Rapid relative increase of crustose coralline algae following herbivore exclusion in a reef of El Salvador

Study site

The study was carried out on Los Cóbanos reef (13°31′25.6″N 89°48′24.6″W), El Salvador, from March to May 2018. The reef lies within the nature reserve “Complejo Los Cóbanos”, 11 km east of the city Acajutla (Fig. 1). The Ministry of Environment and Natural Resources of El Salvador (MARN, initials in Spanish) approved the fieldwork for this study inside the nature reserve. Los Cóbanos reef consists of a heterogeneous basalt shore with a tidal variation of ∼3 m (Segovia, 2016). Colonies of the hermatypic coral species Porites lobata grow inter- and subtidally, covering only 2–7% of the benthos. The region is characterized by two seasons: a dry (December to May) and a rainy (June to November) season. The benthic community at Los Cóbanos is exposed to sedimentation impacts during the rainy season, when river runoff brings sediments and nutrients to the reef, significantly increasing water turbidity.

Benthic community survey

The benthic community was evaluated using the Line-Point-Intercept-Transect method via snorkelling, as described by English, Wilkinson & Baker (1997). In April 2018, seven 50 m long transects were placed parallel to the coastline at 2–3 m water depth, with at least 25 m between individual transects. During low tide, for a total of 100 points per transect (one every 0.5 m), the benthic organism underneath each point was identified in situ to genus level, except for turf and crustose coralline algae (CCA) that could not be identified to this level and were classified generically. If no organism was found on that point, non-living structures were categorized as: ‘sand’, ‘rock’ or ‘Pocillopora sp. skeleton’. The relative benthic cover of each category was calculated using the resulting 100 points per transect.

Experimental cage set-up

The experimental design followed the Before-After Control-Impact/Treatment (BACI) design (Stewart-Oaten, Murdoch & Parker, 1986). Four 70 × 70 × 50 cm³ cages constructed from 2.5 × 2.5 cm² galvanized wire mesh (as used by Smith, Smith & Hunter, 2001) were deployed on the reef with at least 2 m distance from each other. Each cage enclosed one Porites lobata colony of a diameter of 8–12 cm and surrounding algae. Even though P. lobata only covered up to 4% of the benthos, it was targeted in the experiment to assess how simulated overfishing affects coral-algae interactions. Four more plots with a similar benthic community composition to the enclosed ones were selected as controls (70 × 70 cm²). The experiment ran for seven weeks, from April to May 2018. Each week the cages were cleaned using a plastic washing brush to avoid algae growth. The plots were inspected for small fish, snails, crustaceans and echinoderms every week during the cleaning. None were observed in the enclosed plots. Also, the control plots were carefully inspected weekly, but no macroinvertebrates were observed at any time. The cover of the different algae functional groups in the enclosed and control plots was estimated in situ using quadrats that indicated the cage limits (70 × 70 cm²) to the nearest 1% at the beginning and the end of the experiment. The same categories used during the benthic surveys were identified in the exclusion experiment. Only the uppermost benthic community layers were analyzed. The cover of P. lobata was estimated by comparing the projected area of the colony (calculated by measuring the length and width of the colony at the beginning and end of the experiment) to the total selected cage and control area (0.49 m²). All surveys were conducted by the first author.

Due to logistical constraints, it was not possible to include semi-closed cages to assess potential physical effects of the cages in this experiment. CCA often perform better under lower light availability than other algae (Van den Hoek et al., 1978; Vásquez-Elizondo & Enríquez, 2017), while algal turfs increase their net primary productivity with increased water flow (Carpenter, Hackney & Adey, 1991; Carpenter & Williams, 2007). The cage structure could have reduced light availability and water flow within the enclosed plots, benefiting CCA. In order to quantify the potential effects of the cages on water flow and light attenuation, gypsum cards and HOBO light sensors were placed inside and outside cages with the same design as those used in the experiments, at the experimental sites for 24 h in April 2019. The gypsum cards were weighed before and after 24 h exposure to obtain an aggregate measure of water flow over that period. The HOBO light sensors measured light intensity within and outside the cages in lum/ft⁻² every 30 s.

