Ericaria selaginoides (Linnaeus) Molinari & Guiry (formerly known as Cystoseira/Carpodesmia tamariscifolia) is a monopodial and perennial species growing in the littoral and sublittoral zones, under moderate and high wave exposure (Roberts, 1970). E. selaginoides is distributed throughout the Atlantic, from the British Isles to Mauritania (including the Macaronesia Islands), and in some Mediterranean areas under Atlantic influence, such as southern Spain, Sicily, Morocco, Algeria and Tunisia (Roberts, 1970; Gómez-Garreta, 2000). E. selaginoides is one of the most common canopy-forming algal species along the central and eastern islands of the Canary archipelago (Wildpret et al., 1987; Rodríguez et al., 2008). E. selaginoides is a monoecious species in which the apical branches of mature receptacles contain conceptacles with both oogonia and antheridia (Roberts, 1970). Reproduction is oogamic (i.e., large non-motile eggs and biflagellate sperm), and fertilization is external (Guern, 1962; Gómez-Garreta, 2000). After fertilization, large and free-living zygotes (∼70–100 μm) rapidly sink to the bottom, where they are attached to hard substrate during the first 12–24 h (Verdura et al., 2021). This gives the species a low dispersal ability (<20 cm; Mangialajo et al., 2012). Although this is a perennial species, receptacles are most developed in spring and summer (Gómez-Garreta, 2000).

On Gran Canaria Island (Canary Islands, eastern Atlantic Ocean, Figure 1A), E. selaginoides populations are distributed along the northern coast, which is exposed to substantial swells (Supplementary Figure S1). Historically, populations were found in relatively abundant patches in the western region, which became rare and scattered to the east due to their proximity to areas of intense urban development (Rodríguez et al., 2008; Valdazo et al., 2017; Supplementary Figure S1). Populations of E. selaginoides have undergone a marked reduction in both distribution and population size and now predominantly exist as rare and scattered patches. This phenomenon is currently being investigated in an ongoing study (unpublished data), that aims to elucidate the causes of this decline, as has already been documented in Macaronesia (Bernal-Ibáñez et al., 2021a; Martín-García et al., 2022; Valdazo et al., 2024). The environmental parameters, sea and air temperature as well as irradiance, derived from satellite climate data (Supplementary Table S1), exhibit homogeneity across the entire distribution range of E. selaginodes (Supplementary Figures S2-S5).

We used the same area as donor and receptor for our restoration pilot experiments, which is located along the northwestern coast of Gran Canaria Island (Figure 1A). The study site at Punta de Gáldar is located within the marine SAC “Costa Sardina del Norte”, where local impacts such as wastewater discharges, eutrophication, pollution and habitat destruction are controlled. This area is dominated by abrupt basaltic cliffs, under significant annual tidal variations (ca. between 1 and 3 m; Ramírez et al., 2008) and wave exposure; sea surface temperature ranges between 18 °C in March and 24 °C in October (Tuya and Haroun, 2006; Valdazo et al., 2017). Fertile material was sourced from Punta de Gáldar (28°10′12.18″N, 15°41′25.46″W; Figures 1A,B), selected because of the observed well-conserved Cystoseira s.l. abundant patches in the present study. Historically, this site supported dense populations of Cystoseira forests, with species such as E. selaginoides coexisting with Gongolaria abies-marina (Wildpret et al., 1987; Rodríguez et al., 2008; Valdazo et al., 2017).

The density of live embryos per Gram of receptacle DW (n° alive embryos mm^-2 g^-1 DW) was used as the dependent variable to compare settlement patterns across treatments at Day 3. By this time, fertilised embryos had already divided and begun rhizoid formation, allowing a clear distinction between viable and non-viable embryos. For each replicated treatment, one glass slide was selected, and live embryos were counted in five 1.50 mm² quadrats randomly selected from the photographs. A univariate Generalized Linear Model (GLM) was then fitted to the data (i.e., density of settled live embryos per Gram of receptacle DW), considering “Culture” (three levels) and “Light” (two levels), as fixed factors. The model was fitted using a ‘Gamma’ distribution and a ‘log’ link function in the R statistical package, which is suitable for continuous, positive data.

Counts of living embryos were recorded at Day 3, Day 6 and Day 10, which were conducted by analyzing images of a 1.5 mm² area randomly selected from the captured images at each time. The density of living embryos was then analyzed using a Generalized Linear Mixed Effect Model (GLMM) with a ‘Gaussian’ distribution and an ‘identity’ link function, with “Culture” (three levels) and “Light” (two levels), as fixed factors, and “Time” as a random factor. Models were fitted through the R ‘glmmTMB’ package (Brooks et al., 2017).

