Reef Structural Complexity Inﬂuences Fish Community Metrics on a Remote Oceanic Island: Serranilla Island, Seaﬂower Biosphere Reserve, Colombia

: Serranilla is a protected island of the Seaﬂower Biosphere Reserve, far from dense human population. These characteristics could help sustain structurally complex coral reefs, often associated with higher biodiversity, abundance, and biomass of reef-associated organisms, including reef ﬁsh. However, the multiple threats present in Serranilla, including intense illegal ﬁshing, can impact coral ecosystems generally and also speciﬁc key groups, such as the parrotﬁsh, in particular. During the “Seaﬂower Research Expedition 2017”, we assessed how structural habitat complexity inﬂuences reef ﬁsh assemblages. In addition, we explored differences in parrotﬁsh species (family: Scaridae) between Serranilla and San Andr é s, the most populated island in the Archipelago. On Serranilla, we found that habitat structure, rugosity, and coral cover accounted for up to 66% of variation in reef ﬁsh diversity, abundance, and biomass, with values being higher on more complex reefs. Parrotﬁsh species differed between the islands, with larger species supporting higher biomasses at Serranilla, by comparison with San Andr é s; however, the abundance, biomass, and lengths of parrotﬁsh species were low in both areas compared with those reported from other protected Caribbean reefs. Our study indicates that despite the evident relationship between structurally complex habitats and reef ﬁsh, other threats in Serranilla could be affecting parrotﬁsh populations, such as illegal ﬁshing, a widespread activity in the area.


Introduction
Coral reefs are widely recognized as one of the most diverse marine ecosystems, providing essential habitats for numerous species; although coral reefs cover only 0.02% of the ocean's surface, they are believed to harbor nearly 30% of all known marine species [1][2][3][4]. Reef ecosystems provide multiple ecosystem services [5][6][7], including coastal protection [8,9] and the provision of food and livelihoods through fishing activities [10]. However, in recent decades, coral reef ecosystems worldwide have been facing degradation due to multiple environmental and human stressors [11,12]. Currently,~60% of the world's coral reefs are considered to be threatened [13]. In the Caribbean, by 2003, at least 80% of the coral cover registered in the 1970s was lost [14,15], while in the Atlantic, 75% of coral reefs are considered at risk [12]. This loss of coral cover has often produced a decline in the reef's

Study Site
Serranilla island (15 • 50 N, 079 • 50 W), officially named Isla Cayos de Serranilla, is located at the northwest end of the Seaflower Biosphere Reserve (Figure 1). Seaflower BR, with an area of 180,000 km 2 , was declared as a Biosphere Reserve by UNESCO in 2000; within this larger area, the Colombian Government designated 65,000 km 2 as Marine Protected Area in 2005 [33]. In fact, Serranilla is not a single island, but a reef complex formed on a submarine mountain, on which emerged cays are scattered. The principal cay is Beacon Cay, located in the south east of the complex (Figure 1c). The underwater seascape consists of a carbonate platform with reef habitats such as reef crests, coral patches, sand flats, hardground areas, and also seagrass beds [34]. This island is located a long way (>400 km) from the most populated island of the archipelago, San Andrés, and lies only 320 km south of Jamaica (Figure 1a,b; Figure S1). San Andrés, in contrast, has a terrestrial area of 27 km 2 , with a maximum length of 12.6 km and maximum width of 3 km. On the west side of San Andrés there is a submarine hard bottom platform extending seawards for 200-500 m, and a mainly continuous rocky shore, but no barrier reef. On the east side of the island, the platform extends seawards for more than 2 km with some soft sand bottom areas, small coral patches, mangroves, and sandy beaches on the coast, and, at the east edge of the northern part, a discontinuous barrier reefs of about 9 km in length [33].
Oceans 2021, 2, FOR PEER REVIEW 27 km 2 , with a maximum length of 12.6 km and maximum width of 3 km. On th side of San Andrés there is a submarine hard bottom platform extending seawa 200-500 m, and a mainly continuous rocky shore, but no barrier reef. On the east the island, the platform extends seawards for more than 2 km with some soft sand areas, small coral patches, mangroves, and sandy beaches on the coast, and, at t edge of the northern part, a discontinuous barrier reefs of about 9 km in length [33 including San Andrés, Old Providence, and Serranilla island, and (c) the locations of the survey sites at Serranilla (red points: S1-S10) together with (inserted) an aerial image of Beacon Cay, the largest cay on Serranilla (top left).

