Does reef crest zone selection influence Acropora palmata (Lamarck, 1816) fragment survival and growth?

Study area

The study was carried out in four shallow reef crests in Cuba. Two of them are in the northwest region, in Playa Baracoa, Artemisa Province (23°03′20″N, 82°33′10″W) and Rincón de Guanabo, La Habana Province (23°10′23.63″N, 82°05′57.46″W). The other two, Mariflores (20°46′17.46″N, 78°53′44.34″W) and El Peruano (20°50′46.74″N, 79°1′4.32″W) lie to the south of the central region of Cuba, in Jardines de la Reina National Park (Fig. 1). These crests differ in distance from shore, abundance, and diversity of species such as fish and sea urchins, anthropogenic stressors, management and protection (Pina-Amargós et al., 2014; Rey-Villiers, Sánchez & González-Díaz, 2021; Ramos et al., 2024).

The reefs in the northwestern region are impacted by their proximity to the capital city and to coastal development (González-Díaz et al., 2018; Ramos et al., 2024). Pollution from heavy metals and fertilizers, street runoff, the Almendares and Quibú rivers and Havana Bay are the main anthropogenic factors that affect reefs near the city (González-Díaz, De la Guardia & González-Sansón, 2003; Rey-Villiers et al., 2020; Rey-Villiers, Sánchez & González-Díaz, 2021; Ramos et al., 2024). The Playa Baracoa crest is located approximately 230 m from the shoreline and 2 km east of Santa Ana River where untreated wastewater from a local educational institution (Latin American School of Medicine with an average annual enrollment of 10,000 students) is released (Ramos et al., 2024). Rincón de Guanabo is a marine protected area (Protected Natural Landscape/seascape similar to category V IUCN) located 800 m from the coastline. However, there is no effective management plan for this crest. The crest is nearly three km east of an oil drilling and extraction area (Boca de Jaruco thermoelectric power station), but data on nutrient load or pollutants (e.g., hydrocarbons) are either absent or unavailable. In these crests, coral cover is low (<17%) and algal cover is high (>87%) (Ramos et al., 2024). Fish biomass is low (∼12 g m⁻²), due to subsistence fishing (Duran et al., 2018; Gil-Agudelo et al., 2020).

On the other hand, Mariflores and El Peruano crests are located approximately 80 km from the main island of Cuba in the Jardines de la Reina National Park. This National Park has low human impact and is in good condition in terms of conservation (Beyer et al., 2018). It has been classified as an oligotrophic system where nutrient input is due to organic matter from mangroves, muddy sediments, and open waters such as the Gulf of Ana María and the Caribbean Sea (Pina-Amargós, Figueredo-Martín & Ross, 2021). In Jardines de la Reina National Park, the reef crests are characterized by A. palmata cover ranging from 22% to 45%, algal cover between 22% and 49%, and Diadema antillarum densities of 0.3 to 4.7 ind m⁻² (Hernández-Fernández, López & Sotolongo, 2016). Fish abundance is high due to the absence of overfishing (Pina-Amargós, Figueredo-Martín & Ross, 2021). According to Pina-Amargós et al. (2014), the crests showed high abundances of commercially important fish species, including Epinephelus striatus (15–65 cm in length, 0.2 ± 0.02 ind 1,000 m⁻²), Lutjanus cyanopterus (25–85 cm, 0.2 ±0.03 ind 1,000 m⁻²), and L. apodus (10–55 cm, 53.2 ± 2.2 ind 1,000 m⁻²). Large herbivorous fish were also present, such as Scarus guacamaia (45–115 cm, 0.1 ± 0.02 ind 1,000 m⁻²) and Sc. coelestinus (39–105 cm, 0.08 ± 0.003 ind 1,000 m⁻²).

Experimental design

Along the crest, we identified two zones relative to the breaking wave and parallel to the shoreline: the fore zone, where wave breaks first, and the back zone, the opposite side of the crest, where the energy of the wave is dissipated by the fore zone (Fig. 2). To estimate water flow intensity in the fore and back crest zones we used the dissolution of plaster discs (or ‘clod cards’ Doty, 1971) as an indicator (Jokiel & Morrissey, 1993). On each crest, we placed eight plaster discs weighing approximately 144 g each. Four plaster discs were placed parallel to the shoreline in the fore crest zone and another four were placed in the back crest zone. The discs were mounted on steel rods 10 cm off the substrate using a wire. The plaster discs were kept for 48 h in Playa Baracoa and Rincón de Guanabo crests, during April and June 2023, respectively. Whereas in El Peruano and Mariflores crests, the discs were placed twice, in December 2022 and April 2023, for 48 h. An exception was in December in Mariflores when the discs were retrieved after 24 h for logistical reasons. Once the discs were removed, they were left to dry and reweighed. The weight lost from each plaster disc provided an indicator of energy on each crest zone.