During the culture period, five distinct embryonic stages were observed: (I) embryos with four primary rhizoids (Figure 2A), (II) embryos with more than four rhizoids (Figure 2B), (III) elongated embryos with short apical hairs (Figure 2C), (IV) elongated embryos exhibiting branching and long apical hairs (Figure 2D), and (V) irregularly shaped embryos (Figure 2E). The proportion of embryos at each embryonic stage was then compared at Day 10 through a univariate GLM, considering “Culture” (three levels) and “Light” (two levels), as fixed factors. Models were fitted using the ‘betareg’ function, implemented via the ‘betareg’ R package (Cribari-Neto and Zeileis, 2010), which assumes a ‘logit’ link function and a residual distribution appropriate for continuous, positive, data that are proportions bounded from 0 to 1.

Ten days after fertilization (Day 10), embryos had progressed into the distinguishable developmental stages described above. To assess embryonic size, the area of five randomly selected individuals per embryonic stage was measured by analyzing five randomly selected images of 1.5 mm² area (Valdazo et al., 2024). Differences in area per embryonic state were then analyzed using a GLM with a ‘Gamma’ residual distribution family and a ‘log’ link function, with “Culture” (three levels) and “Light” (two levels), as fixed factors.

Approximately 500 receptacles (Figure 1E) were harvested on 4 August 2024, following the same transport and release protocol describe above. A total of 40 hexagonal concrete clay tiles (17.50 cm² in area, side length of 2.5) with a porous and rough surface were used. The tiles were commercially purchased from Silica Studio, Spain; the rough, porous, texture facilitates embryo adhesion and settlement (Falace et al., 2018; Supplementary Figure S6E–H). Tiles were maintained under culture conditions optimized in the previous phase to promote embryo growth, using Von Stoch culture medium, a 15:9 h light:dark photoperiod, 125 μmol photons m⁻²s⁻¹, and a constant temperature of 20 °C using a temperature-controlled water bath (Figure 1D). Two aquariums (10 L) were used, which were initially filled with 1 L of culture medium for the first 24 h to facilitate embryo settlement. After the receptacle removal, the volume was increased to 8 L per aquarium, with medium renewal every 4 days to mitigate any possible effects of nutrient limitation. Continuous aeration was maintained using air pumps and bubblers to ensure adequate oxygenation.

During laboratory culture, embryo development on tiles was periodically monitored. Twenty-four hours after seeding, receptacles were removed and the dry weight (DW) of the receptacles associated with each tile was measured; this time was considered the fertilization point (Day 0). Release efficiency at Day 0 was calculated as the number of attached embryos per tile divided by the DW of the corresponding receptacles (embryos g^-1 DW). Release efficiency was obtained through photographic sampling. Each tile was photographed perpendicularly from above using an OM System TG-7 camera (OM Digital Solutions, Tokyo, Japan) mounted on a fixed tripod at a known distance to ensure consistent scale and perspective across images. Image analysis was carried out in ImageJ, and embryos were counted manually by overlaying a grid on each photograph and visually tallying the items within each grid cell (Savonitto et al., 2021).

As development progressed, early juveniles became densely packed and formed aggregated patches on the tiles, preventing individual counts. Therefore, embryo survival was quantified as the percentage of tile area occupied by embryos (cm²). Survival was measured on days 5 (Day 5) and 12 (Day 12) in 20 randomly selected tiles, and on day 33 (Day 33) for all tiles prior to outplanting. Survival was obtained through photographic sampling using an OM System TG-7 camera (OM Digital Solutions, Tokyo, Japan). Image analysis was carried out in ImageJ to quantify the area occupied by embryos, calculating survival as the proportion of tile surface covered by embryos relative to the total tile area (Schneider et al., 2012; Malfatti et al., 2023). The percentage of tile area occupied by embryos was analysed using a GLM with a ‘Gamma’ residual distribution and a ‘log’ link function, with “Time” as a fixed factor with four levels: Day 5, Day 12, Day 33 and the outplanting day (Day 0-OT).

