesh cell-strainers (SWiSH Premium; Stellar Scientific, Baltimore, MD, USA) (Fig. 2). Fertilization was initiated when placing ~200 eggs into each of the cell-strainers immersed in the sperm solution. We included five or six technical replicates per treatment for D. labyrinthiformis and three for all other species. At the end of each assay, we stopped fertilization by rinsing eggs in a series of water baths consisting of 150-mL plastic beakers filled with 100 mL of FSW (Fig. 2). We carefully lifted each cell strainer containing eggs from the culture plate, partially lowered it into a water bath, and gently swirled it to remove residual sperm. We repeated this process three times in new FSW before transferring eggs into a final beaker filled with clean FSW. When all fertilized eggs had at least reached a 64-cell stage (roughly 3 to 4 h after fertilization), we photographed a subset of eggs (mean of 72, range of 10 to 228) from each replicate under a dissecting stereomicroscope (MBS-10; LOMO, St. Petersburg, Russia) and assessed fertilization rates using the resulting images (Fig. 2), whereby we considered eggs showing no signs of cleavage at this point to be unfertilized. Only intact, full-sized embryos or eggs (and not fragments) were considered during photo-analyses.

Data analysis

We performed all statistical analyses in R version 4.2.2 (R Development Core Team, 2020). We used one-way Kruskal–Wallis non-parametric tests to determine statistical differences in fertilization success among treatments within each assay. When significant differences were detected, we ran Dunn’s post hoc pairwise comparisons to identify different treatment groups. All statistical results are available in Table S3.

Sperm concentration

Fertilization reached a maximum at sperm concentrations between ~10⁵ and 10⁸ cell mL⁻¹ in D. labyrinthiformis, between ~10⁶ and 10⁹ cell mL⁻¹ in C. natans and P. strigosa, and between ~10⁶ and 10⁷ cell mL⁻¹ in O. faveolata (Fig. 3, Tables S2a, and S3a), though for O. faveolata we did not test for concentrations above 10⁷ cell mL⁻¹. At least 97% of eggs from the three brain coral colonies were fertilized at these concentrations, and fertilization success for O. faveolata reached 89% ± 0.7 (n = 3) (mean ± SE) at a concentration of 6.4 × 10⁵ cell mL⁻¹. At 10⁴ cell mL⁻¹ or less, fertilization almost entirely failed in D. labyrinthiformis (<5%), and was below 55% in dams of all three other species (Fig. 3, Tables S2a, and S3a).

We obtained the highest sperm concentrations in collections from P. strigosa and C. natans (~10⁹ cell mL⁻¹) (Table S2a). While maximum fertilization was obtained at this concentration for P. strigosa (99% ± 1 (n = 3)), fertilization success was slightly (albeit not significantly) lower at this concentration in C. natans (89% ± 2 (n = 3)) compared to the next highest concentration of 10⁸ cell mL⁻¹ (100% ± 1 (n = 3)) (Fig. 3, Tables S2a, and S3a).

None of the control replicates where eggs were rinsed and kept in FSW without sperm presented significant levels of fertilization (≤5%) (Fig. 3, Tables S2a, and S3a). Therefore, parthenogenesis, self-fertilization, and sample contamination across all other experimental conditions were considered negligible.

Gamete age

Near complete fertilization (95–100%) was achieved in the three brain coral crosses when gametes were combined between 2h20 and 4h20 AS (Fig. 4, Table 1, Tables S2b, and S3b). In D. labyrinthiformis, 99% ± 1 (n = 6) of eggs were fertilized in the 4h20 AS treatment, relative to 4% ± 2 (n = 6) when gametes were introduced 6h20 AS, suggesting that gametes produced by D. labyrinthiformis ceased to be viable within that time window (Fig. 4, Table 1, Tables S2b, and S3b). Fertilization success declined more gradually between 4h40 and 6h50 AS in both C. natans and P. strigosa, with 68% ± 8 (n = 3) and 85% ± 5 (n = 3) of eggs from these dams being fertilized when mixed 6h50 AS, respectively (Fig. 4, Table 1, Tables S2b, and S3b).

