Introduction

Coral reefs are among the most biodiverse and productive ecosystems in the world (Hoegh-Guldberg 1999). However, their prevalence has dramatically declined over recent decades (e.g., Pandolfi et al. 2003; Burke and Maidens 2004; Schutte et al. 2010; Burke et al. 2011; Eddy et al. 2021). This decline has been largely attributed to the effects of various stressors, both natural and anthropogenic (Gardner et al. 2003; Carpenter et al. 2008; Vega Thurber et al. 2014). Natural stressors often function as transitional disturbances and are recognized as a component of reef growth and development, including events such as hurricanes that can alleviate thermal stress (Heron et al. 2008) or facilitate propagation (Vroom et al. 2005). However, many of the human-induced stressors impacting coral reefs are considered chronic and do not allow sufficient time or appropriate conditions for coral reef-associated organisms to recover, resulting in the loss of key ecosystem components (Spalding and Brown 2015). Notably, since the 1980’s, increases in sea water temperatures linked to global climate change have triggered mass coral bleaching events, including three widespread occurrences in 1998, 2010 and 2015–2016 that affected all tropical regions within both hemispheres (Hughes et al. 2017).

Within the Caribbean Sea, the Mesoamerican Barrier Reef System (MBRS) extends for ~ 1000 km from Mexico to Guatemala, Belize and Honduras and is the largest coral reef system in the Northern Hemisphere (Chollett et al. 2017). This system provides subsistence for many people within these countries (Gress et al. 2019) but has been heavily impacted by multiple stressors, including overfishing, coral disease outbreaks, hurricanes, coral bleaching, sedimentation, and land-based pollution (Mumby et al. 2007; Smith et al. 2008; Pandolfi 2010; Perry et al. 2013; Jackson et al. 2014; Muñiz-Castillo et al. 2019; Cáceres et al. 2020; França et al. 2020). Over the last 30 years, the average estimated coral cover in this region declined from approximately 50% to 10% (Gardner et al. 2003; Cramer et al. 2020). In particular, the precipitous decline of the three acroporid corals, Acropora cervicornis, A. palmata and A. prolifera, considered the building blocks of Caribbean coral reefs, and the mass mortality of the sea urchin, Diadema antillarum, resulted in a dramatic reduction within three-dimensional reef complexity (Carpenter 1988; Alvarez-Filip et al. 2009; Alvarez-Filip, et al. 2011a, b). These events are thought to have created phase shifts in the composition of the benthic community, moving from hard (scleractinian) coral as the dominant organism to octocorals, sponges, and/or macroalgae, with cascading impacts on the biodiversity and productivity across this region (Carpenter 1988; Hughes 1994; Miller et al. 1999; Aronson et al. 2002; Patterson et al. 2002; Burke and Maidens 2004; Maliao et al. 2008; Wulff 2012; Smith et al. 2016a, b).

Given the alarming rate at which coral ecosystems around the world are declining, coral reef surveys represent a fundamental first step to assess the overall health of the reef ecosystem by measuring benthic cover, species diversity, biomass, and surface complexity. Annual trends in these measures serve as a tool to evaluate reef resistance and resilience to disturbances and to identify areas of particular concern where management strategies need to be implemented. Although the Caribbean region is one of the most extensively studied (Hughes 1994; Burke and Maidens 2004; Jackson et al. 2014), a lack of long-term monitoring in many locations makes it difficult to estimate benthic population and cover trends and develop effective models. In this study, we present results from a decade of surveys of shallow coral reefs off of the southern coast of Roatán, Honduras, providing information on changes that occurred in percentage coverage of benthic invertebrates over a ten-year period. These data represent the first long-term monitoring for this location and provide a baseline for future studies.

Materials and methods

Study sites

Las Islas de la Bahia (The Bay Islands) within the MBRS lie ~ 65 km off the northern shoreline of Honduras, with Roatán being the largest (approximately 60 km long and 8 km wide) and most populated of the three main islands (Biggs and Olden 2011) (Fig. 1a, b). The islands of Roatán, Guanaja, and Utila are part of the Bonacca Ridge, a mountainous uplift that is thought to have formed along the tectonic plate boundary between the North American and Caribbean plates and separates the region from the Cayman Trough to the North (Sutton 2015). Sea-surface temperatures range from 26 to 30 °C and salinity between 35 and 35.5 PSU (Heyman and Kjerfve 2001; Brenes et al. 2009; Brenes et al. 2017), while processes such as sedimentation, eutrophication, larval dispersion and recruitment are influenced by the cyclonic gyre off the coast of Honduras (Soto et al. 2009). Roatán is part of the Bay Islands Marine National Park, also known as PNMIB (Parque Nacional Marino Islas de la Bahia), a marine protected area (MPA) established in June 2010 by the National Congress of the Republic of Honduras as a management tool to preserve and protect a fundamental portion of the MBRS ecosystem (Carrasco et al. 2013). Four reef sites within the nearshore waters of the southern coast of Roatán were selected for this study: Coco View Wall, Two Tall Two Small, Menagerhea, and Gold Chain (Fig. 1c). Each selected reef was shallow (3–6 m) and represented the diversity of scleractinian corals found within the region (Olson & Radawski, pers. observ.).