Data Collection
The surveys were conducted at ten sites at Serranilla during the "Seaflower Research Expedition 2017". Ten sites were selected to capture the heterogeneity of Serranilla (Table S1). At each site, habitat structure (benthic components and rugosity) and fish assemblage structure and composition were assessed using underwater visual census (UVC) techniques conducted by a paired-diver team along 50 m × 2 m belt transects (n per site = 2 or 3). On each transect, the fish species, abundance, and visual estimated size per individual (interval length) were recorded by the first diver, following established sampling methodologies [35][36][37]. To highlight the specific relationships of the key fish family of parrotfish (Scaridae) with habitat, the abundance of parrotfish according to the size was also calculated; we distinguished large species (>50.1 cm total length) and small species (<50 cm total length) based on available values from FishBase [38]. Valid species names were verified, according to Fricke et al. [39].
Habitat structure was evaluated as a combination of the percentage cover of the main benthic components and of reef rugosity. To characterize the coral community, we selected the percentage cover of five morpho-functional benthic groups: branching corals, brain corals, submassive corals, spherical corals, and algae. The variables were recorded with a video camera that was held 0.4 m above the seabed while being moved along each of the 50 transects. From each such video transect, 40 frames were selected, and 50 points in each frame were sampled to estimate the percentage of the benthic groups. Rugosity was estimated with the rugosity index (RI), obtained using the chain-link method [40]. Within each belt transect, the second diver laid down a 10 m chain over the seabed closely following the substrate contour and then measured the linear distance occupied by the chain. This was done three times within each transect. RI was then calculated as RI = 1 − d/L, where d is the horizontal distance covered by the chain, and L is the chain's true (i.e., stretched) length. This method is quick, objective, and highly replicable [36].
Fish data for San Andrés were obtained during the same sampling period as the Serranilla surveys (September 2017). Data were collected from transects at three sites on the west side (Luna Verde, Wild Life and Bajo Bonito) and one on the east side (Bahía Honda) of the island. Fish species, abundance, and estimated fish size in San Andrés were recorded following the same UVC described above for Serranilla. To compare parrotfish, we selected 12 transects for each location (Serranilla and San Andrés), all placed parallel to the coast, with comparable values of RI (0.19-0.49).

Analysis
We used two approaches to analyze the reef fish assemblages at Serranilla and San Andrés. The first approach aimed to characterize the relationship of fish assemblages to habitat structure. The second focused on the parrotfish assemblage, evaluating differences between Serranilla and San Andrés.

Relationship of Reef Fish Assemblages to Habitat Structure
For each transect, we estimated diversity, abundance, and biomass. Biomass (g m −2 ) was calculated using the allometric weight equation: where L is the weighted mean of the estimated length for each species per transect, and a and b are the constants of length-weight obtained from FishBase [38]. We evaluated diversity using Hill's effective number of species of order 0, 1, and 2 [41], where 0 D is species richness, 1 D considers all species and their abundance, 2 D reflects the most abundant species, and 2/1 D reflects their evenness. To explore the relationships of reef fish assemblages to habitat, we performed multiple linear regressions of fish assemblage measures against the main explanatory variables that define habitat structure (benthic cover and rugosity). The preferred models were selected based on the lowest values of Akaike's Information Criterion (AICc). The models' variables included the average parameters of the models with <2 ∆AICc [42]. The p-value and adjusted R 2 of the model with all the explanatory variables and the selected variables were reported. For each model, the validity of the linear models was examined with the normality of the residuals [43].