We observed a different pattern in O. faveolata, whereby a delay AS was required to obtain maximum fertilization success. In this dam, when gametes were mixed 2h10 AS, 83% ± 2 (n = 3) of eggs were successfully fertilized compared to 89–92% when gametes were aged 2h40 to 4h40 AS (Fig. 4, Table 1, Tables S2b, and S3b). Similar to C. natans and P. strigosa, a 15% decline in fertilization was observed between 4h40 and 6h40 AS for O. faveolata, with 74% ± 3 (n = 3) of eggs fertilized at this last time point (Fig. 4, Table 1, Tables S2b, and S3b).

Gamete co-incubation period

The duration of gamete co-incubation did not influence fertilization success in the D. labyrinthiformis, C. natans nor P. strigosa colonies under study. Maximum fertilization was reached in all conditions, including co-incubation times as short as 15 min (≥97%) (Fig. 5, Tables S2c, and S3c). In contrast, O. faveolata gametes required a co-incubation period of 1h to 2h to achieve similar fertilization success rates (87–93%) (Fig. 5, Tables S2c, and S3c).

Sperm concentration

Sperm concentration was a strong driver of fertilization success in dams of all four species, with clear minimum thresholds to achieve the highest fertilization potential, namely ~10⁵ cell mL⁻¹ in Diploria labyrinthiformis and ~10⁶ cell mL⁻¹ in Colpophyllia natans, Pseudodiploria strigosa, and Orbicella faveolata (Fig. 3). In all crosses, fertilization success decreased dramatically at ~10⁴ cell mL⁻¹ or less (Fig. 3). These findings are consistent with previous reports of IVF success in other hermaphroditic broadcast-spawning scleractinians, requiring high sperm concentrations in the range of 10⁵–10⁷ cell mL⁻¹ (Fig. 6, Table S4). In fact, in 21 of 26 species for which information on sperm density and IVF is available, crosses conducted at concentrations below 10⁴ cell mL⁻¹ resulted in poor fertilization success (<50%) (Fig. 6, Table S4). Exceptions include Acropora latistella, A. tenuis, Favites colemani, and Platygyra daedalea, all of which can maintain intermediate (50–75%) to high (>75%) fertilization success under much lower sperm densities of 10³–10⁴ cell mL⁻¹ (Fig. 6, Table S4). Such large differences in effective sperm concentrations could be driven by various, species-specific characteristics such as differential chemoattraction of sperm towards egg chemical signals (Morita et al., 2006), varying levels of gamete incompatibilities (Willis et al., 1997; Baums et al., 2013; Miller et al., 2018), as well as variations in egg size (Levitan, 2006). In other marine invertebrate species, larger eggs require fewer sperm encounters to achieve fertilization (e.g., sea urchins: Levitan, 1996; Rahman & Uehara, 2004), bivalves: (Luttikhuizen, Honkoop & Drent, 2011), gastropods: (Huchette et al., 2004), providing a larger physical target for sperm in the water column. The four species included in this study produce eggs in the size range of ∼300 to 400 µm (Table S1) and exhibited similar sperm concentration requirements for successful fertilization during IVF. Further empirical studies are needed to determine the relationship between sperm limitation and variation in egg size in corals.