To test the effect of the crest zones associated with the water flow gradient on the survival and growth of A. palmata fragments, we established a field experiment. The fragments were collected from randomly selected colonies within each study crest. They were cut with forceps from the apical portions of different branches of several colonies, exhibiting a roughly rectangular shape and lacking secondary branching. These fragments were subsequently outplanted onto the same reef crest as the parental colonies, rather than in a common garden. A total of 50 fragments with a size (width and height) between one and seven cm were placed in each crest, 25 fragments in the back and 25 in the fore crest zones, one m apart each and parallel to the shoreline (Fig. 2). The A. palmata fragments were attached to the substrate with epoxy (KLIPTON ACUAPLAST) and each was labeled and monitored over time. The depth at which the A. palmata fragments were placed varied among crests and between zones within each crest. In Playa Baracoa, the depth is approximately 1.5 m and in Rincón de Guanabo, around 2 m, in both zones. In El Peruano, the depth is one m in the back crest zone and 2.5 m in the fore crest zone, while in Mariflores, the back crest zone has a depth of approximately one m and the fore crest zone about 1.8 m.

In Playa Baracoa, the experiment with A. palmata fragments was carried out in January 2022. The maximum heights and widths of the fragments were measured with a Vernier caliper at 0, 152, 279, and 453 days. In Rincón de Guanabo, the experiment began in January 2022, and the fragments were monitored at 0, 136, 261, and 433 days. In El Peruano, the experiment started in February 2022 and fragments were measured at 0, 172, 291, and 423 days. In Mariflores, the experiment began in February 2022, and the fragments were measured at 0, 168, 291, and 423 days. The number of fragments measured varied in each period, because some died, were detached from the substrate or were not found on the reef (due to human error) but were relocated in the following monitoring period.

The temperature was recorded every 30 min during the monitoring period using a data logger (HOBO UA-002-64 Pendant) deployed at Playa Baracoa and Rincón de Guanabo crests. In Jardines de la Reina, data were obtained from a logger located on a crest between El Peruano and Mariflores, as it was not possible to deploy loggers on each crest. The data logger records do not precisely align with the coral outplanting start date, either due to the unavailability of loggers at that time or because some units failed, resulting in the loss of temperature data. The mean monthly temperature was determined considering all records obtained every 30 min during each month.

Data analysis

Statistical analyses were conducted using the R program, created by the R Core Team (2016, version 4.0.5). The non-parametric Wilcoxon-Mann–Whitney test was used to test differences in the dissolution of the plaster discs between the crest zones. The median dissolution percentage was calculated from the four plaster discs in each zone. The Student t test (parametric) or the Wilcoxon-Mann–Whitney test (non-parametric) were used to determine differences in the initial size of the fragments between the back and fore crest zones. The Kruskal–Wallis test (non-parametric) and the Dunn post hoc test were used to determine differences in the initial sizes of the fragments between the four crests. A Cox regression model, coxph(Surv(survival time, survival) ∼ height + width, data = Data) was used to analyze whether the initial size of the fragments influenced survival, using the survival package (Therneau & Grambsch, 2000). The survival probability (s_p) of the fragments was evaluated using the nonparametric Kaplan–Meier estimator with the survminer package (Kassambara, Kosinski & Biecek, 2021). This estimator models survival probability over time by accounting for the exact timing of mortality events. It partitions the follow-up period into intervals defined by the occurrence of events (e.g., fragment mortality) and estimates the conditional survival probability for each interval. These conditional probabilities are then multiplied sequentially to obtain the cumulative survival function. This approach incorporates both time-to-event information and censored data.

The growth rate of the fragments was determined for both width and height during each study period, using the formula of Mercado-Molina, Ruiz-Diaz & Sabat (2014): ${\displaystyle \text{Growth rate}=\frac{\left(\text{initial size}-\text{final size}\right)}{\text{time}}}$ where:

initial size: is the height and/or width(cm) of fragment at the beginning of each period.

final size: is the height and/or width (cm) reached by fragments at the end of each period.

time: is the number of days included in each study period.

We tested for homogeneity of variance and normality using Levene’s and Shapiro–Wilk’s tests, respectively (R package nortest). The growth rate data did not fit a normal distribution, and we performed a square root transformation. The t-Student (parametric) test was used to determine differences in the growth rates between the back and fore crest zones. A generalized and linear mixed model was used to estimate the effect of: (1) the water flow represented by the dissolution of the plaster discs, (2) mean temperature in each period, (3) depth, (4) period, (5) crest zone and (6) crest site on fragment survival and growth rate, respectively with the lme4 package (Bates et al., 2015). For this model the reference level for the crest site and zone factors were Mariflores crest and the back zone respectively, with an estimate value of 9.4. A segmented model was used to identify dissolution values from which the effect on growth begins to change using the segmented package (Muggeo, 2008).