Differences in Parrotfish Assemblages between Islands (Serranilla and San Andrés)
We evaluated variation in the metrics ( 0 D, biomass, and abundance) and composition (abundance and biomass) of the Scaridae species present between the two islands, Serranilla and San Andrés. We also examined the mean sizes observed for each species at Serranilla and contrasted these with the available information about common and maximum size obtained from FishBase [38].
Two ANOVAs based on permutations (10,000) were performed to test the differences in biomass and abundance using Euclidean matrices from fourth-root transformed data. We used a model with one factor: where µ is the general mean; Island i is the factor with two levels (Serranilla and San Andrés) and 12 replicates per site, and e ij is the associated error. Two permutational multivariate analyses of variance (PERMANOVA) were performed to assess parrotfish assemblage composition differences in biomass and abundance. We used the previous ANOVA model and verified the homogeneity of dispersion with the PERMDISP test and non-metric multidimensional scaling (NMDS) [44]. Similarity percentage analysis (SIMPER) were used to detect the parrotfish responsible for the dissimilarities between the islands. We plotted in the NMDS the small and large parrotfish abundance and biomass to visualize the differences in composition. For PERMANOVA, PERMDISP, SIM-PER, and NMDS, we used a Bray-Curtis similarity matrix with a fourth root transformed data. Statistical analyses were performed in R and Primer v6.1 PERMANOVA+ [45,46], and plots for SIMPER results generated in SigmaPlot v11 software.

Results
A total of 8137 individual fish belonging to 68 species and 22 families were recorded across the sites on Serranilla. The families represented by the most species were: Scaridae (12 spp), Pomacentridae (9 spp), and Labridae (7 spp). Three wrasses (Labridae) were the most abundant species, representing 50% of the total fish abundance: Halichoeres bivittatus, H. garnoti, and Thalassoma bifasciatum. However, because of the larger sizes of individuals, the highest mean biomasses were represented by Haemulon album, Melichthys niger, and Mulloidichthys martinicus, which together contributed >30% of the total biomass (Table S2).

Relationship of Reef Fish Assemblages to Habitat Structure
The results of the multiple linear regressions supported the interpretation that at Serranilla overall fish abundance, biomass, and diversity were related to habitat structure (Table 1). Up to 66% of the variation in measures of reef fish assemblage structure was significantly explained, except for variation in biomass and evenness. Among all the measured explanatory variables, rugosity, and coral cover of branching and brain corals were most often significant. Of the diversity indices, species richness ( 0 D) was the measure with the greatest percentage of variation described (59%), in that case by only one variable (rugosity). In contrast, 66% of variation in the abundance of small parrotfish species was explained by the combination of all six measures of habitat structure (Table 1).