Previous studies have reported declining fertilization success attributed to polyspermy or reduced water quality at higher sperm concentrations, including for A. cervicornis, O. franksi, M. digitata, and F. pentagona (Fig. 6, Table S4). Here, we found no clear evidence for depressed fertilization at higher sperm concentrations as fertilization success remained high at concentrations equal or above 10⁸ cell mL⁻¹ for our crosses with P. strigosa (99%), D. labyrinthiformis (97%), and C. natans (99%), though in C. natans fertilization success decreased slightly (albeit not significantly) between 10⁸ and 10⁹ cell mL⁻¹ (Fig. 3). These results suggest effective mechanisms preventing polyspermy in eggs of these brain coral species (i.e., fast-block via egg membrane depolarization and/or slow-block via cortical granule reaction) (Gould & Stephano, 2003). Overall, risks of fertilization failure appear more common when cultures are too diluted rather than too dense. Therefore, in accordance with previously established guidelines for coral breeding and restoration practitioners (Guest et al., 2010; Marhaver, Chamberland & Fogarty, 2017; Banaszak et al., 2018; Severati et al., 2024), we recommend conducting IVF with sperm mixtures no more dilute than 10⁶ cell mL⁻¹ when possible for these species. Our results further support the likelihood of sperm limitation on sexual coral reproduction in situ, whereby gamete dilution may limit the fertilization potential of coral eggs during natural spawning events (Oliver & Babcock, 1992; Levitan et al., 2004; Mumby et al., 2024). Concentrations as high as achieved here (≥10⁸ cell mL⁻¹) are unlikely to occur in the open ocean where sperm rapidly disperse in the water column. Thus, risks associated with polyspermy may be marginal in the natural environment relative to ex situ fertilization cultures carried out in small volumes of water.

Gamete age

Our results suggest that combining sperm and eggs as quickly as possible is not essential in D. labyrinthiformis, C. natans, and P. strigosa, and is not recommended in O. faveolata. In our O. faveolata cross, a delay of 2h40 AS before mixing male and female gametes was required to obtain optimal fertilization (>90%, Fig. 4). Favites pentagona displays a similar pattern whereby fertilization success increases gradually, from ~75% to 90%, between 30 min and 2 h after spawning (Oliver & Babcock, 1992) (Table 1). In these species, it is possible that fertilization is contingent upon a delayed, final maturation division of the eggs and their release of polar bodies (Heyward & Babcock, 1986), or that maximum sperm flagellar motility is achieved after a longer period of time following spawning and/or requires a longer exposure to activation factors from eggs (Morita et al., 2006), though we did not record sperm activity in this study. Thus, when O. faveolata gametes are mixed soon after spawning, allowing a longer co-incubation period would be recommended, during which time more gametes could become competent.

The extended gamete viability (>4 h) observed in the current study may increase chances of egg and sperm encounters in situ under conditions of low sperm availability, or during imperfectly synchronized spawning events by allowing more time for gametes to meet (Chui et al., 2014). Together, these findings present restoration opportunities whereby gametes could be collected and later combined from sites representing more distant and distinct habitat types, thus enhancing potentially adaptive genetic diversity in larval cultures (Baums et al., 2019, 2022).

Co-incubation period

The high success achieved with co-incubation times as short as 15 min indicates that fertilization takes place quickly in D. labyrinthiformis, C. natans, and P. strigosa (Fig. 5) at the egg: sperm ratios tested here. Co-incubation times of 10 min have also yielded adequate fertilization success in other species including A. digitifera, A. papillare, A. tenuis, Goniastrea aspera, Platygyra daedalea, and P. ryukyuensis (Nozawa, Isomura & Fukami, 2015; Buccheri et al., 2023). Shorter co-incubation periods in larval propagation pipelines allow for earlier washing of the eggs following the mixing of gametes, before they undergo mitotic divisions and hence remain less susceptible to cell breakage during handling (Heyward & Negri, 2012; Severati et al., 2024). In contrast, O. faveolata gametes required a co-incubation period of 1h to 2h to achieve the highest fertilization rates (87–93%) (Fig. 5), though this could be due to a portion of the gametes needing more time to become competent (see Assay 2: gamete age).

Short in vitro co-incubation periods could be a function of egg-regulated sperm motility. Concentrating gametes in this way could increase the exposure of sperm to chemo-attractants that stimulate flagellar activity and aid sperm in locating an egg. Coral eggs also release inhibitors that reduce sperm motility once fertilized (Coll et al., 1994; Morita et al., 2006; Morita, Iguchi & Takemura, 2009). Thus, differing minimum co-incubation periods across species could result from varying levels of sperm motility and chemotaxis, influencing the time required for sperm to locate, swim towards, and fertilize eggs. Shorter periods required for fertilization to take place could have evolved so that fertilized eggs are able to immobilize sperm attempting to penetrate them before the physical interaction causes lethal damage, or to reduce the risks of polyspermy (Oliver & Babcock, 1992; Franke, Babcock & Styan, 2002). Alternatively, species-specific differences in the ‘fertilizability’ of eggs such as differences in the egg mucous coat or other processes following bundle breakup might affect required co-incubation time.