After 33 days in culture (Day 33), tiles were transported to Punta de Gáldar, which is located ca. 50 km away, in refrigerated boxes filled with sea water. To examine how exposure to environmental stress in the donor habitat affected juvenile survival and success, we outplanted juveniles in different physical conditions within the natural vertical distribution of E. selaginoides. We implemented two treatments: height on the shore and pool presence. Height on the shore included two levels within the low-intertidal zone: upper and lower (U and L, respectively). The lower level is exposed and accessible for only a few days during spring tides, which occur bi-monthly. Conversely, the upper level is exposed every day at low tide. The upper level was assumed to have increased temperature and desiccation stress (Ramírez et al., 2008; Betancor et al., 2015). The “pool” treatment (inside vs. outside of pools, +P/-P) was applied to half of the tiles within each zone, resulting in four treatment combinations. Each combination included 10 tiles, from a total of 40 initially seeded. We selected four pools with similar physical characteristics, in terms of size and depth (ca. 3 m² surface area and 0.5 m deep; Supplementary Figures S6A–D). We verified the presence of Ericaria selaginoides in the four pools. Herbivore pressure is not considered locally relevant, predominantly arising from mesograzers sheltering in the algal fringe (authors’ personal observation). Seeded tiles were distributed across the four treatments (n = 10 per treatment: P + U, P−U, P + L, and P−L) to ensure homogeneous initial embryo coverage across treatments. Tiles were affixed to the rock using epoxy putty, stainless steel bolts, nuts, washers and screws (Supplementary Figures S6E,F). On the outplanting day (Day 0-OT), measurements were assessed immediately after tiles were attached to the substrate, to account for any potential losses associated with transport and deployment. Tiles were subsequently monitored twice over 3 months, as allowed by weather conditions: at 57 days (Day 57-OT) and 91 days after outplanting (Day 91-OT). Adverse climatic conditions led to the loss of several tiles, reducing the sample size in some treatments, especially in the lower eulittoral treatments. At the lower level, we lost five epoxy bolts at Day 0-OT in both P+ and P− treatments, and an additional two tiles were lost in P− at Day 57-OT (Supplementary Table S2).

At each outplanting monitoring time (Day 0-OT, Day 57-OT and Day 91-OT), several measurements were taken, including: (a) juvenile survival (i.e., % area covered by embryos, cm²), (b) morphometric analysis, by measuring maximum juvenile length, and (c) juvenile occurrence, calculated as the percentage of tiles in which juveniles were alive. All analysis were done through photographic sampling with an OM System TG-7 camera (OM digital Solutions, Tokyo, Japan). Tiles were photographed perpendicularly from above to estimate survival, and laterally to obtain morphometric data; both orientations included the same scale reference. Image analysis was performed in ImageJ, where survival was calculated as the proportion of juvenile-occupied surface relative to the total tile area, maximum juvenile length was measured from the lateral images and juvenile occurrence was computed as the % of tiles showing juvenile presence (Schneider et al., 2012; Malfatti et al., 2023). We implemented GLMMs with a ‘Tweedie’ residual distribution family and a ‘log’ link function to assess the effect of shore level (two levels: Upper and Lower) and pool presence (two levels: absence, P-, and presence, P+), as fixed factors. We used “Time” as a random effect and the percent cover at the time of deployment, as a covariate, to evaluate their effect on independent variables.

All modeling and testing were conducted using R (Rstudio Team, 2022). For all fitted models, diagnosis plots of residuals and Q-Q plots (Supplementary Figures S7-S17) were visually inspected to check the appropriateness of the fitted models (Harrison et al., 2018).

Gamete release occurred shortly (0–3 h) after induction, and the release of oospheres and antherozoids was evident by the presence of numerous spherical cells marked by the formation of an orange-colored mound (Figure 3A). Fertilization occurred externally, leading to the formation of a fertilization membrane that facilitated adhesion of the zygotes to the substrate, with a mean diameter of 145.55 μm ± 6.10 (mean ± SD; n = 20). Then, 24 h after fertilization, the zygote cytoplasm, initially homogeneous, became clearly visible along with the nucleus (Figure 3B). Subsequently, cytoplasmic differentiation (polarization) was observed, characterized by the establishment of a vertical growth axis. Approximately 48 h after fertilization, the first division occurred perpendicular to this axis, generating two daughter cells of equal size (Figure 3C). The second division took place 20 h later (Figure 3D), and then a third division (Figure 3E). Between 48 and 72 h post-fertilization, multiple divisions occurred with an increase in the total embryonic volume, leading to the formation of rhizoidal bud protrusions (Figure 3F). After 72 h, nearly all embryos displayed initial rhizoid formation. The rhizoid mother cell divided, giving rise to 4 cells that differentiated into primary rhizoids (Figure 3G). These structures continued to elongate, forming long filamentous projections (Figure 3H). Within 5 days after fertilization, most embryos were detached from the fertilization membrane, coinciding with the emergence of secondary rhizoids (Figure 3I). As development progressed, the embryos exhibited significant elongation and an apical invagination, from which short apical hairs emerged (Figure 3J). By the end of the first week, the embryos had undergone notable morphological changes, measuring 541.91 ± 64.02 μm in length and 260.81 ± 47.81 μm in width (n = 20). Further elongation, the initiation of branching, and the presence of prominent, elongated, apical hairs were observed (Figure 3K), pointing to an advanced stage of early thallus development.