Data analysis

The mean relative cover of each functional group identified in the benthic community survey was calculated using the relative cover of all transects. To compare the community composition of the control and enclosed plots at the beginning and end of the experiment, Bray–Curtis Dissimilarity was calculated on untransformed data, and a PERMANOVA test was conducted using PRIMER 7. For the pair-wise comparisons, Monte Carlo P-values were obtained using PRIMER 7 due to the small number of permutations resulting from the PERMANOVA test, as suggested by Anderson, Gorley & Clarke (2008). All additional tests were conducted in the statistics program R version 3.5.1 (R Core Team, 2018). Differences in coral, CCA, and turf benthic cover were tested with repeated measures ANOVA (rmANOVA from the ez package, Lawrence, 2016). In order to test the weight difference of the gypsum cards after the 24 h exposure, a t-test (from the stats package) was conducted. To test for potential differences in the measured light intensities, a Wilcoxon Rank Sum test (stats package) was used, as normality and sphericity assumptions were not met. An Anderson-Darling-Test (ADGofTest package) was used to test for data normality and Levene’s-test (from the car package) was used to test sphericity assumptions. The jitter function (from ggplot2 package, Wickham et al., 2019) was used to add random variation to the non-metric multidimensional scaling (nMDS) plot and reveal the points with the same community assemblage that were overlapping.

Benthic community survey

The benthic community composition of Los Cóbanos reef was dominated by algae. Different types of algae comprised 72–98% of the benthic community. Turf algae were the most dominant algae group, with a mean benthic cover of 26.6 ± 8.8 (SD) %, followed by green algae with 22.8 ± 21.2 (SD)  % (Table 1). Only one hermatypic coral species was found alive (Porites lobata), which had a benthic cover of 2.0 ± 2.8 (SD) %. Calcium carbonate skeletons of the branching coral genus Pocillopora were widely observed. Mobile benthic macroinvertebrates were not observed within the transects; however, the nudibranchs Elysia diomedea and Glossodoris sedan, and the echinoderms Ophiocoma aethiops, Echinometra vanbrunti, and Holothuria (Halodeima) kefersteinii were observed on the reef during the benthic surveys (Table S1).

Fish herbivore exclusion

No significant effect of cage structures was observed on either light intensity (Wilcoxon Rank Sum test, p = 0.520) or water flow (t-test, t (_2.08) = − 1.21, p = 0.343), indicating that the physical structure of the cages did not affect light or water flow.

There was a significant interaction effect of treatment and time on the benthic community composition (PERMANOVA, P(perms) = 0.001, perms = 996). The benthic community composition in the enclosed and control plots only differed significantly at the end of the experiment (PERMANOVA, P(perms) = 0.027, perms = 35, P(MC) = 0.007) (Fig. 2, Tables S2 and S3). The benthic community composition in the enclosed (PERMANOVA, P(perms) = 0.030, perms = 35, P(MC) = 0.001) and control (PERMANOVA, P(perms) = 0.028, perms = 35, P(MC) = 0.003) plots changed significantly between the beginning and the end of the experiment. Crustose coralline algae cover increased by a total 57.5% (from 0 to 57.5 ± 9.6 (SD) %) in the enclosed plots and 21.5% (from 1 ± 1.4 (SD) % to 22.5 ± 8.66 (SD) %) in the control plots, while the turf algae benthic cover decreased by 81% (from 83.8 ± 4.7 (SD) % to 2 ± 2.3 (SD) %) in the enclosed plots and 63% (from 79.75 ± 4.3 (SD) % to 16.5 ± 7 (SD) %) in the controls (Fig. 3). There was a significant interactive effect of treatment and time on both the CCA (rmANOVA, F_(1,6) = 27.47, p = 0.003) and turf algae benthic cover (rmANOVA, F_(1,6) = 14.69, p = 0.009).

Benthic community survey

This study aimed to describe the current state of the benthic community of Los Cóbanos reef, El Salvador, and to investigate the effect of piscine macroherbivore exclusion on benthic community composition. As expected (H1), the survey results show that the reef was dominated by algae (mostly turf and green algae) and that the coral cover was extremely low. Compared to the coral diversity described for the same reef by Reyes-Bonilla & Barraza (2003), the number of living hermatypic coral species present decreased from five to one in less than two decades. Past unpublished surveys conducted at Los Cóbanos yielded similar results; however, the reported algae cover was slightly lower two years prior the present study (∼72%) (Herrera, 2017). Yearly surveys conducted by J. Segovia using permanent transects reported a coral cover decrease from 6% to 2% between 2014 and 2016. Segovia, Trejo & Ramos (2019) and Segovia (2020), using haphazardly deployed transects, documented a P. lobata cover of 2% in 2019 and 2020, in line with the value observed in this study for 2018. In contrast to the present study, the benthic community in Los Cóbanos has been mainly surveyed using 30-meter long transects at three different distances from the shore and a 1 m² quadrat (Segovia, 2016; Segovia, Trejo & Ramos, 2019; Segovia, 2020). Regardless of the difference in the monitoring methodology, the results of the different surveys yield very similar results, indicating a clear decrease of hard coral cover over time up to 2016, mainly attributed to El Niño events in 2014 and 2016 (Segovia, Trejo & Ramos, 2019). Moreover, according to personal communication with fishers from the local community, 20–30 years ago, Pocillopora sp. cover in the shallow parts of the reef was high. According to the fishers, the only way to access the reef was during high tide with the help of boats, suggesting a much higher coral cover and a higher hard coral diversity than the present one. A vast number of calcium carbonate skeletons, some of them still attached to the substrate, belonging to the coral genus Pocillopora were observed during the surveys, indicating this genus comprised a significant component of the shallow benthic community in the recent past.