Broadcast spawning coral species release large quantities of gametes with a high degree of synchrony and this enhances chances of compatible gamete encounters. The longer the contact time required between sperm and egg for fertilization to take place, the higher the likelihood that gametes will disperse before fertilization has occurred, especially when strong ocean currents occur (Oliver & Babcock, 1992; Levitan et al., 2004; Mumby et al., 2024). Thus, rapid fertilization of eggs in the field may be an adaptation to facilitate fertilization in situations where sperm limitation is likely. Trade-offs may result, however, by reducing the effective number of potential mates that are slightly more distant.

Possible interactions between factors

In this study we examined three factors separately, holding the other two factors constant at an expected effective level. In practical situations applying IVF, there are likely additional factors that may affect fertilization success. For example, gametic incompatibilities between individual parents have been demonstrated in some species, including O. faveolata (Miller et al., 2018). For this reason, the effective sperm concentration (i.e., of compatible sperm) in a particular cross may be less than the absolute sperm concentration, and it is possible that this may account for the somewhat lower fertilization levels achieved here for O. faveolata overall. Maximizing the number of possible genet pairings in a fertilization cross may help alleviate the risks of fertilization failure due to genetic incompatibilities, though spawn collections from many egg and sperm donors might not always be feasible (Baums et al., 2022). Nonetheless, the best mitigation in cases of sperm limitation is likely to extend the co-incubation time, as basic kinetics indicate that more contacts between compatible egg-sperm pairs will occur over time, and no negative effects of extended, two-hour co-incubation times were observed here. For example, observations with Acropora palmata, a species with significant individual incompatibilities (Baums et al., 2013; VF Chamberland, 2019, personal observation), suggest that fertilization success at very low sperm concentrations (≤10³ cell mL⁻¹) can be improved with longer contact times (overnight; ~6 h) (MW Miller, 2014, personal observation). Other complex interactions between sperm concentration and gamete contact time have recently been demonstrated in several Indo-Pacific species. For instance, in A. digitifera, 10 min are required for sufficient sperm to find and fertilize eggs under concentrations between 10³ and 10⁴ cell mL⁻¹ (Table 2), whereas much shorter contact times of 10 to 60 s can yield acceptable fertilization success at 10⁵ cell mL⁻¹ in this species (Buccheri et al., 2023). Significant within-species variation in fertilization success among different gamete collections or cohorts has also been observed. For example, fertilization success for Platygyra ryukyuensis surpassed >75% at Magnetic Island, Australia when conducting crosses at concentrations of 10⁴ sperm mL⁻¹ (Miller & Babcock, 1997), whereas two orders of magnitude more sperm was required to achieve similar success with spawn collections from Green Island, Taiwan (Fig. 6, Table S4). The current study did not capture potential intra-specific, spatio-temporal variation, since only a single spawn collection was assayed for each species.

Extrapolating our findings to the reef environment requires caution given the multitude of parameters influencing coral sperm and egg encounter rates and fertilization success in the open ocean, which were absent in our assays (e.g., predation: Westneat & Resing (1988), water motion: Riffell & Zimmer (2007), sedimentation: Ricardo et al. (2015), acidification: Albright & Mason (2013)). Nonetheless, it is clear that with shrinking coral populations, potential mates have become more and more distant on reefs and that sperm limitation may now prevail during natural spawning events, thus increasing risks of reproductive failure through Allee effects (Mumby et al., 2024). All four species under study have suffered significant recent population declines, three of which are currently listed as Critically Endangered (IUCN, 2022). Thus, sperm densities are likely to fall below the 10⁴ cell mL⁻¹ threshold established in this study by the time eggs meet conspecific sperm, which may require greater travel distances and increased gamete aging, further raising risks of fertilization failure.