Parrotfish Differences between Islands (Serranilla and San Andrés)
We recorded 11 parrotfish species present within one or both of the two islands, 11 on Serranilla and six on San Andrés (Scarus coelestinus, S. coeruleus, S. guacamaia, S. iseri, S. taeniopterus, S. vetula, Sparisoma atomarium, S. aurofrenatum, S. chrysopterum, S. rubripinne, S. viride). S. viride was the species with both the highest abundance and biomass on Serranilla, and on San Andrés, S. taeniopterus and S. aurofrenatum had the highest abundance and biomass, respectively (Table S3). Comparison of metrics between Serranilla and San Andrés showed that there was no statistical difference in overall abundance (Pseudo-F = 2.137, p = 0.146) or species richness per transect ( 0 D; Pseudo-F = 2.057, p = 0.105) between the two island areas, but there was a statistically significant difference in parrotfish biomass per transect (Pseudo-F = 4.637; p = 0.040).
Nevertheless, parrotfish assemblage composition did appear significantly different between the two islands when examined both in terms of abundance (Pseudo-F = 16.031, p < 0.001) and in terms of biomass (Pseudo-F = 12.170, p < 0.001) with NMDS plots; these showed, in relation to both measures, a strong spatial segregation between sites on the two islands and a notably higher dispersion among the Serranilla sites (Figure 2a,b). As shown by the sizes of the bubbles, at Serranilla, most of the parrotfish contributing to abundance and biomass were larger species, whereas at San Andrés these were mainly small-size parrotfish (Figure 2c-f).
The SIMPER analysis showed that just three species were responsible for >50% of island dissimilarity, two of which had higher abundance and biomass in San Andrés (S. taeniopterus and S. aurofrenatum), and a third which had a considerably higher abundance in Serranilla (S. viride) (Figure 3; Table S4). Overall, the mean parrotfish size of each species was considerably smaller than the common and maximum lengths reported in FishBase ( Figure S2). The SIMPER analysis showed that just three species were responsible for >50% of island dissimilarity, two of which had higher abundance and biomass in San Andrés (S. taeniopterus and S. aurofrenatum), and a third which had a considerably higher abundance in Serranilla (S. viride) (Figure 3; Table S4). Overall, the mean parrotfish size of each species was considerably smaller than the common and maximum lengths reported in FishBase ( Figure S2).

Discussion
Our study demonstrates that variation in reef fish diversity, biomass, and abundance in Serranilla can be partly explained by two benthic components, coral cover and rugosity, that contribute to habitat complexity in coral reefs [18,19]. Nevertheless, the mean values of parrotfish biomass and length at Serranilla were low, which could reflect the effect of anthropogenic pressure in the area, mainly through illegal fishing [30][31][32].
Structural complexity in coral reefs has been evaluated for its ecological relevance, enhancing abundance, biomass, and biodiversity [47,48]. Multiple studies have found that the complexity of healthy coral reefs favors coastal protection, reducing the effects of wave action and other environmental phenomena and maintaining shoreline equilibrium [9,49,50]. A relationship of fish assemblage with the structural complexity, such as we observed at Serranilla in the Seaflower BR, has also been observed at other locations in the Atlantic [51,52] and in other regions of the world [53,54].
Reef habitat structures provide microhabitats and refuge areas that benefit multiple organisms, including reef fish [55,56], with different rugosity levels favoring multiple fish groups [52]. Loss of habitat complexity can negatively affect multiple species [47,57], including reef fish [58], leading to species losses [11] and localized extinctions [59], including species of commercial interest such as parrotfish. Rogers et al. [52] predicted that a loss of complexity could cause a more than three-fold reduction in fishery productivity. Therefore, preserving complexity in coral reefs is crucial, particularly since structural complexity is currently in rapid decline in the Caribbean [17], with one of the largest protected areas in the Caribbean, the Seaflower BR, remaining highly vulnerable to anthropogenic pressure [33].