Studies of other reefs in the ETP showed a higher hard coral species richness (2–3 coral species) than at Los Cóbanos (Guzmán & Cortés, 1993; Alvarado et al., 2010; Stuhldreier et al., 2015). Turf algae also seem to dominate other ETP reefs (Cortés et al., 2017; Stuhldreier et al., 2015). Yet, these reefs display a lower algae diversity, with only two or three macroalgae genera dominating the reef benthos (Stuhldreier et al., 2015), compared to nine different genera found at Los Cóbanos.

Cortés et al. (2017) state that the increasing temperature and length of El Niño events, sedimentation, and local human activities have caused reef degradation in the region. The coral cover of Los Cóbanos reef decreased drastically in the last decades. Two oil spills in El Salvador in 1993 together with extreme El Niño events may have caused the disappearance of most of the coral community of Los Cóbanos (Cortés et al., 2017; Molina, 1996; Alvarado, 2012; Segovia, Trejo & Ramos, 2019). Currently, the reef is exposed to high nutrient input from the local human population, intense fishing, and sedimentation (Herrera, 2017, unpublished; Reyes-Bonilla & Barraza, 2003). The hard coral population is limited to a small area at 0–15 m water depth extending for about 7,000 m along the coastline between Punta Remedios and Acajutla (Fig. 1) (pers. obs. J Segovia). The results of this study, together with the aforementioned literature, suggest that Los Cóbanos reef may have undergone a shift from a reef with high coral cover towards a rather algae-dominated reef, as described by Pandolfi et al. (2005) for other regions of the world, at least in the first 600 m from the coast. Bellwood et al. (2004) defined coral reefs as “three-dimensional shallow-water structures dominated by scleractinian corals”. The low coral cover and high algae cover and diversity underline that Los Cóbanos is no longer a coral reef but rather an algal-dominated reef. Concordantly, recent literature refers to Los Cóbanos reef as rather a rocky reef featuring hard coral and algal communities (Herrera, 2017, unpublished; Segovia, 2017). There is, however, no historical monitoring data that verifies the assumption that Los Cóbanos reef was indeed a coral reef in the past. Core drilling and examination of the underlying matrix might be able to resolve this question.

Herbivore exclusion

Unexpectedly, CCA cover increased in both enclosed (by a total of ca. 58%) and control areas (ca. 21%), while turf algae decreased in both treatments (enclosed ca. 82%, control ca. 63%) at a remarkable speed. As CCA are among the slowest-growing marine algae, it is unlikely that this result reflects actual growth of CCA. Rather, at least some CCA may have been covered by turf-forming fouling epiphytes, the removal of which would have led to an apparent short-term increase in CCA. CCA are often considered as subordinate in their capacity to compete for space and are often overgrown or shaded by turf or macroalgae (Dethier, 1994; Littler & Littler, 1980). In some cases, this overgrowth even provides protection to CCA from harmful environmental conditions (Figueiredo, Kain & Norton, 2000). However, our results cannot confirm this, as only the uppermost layer of the benthic community was analyzed at the beginning and end of the experiment. No observations were made on whether CCA could indeed be found living under the turf algae. The exclusion of large piscine herbivores through caging had a significant effect on the benthic community composition. Surprisingly, there was a significantly higher apparent increase in CCA cover in the enclosed areas, whereas turf algae decreased more in the absence of piscine macroherbivores (Fig. 3). This is remarkable, as most literature indicates that herbivore exclusion causes an increase of macro- and turf algae (e.g., Hughes et al., 2007; Roth et al., 2015; Thacker, Ginsburg & Paul, 2001; Zaneveld et al., 2016). No effect of the cage structures on the current regime or light availability was detected through the 24 h measurements, meaning that the stronger observed benthic community changes in the enclosed plots were presumably caused by biotic factors. Nevertheless, the extent to which these results can be transferred to the seven-week experimental period is limited. High frequency data loggers were used to monitor light changes over a 24 h period on a day that was representative for the study period during the dry season. However, these measurements may not be representative for the study period which marked a transition from the dry to rainy season. Our study thus is not able to directly determine the biotic and abiotic processes behind the phenomenon observed at Los Cóbanos. Therefore, potential explanations for these counterintuitive results are discussed in the following paragraphs.