The density of alive embryos (per unit of receptacle mm^-2 g^-1 DW) ranged from 192.40 ± 124.10 in L + AL treatment to 377.70 ± 334.70 in L + VS, with intermediate values in the remaining treatments (L–AL: 235.0 ± 148.6; L–SW: 245.1 ± 141.3; L–VS: 263.7 ± 109.3; L + SW: 240.0 ± 126.4; mean ± SD), and did not significantly vary across treatments (Figure 4; Supplementary Table S3).

Embryo density declined over time among all treatments (Figure 5), with no significant differences observed between treatments (Figure 5; Supplementary Table S4).

Significant differences in embryonic developmental stages were observed among treatments on Day 10 after fertilization (Figure 6; Supplementary Table S5). The highest proportions of individuals at the last embryonic stage (stage IV) were achieved in the VS and SW culture media (Figure 6), both showing significantly higher values compared to AL (p < 0.001; Supplementary Table S4), with L+ conditions showing a negative effect on stage IV (p < 0.001; Supplementary Table S5).

On Day 10, embryo size varied across treatments and developmental stages (Figure 7; Supplementary Table S6). At stage II (Figure 7; Supplementary Table S6), no statistically significant effects across treatments were detected. At stage III, the culture media had a significant effect, with VS and SW treatments having smaller areas than the AL treatments (p = 0.05 and p < 0.05, respectively; Supplementary Table S6). At stage IV, the juvenile area was significantly influenced by the culture media, with an inconsistent effect across light levels (Supplementary Table S6). Embryos under the VS culture showed significantly higher area than both AL and SW treatments (p < 0.001 and p < 0.01, respectively; Supplementary Table S6). Interaction effects were also significant, with L−VS showing higher area than both L + AL and L + SW (p < 0.001and p < 0.05, respectively; Supplementary Table S6), and L + VS showing higher area than L-AL p < 0.01; Supplementary Table S6).

Zygote release efficiency in culture averaged 3,038.66 ± 2,548.56 (mean ± SD; n = 10) zygotes DW⁻¹. Juvenile cover increased progressively over time, with a mean tile cover of 2.17% ± 1.44% (mean ± SD; n = 10) on Day 5 (Supplementary Figure S18). This value rose to 6.02% ± 4.8 (mean ± SD; n = 10) on Day 12, and further increased to 13.99% ± 4.84% (mean ± SD; n = 10) on Day 33.

At the time of outplanting (Day 0-OT), cover stabilized at a mean of 16.61% ± 11.78 (mean ± SD; n = 40), indicating no mortality during transportation (Supplementary Table S7). The average juvenile length at this point was 0.66 ± 0.30 cm (mean ± SD; n = 40). Mean percent cover and juvenile length varied over time. On Day 57-0T, mean cover had decreased to 6.19% ± 9.78, with a mean length of 0.51 ± 0.66 cm (mean ± SD; n = 28). On Day 91-OT, cover slightly increased to 7.76% ± 11.35, while mean juvenile length reached 0.75 ± 0.94 cm (mean ± SD; n = 28). Alive juvenile occurrence varied among treatments (Supplementary Figure S19). Juveniles allocated to the “Lower/P+” treatment exhibited the highest and most stable occurrence of alive juveniles across time. Juveniles from the “Upper/P+” treatment showed an initial mortality but remained relatively stable afterwards. Juveniles from the “Lower/P-” treatment experienced a reduction and maintained a lower but consistent presence, whereas in the “Upper/P-” treatment exhibited a rapid loss of juveniles, leading to total disappearance by Day 57-OT.

Juvenile percentage cover (Figures 8A–C) and length (Figures 8D–F) over time were significantly influenced by pool presence, with individuals in “P+” showing higher values compared to those in “P−” (p < 0.001 and p = 0.001, respectively, Supplementary Table S8). Cover and length were also significantly higher in the “Lower” compared to the “Upper” zone (p < 0.001 both of them, Supplementary Table S8). The covariate (initial cover percentage) had a marginally significant effect on cover (p = 0.05, Supplementary Table S8), indicating an influence of initial embryo density on persistence at this stage. However, the effect on length was not significant (p = 0.20; Supplementary Table S8).