Discussion
Our study demonstrates that variation in reef fish diversity, biomass, and abundance in Serranilla can be partly explained by two benthic components, coral cover and rugosity, that contribute to habitat complexity in coral reefs [18,19]. Nevertheless, the mean values of parrotfish biomass and length at Serranilla were low, which could reflect the effect of anthropogenic pressure in the area, mainly through illegal fishing [30][31][32].
Structural complexity in coral reefs has been evaluated for its ecological relevance, enhancing abundance, biomass, and biodiversity [47,48]. Multiple studies have found that the complexity of healthy coral reefs favors coastal protection, reducing the effects of wave action and other environmental phenomena and maintaining shoreline equilibrium [9,49,50]. A relationship of fish assemblage with the structural complexity, such as we observed at Serranilla in the Seaflower BR, has also been observed at other locations in the Atlantic [51,52] and in other regions of the world [53,54].
Reef habitat structures provide microhabitats and refuge areas that benefit multiple organisms, including reef fish [55,56], with different rugosity levels favoring multiple fish groups [52]. Loss of habitat complexity can negatively affect multiple species [47,57], including reef fish [58], leading to species losses [11] and localized extinctions [59], including species of commercial interest such as parrotfish. Rogers et al. [52] predicted that a loss of complexity could cause a more than three-fold reduction in fishery productivity. Therefore, preserving complexity in coral reefs is crucial, particularly since structural complexity is currently in rapid decline in the Caribbean [17], with one of the largest protected areas in the Caribbean, the Seaflower BR, remaining highly vulnerable to anthropogenic pressure [33].
Coral reef degradation may also result from the loss of critical groups, such as the parrotfish. Their loss, through increased in macroalgae, decreased coral cover, and subsequently a diminished habitat complexity, may lead to a decline in associated organisms [22,60]. At Serranilla it was found that transects closer to Beacon Cay showed lower rugosity and diversity of reef fish. Although greater habitat complexity may have helped maintain higher abundance and biomass on Serranilla's reefs compared with those on San Andrés, direct anthropogenic pressure may still limit any positive effect, and lead to a lower abundance and diversity of reef fish, including parrotfish [22,46,47,52].
Anthropogenic pressure is among the most recurrent threats to coastal ecosystems, since human populations reside nearby and depend on coastal areas for commerce and survival [61,62]. Studies have found a relationship between reef fish assemblages and the distance from dense human populations [63], and found that sites closer to human populations tend to have lower abundance, biomass, and richness of reef fish than areas that are more remote, with an exponential increase in reef fish biomass as the distance from the significant human population increases [64,65].
Considering that Serranilla is far from any significant human population, we expected to find higher biomasses of reef fish, particularly of key coral reef groups such as the parrotfish. Beacon Cay, Serranilla, has only around 12 permanent inhabitants (equivalent to 9.2 inhabitants per 0.1 km 2 ), while San Andrés has more than 65,000 (equivalent to 241 inhabitants per 0.1 km 2 [33]. Additionally, more than 730,000 tourists visit San Andrés each year [65]. However, our data showed that despite the high reef rugosity, most parrotfish at Serranilla had low biomass and abundance values, similar to those found at San Andrés.
Differences in the size and abundance of some parrotfish species between Serranilla and San Andrés seem likely due to a lower fishing pressure at Serranilla. Nevertheless, fishing pressure may have modified the parrotfish assemblages at Serranilla. On San Andrés, fishing is one of the most important commercial activities, and includes spearfishing targeting multiple parrotfish species, such as S. coeruleus, S. coelestinus, and S. guacamaia [66], that were rarely observed during this study. While permitted fishing activity is mainly artisanal, illegal vessels have been reported carrying tons of illegally caught parrotfish [67]. In addition, Bruckner [34] reported illegal fishing boats at Serranilla island during the "Global Reef Expedition" in 2011, and such boats were also observed during the present "Seaflower Research Expedition" in 2017. Fishers have in recent years expressed concern over the absence of large parrotfish in the lagoon as a result of overfishing [68].
Herbivorous fish biomass has been used as one of the indicators for evaluating reef health. In the report on Mesoamerican reefs by the Healthy Reefs Initiative, they recorded 270 kg ha −1 of herbivorous fish on the healthier reefs compared with 83 kg ha −1 on reefs in critical conditions [69]. In particular, a relationship has been reported between fishing pressure and parrotfish biomass and size, making parrotfish abundance a useful biological indicator of fishing pressure [63]. On a small island in the Caribbean, exploited reefs, with areas open to artisanal fisheries, were found to have a parrotfish biomass of around 320 kg ha −1 , with smaller parrotfish contributing most of this biomass (~200 kg ha −1 ).
We found that Serranilla BR had larger parrotfish than San Andrés, though the biomass in Serranilla BR was nevertheless lower (143 kg·ha −1 ) than those values reported from other Caribbean MPAs with adequate protection and low local human populations, in which parrotfish biomass values can exceed 500 kg·ha −1 [70]. In contrast, in areas with high local human population densities and inadequate protection, parrotfish biomass can be rapidly reduced to 50 kg·ha −1 or less [22,29]. However, other factors may also influence herbivorous fish abundance; for example, on the Alacranes reef, an MPA in the southern part of the Gulf of Mexico, where human activities are restricted, and fishing pressure is low, the values of parrotfish biomass were nevertheless relatively low (~58-128 kg h −1 ), although the parrotfish present had larger sizes than those observed on Caribbean reefs [71].
As with many other species, small individuals have become a target of fisheries due to the loss of larger individuals and the decline of other commercial fish species more generally, leading to decreased sizes [67]. Parrotfish size is likely related to reef resilience [23,57], since larger individuals can remove more algae from the reef and have more effect on benthic communities than smaller ones [24,29]. In the present study, we found that 90% of the parrotfish species at Serranilla had smaller mean sizes than the typical size reported in FishBase [38]. Notably, the observed mean size of S. viride and S. chrysopterum were both below the size at first maturity (L50), as recorded in the Colombian Caribbean [72], suggesting a higher fishing pressure at Serranilla than expected.