Dissolution of the plaster discs

The dissolution of the plaster discs was similar between the fore and back crest zones from the northwestern and the central regions of Cuba during April and June. In El Peruano and Mariflores reefs, the dissolution of the plaster discs varied significantly (p = 0.03) between crest zones, during December. In El Peruano, the dissolution was 38% and 49% in the fore and back crest zones, respectively. In Mariflores, the dissolution of the plaster discs was lower than in El Peruano, at 35% in the fore crest zone and 28% in the back crest zone (Table 1).

Initial size of the fragments

The initial mean width and height of the fragments ranged from two to four cm and was similar between the fore and back crest zones of the reefs, except in Mariflores where the initial width of the fragments varied significantly (p = 0.003) between zones (Table S1). The initial size varied significantly across the four reefs (Table S2). The Cox regression model did not show an effect of the initial size of the fragments on the survival of fragments in the four reefs.

Survival

Overall, the survival probability of fragments was high (s_p > 0.6) in the four crests. The survival was similar among the four fore crest zones ranging from 0.9 to 0.6. However, survival varied significantly (p = 0.03) among the back crest zones. In this case, Rincón de Guanabo had the lowest (s_p = 0.3) survival at 347 days in the back crest zone (Table 2, Fig. 3). The generalized linear mixed model did not show a significant effect of dissolution of the plaster discs, mean temperature, depth, or period on fragment survival. However, the survival was positively influenced by the four fore zones (estimate = 5.3; p = 0.02), and Playa Baracoa (estimate = 6.2; p = 0.002) and El Peruano (estimate = 12.9; p = 0.0002) crests, and the interaction between the fore crest zone and the Rincón de Guanabo crest (estimate = 6.9; p = 0.03).

Growth rates

The linear model showed that the growth rates were negatively affected (estimate = −6.1; p = 0.004) in the fore crest zone. The rate of increase in the height of the fragments was four-fold higher in the back crest zone in Playa Baracoa (back: 2.9 ± 2.9 cm year⁻¹, fore: 0.7 ± 2.2 cm year⁻¹, p = 0.03) and five-fold greater in Rincón de Guanabo (back: 3.7 ± 2.2 cm year⁻¹, fore: 0.7 ± 2.6 cm year⁻¹, p = 0.02) relative to the fore crest zone during the third period. In El Peruano, the rate of increase in the width of the fragments was six-fold greater in the back crest zone (2.2 ± 1.8 cm year⁻¹) with respect to the fore crest zone (0.4 ± 2.2 cm year⁻¹). Similarly, in Mariflores, the rate of increase in the width of the fragments of the back crest zone (3.3 ± 2.6 cm year⁻¹) was significantly higher (p = 0.03) than in the fore crest zone (1.1 ± 2.6 cm year⁻¹) during the first period. Also, in this crest the rate of increase in the width (back: 2.6 ± 1.8 cm year⁻¹, fore: −0.4 ± 2.6 cm year⁻¹, p = 0.0002) and height (back: 2.6 ± 1.8 cm year⁻¹, fore: −1.1 ± 2.6 cm year⁻¹, p = 0.001) of the fragments was greater in the back crest zone with respect to the fore crest zone for the second period. However, the growth tended to be higher (p > 0.05) in the fore crest zone for some monitoring periods, except in Playa Baracoa (p < 0.006) and El Peruano (p < 0.02) crests during the first and second periods respectively, where the differences in growth were significant (Fig. 4, Table S3). Additionally, the interaction fore crest zone in Rincón de Guanabo (estimate = 8.6; p = 0.02) and El Peruano (estimate = 5.3; p = 0.04) crests had a positive effect on growth rates of the fragments.

The dissolution of the plaster discs positively influenced (estimate = 0.03; p = 0.008) the growth rates, with values above 68% associated to increased growth of the fragments (Fig. 5). Temperature promoted growth (estimate = 0.002; p < 0.001) during the second period. However, the interaction between dissolution and temperature was negative (estimate = −0.001; p = 0.007). The depth effect on growth depends on the crest site and zone. The interactions depth and El Peruano crest was positive (estimate = 8.6; p = 0.002), while the interaction between depth and the fore zones was negative (estimate = −0.009; p = 0.04).

Temperature

The mean temperature increased from 26.1 ± 0.4 °C in January to 28.1 ± 0.7 °C in May during the first period, in Playa Baracoa and Rincón de Guanabo crests. During the second period (June to September 2022) the mean temperatures were high at 29 and 30 °C. The maximum recorded temperature was 33.6 °C in June. The third period extended from October 2022 to March 2023, when the temperatures decreased, with a minimum temperature recorded of 24 °C (Figs. 6A and 6B). In Jardines de la Reina crests, the temperature was only recorded in July 2022 during the first period, when the mean temperature was 30.6 ± 0.4 °C. During the second period (August to November 2022) the minimum temperature recorded was 28.6 °C in November and the maximum was 31.7 °C in September. During the third period (December 2022 to April 2023) the temperatures decreased till March. The minimum was 26 °C in February and March and the maximum was 32 °C in December (Fig. 6C). The average temperature was approximately 1 °C higher in Jardines de la Reina crests relative to the northwest crests.