Conclusions
This study on Serranilla island, which is part of the Seaflower BR, has demonstrated that the overall reef fish assemblage is spatially variable with habitat structure explaining up to 66% of this variation [53,57]. In addition, focusing on the parrotfish and their sizes has allowed us to compare sizes and biomasses of fish on Serranilla with those on San Andrés. In combination, the results presented here, together with the intense illegal fishing activities observed in the reserve and other drivers not considered in this study, highlight the need to evaluate other factors that may influence reef fish biodiversity beyond structural complexity [26,73]. Meanwhile biological and ecological information about the current status of most of the key groups in the Seaflower BR remains scarce [67]. Maintaining healthy reef ecosystems is becoming critical, especially in areas where human well-being depends on coral reef ecosystem services, such as food production [73], so that the need for effective management and active surveillance at the Serranilla MPA is urgent.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/oceans2030034/s1, Figure S1: Serranilla Island and distances between islands (San Andres 415 km, Providencia 309 km, Jamaica 305 km), Figure S2: Mean observed fish sizes during census at Serranilla Island (SI) compared with the common and maximum (Max) reported in FishBase, Table S1: Site summary of the metrics of fish assemblages in Serranilla Island with average value per metric ± SD, Table S2: Summary of fish assemblage families and species in Serranilla Island. Values of abundance and biomass with the total sum, frequency, and average ± SD, Table S3: Summary of the abundance and biomass of the Scaridae species in Serranilla and San Andrés with the average values ± SD, Table S4: SIMPER analysis summary. Scaridae average contribution and cumulative contribution percentage in biomass and abundance to the dissimilarity between SI and SAI. Registered and common length of those species.  The project under which this research was conducted is "Valoración de servicios ecosistémicos de los arrecifes de coral en los alrededores de la Isla Cayo Serranilla, Reserva de Biosfera Seaflower, Caribe Colombiano," developed and supported by Universidad Nacional de Colombia Sede Caribe. Additional funding included Colciencias Ph.D. scholarship (Conv. 757) and CEMarin (Call 14, 2018) that funded the maintenance of the CEMarin young researcher Julián Prato and the project "Relationships between coral reef complexity and ecosystem services at Caribbean oceanic islands, Seaflower Biosphere Reserve, Colombia". Universidad Nacional de Colombia Sede Caribe funded fish census at San Andrés to Diana Castaño MSc. thesis project "Estructura y Función de peces herbívoros en zonas arrecifales de San Andrés una isla oceánica en el Caribe".

Data Availability Statement:
The datasets generated during the current study are available within the article and its supplementary materials. Additional data are available from the authors on reasonable request.