Fig. 1
figure 1

Maps of the a Mesoamerican barrier reef system, b Roatán island (Honduras), and c the four dive sites used in this study

Full size image

Surveys

In 2010, rebar stakes were inserted into the reef while SCUBA diving to generate 30 m permanent transects. Each 30 m transect was divided into two sub-transects of 10 m (from 5 to 15 m and 20 to 30 m), creating four to six 10 m transects per site. In late May to early June from 2010 to 2019, quantitative and qualitative data on the benthic cover and diversity of corals (both hard and soft), sponges, algae, and other benthic invertebrates were retrieved via SCUBA diving by performing point intercept transects (PIT), where the benthic cover at each 10 cm interval along the 10 m length was recorded. Data were not retrieved in 2014 and 2015. Percent coverage along these transects was calculated for hard corals (Scleractinia), soft corals (Octocorallia), macroalgae, turf, crustose coralline algae (CCA), sponges, sand, recently dead coral, rubble and other (including benthic invertebrates that do not belong to the previous categories) as the number of observations within that category divided by 100 observations per transect. On the same transects, 2 m band surveys were performed, with 1 m on each side of the tape measure. The number of hard coral, soft coral, and sponge individuals was recorded within each 10 m by 2 m transect to provide abundance data.

Where possible, all of the organisms within the hard coral, soft coral, macroalgae and sponge categories were identified to the species level (scleractinian taxonomy was mainly based on Veron (2000) with some exceptions to take into account recent reclassification of certain coral species (Wallace et al. 2012)), which allowed for the determination of percent cover of individual species. However, to minimize possible confounding factors associated with the identification of some coral species in situ, the data were grouped and analyzed at the genus level.

Data analysis

Using the vegan package (Oksanen et al. 2018) in R version 3.4.3 (R Core Team 2017), a Bray–Curtis dissimilarity matrix was generated with untransformed data and examined for homogeneity of variance (betadisper). Differences in the benthic community composition between sites and years were assessed using a two-way permutational analysis of variance (PERMANOVA; adonis) with 999 random permutations (Anderson 2006; Anderson et al. 2006). The function metaMDS was used to generate non-metric multidimensional scaling (nMDS) plots to visualize the data in two-dimensional space. A similarity percentage analysis (simper) was implemented to calculate the individual contribution of each benthic category and coral genus toward the differences noted between the first and last years of the study.

To compare the percentage coverage of the investigated categories and hard coral genera across sites and years of the study, linear mixed effects (LME) models were fitted, taking into account spatial autocorrelation. Using the R package lsmeans (Lenth 2016), pairwise comparisons among yearly least square means using Tukey’s HSD tests followed by Bonferroni corrections were employed to identify significant differences in percent cover of the various categories between years. To evaluate associations between hard corals, turf, macroalgae, and CCA, the cor.test function in R was used to run Pearson’s correlations and the R package ggpubr (Kassambara 2018) was used to visualize results.

The wilcox.test of the R package dplyr (Wickham et al. 2018) was used to compare abundances of median values of hard corals, soft corals, and sponges between the start (2010) and end (2019) of the survey period with Mann Whitney U tests. Bonferroni corrections were applied to the results. All data are shown as the mean \(\pm\) standard error (SE) unless otherwise indicated.

Results

During the 2010 to 2019 Roatán surveys, 29 scleractinian coral species, 14 soft coral species (including gorgonians), 6 macroalgal species, and 20 sponge species were identified (Table 1). The benthic community structure of the study area changed across the study period, with significant differences in overall composition between sites (2-way PERMANOVA; Df = 3, F = 24.08, p = 0.001), years (Df = 7, F = 4.21; p = 0.001), and site by year interactions (Df = 21; F = 1.42; p = 0.011). When nMDS plots of the communities at the sites in two-dimensional space were examined over the study period, a gradual homogenization of the benthic assemblages was apparent, with reduced size and spread of the ellipses between the sites in the later years of the study compared to the early years (Fig. 2). To determine which categories were responsible for these changes between 2010 and 2019, simper analyses showed that hard coral cover, followed by sand, macroalgae, and rubble, were most important (Table 2a). However, looking at percent cover of the various categories by site and year suggests that changes in both hard coral and macroalgal cover were largely driven by two of the four sites (Two Tall Two Small, Gold Chain; Fig. 3; Supplementary Table 1). Within the hard coral category, Agaricia and Orbicella were the genera most responsible for the change in benthic cover (Table 2b), with differences again occurring primarily at two of the four sites (Supplementary Table 2).

Table 1 Benthic sessile organisms identified at the sites during the study
Full size table
Fig. 2
figure 2

Non-metric multidimensional scaling plots of the benthic community composition at the four sites and demonstrating the homogenization that occurred over the study duration

Full size image
Table 2 (a) Simper analysis results for benthic categories
Full size table
Fig. 3
figure 3

Mean percent cover of the five most abundant scleractinian genera found on individual transects at the study sites over time. Trend bars are shown in black and different color circles represent the 4 study sites. No data were collected in 2014 or 2015. * denote significance of the trend lines at p < 0.05

Full size image

LME models were significant by year for the categories hard coral, macroalgae, turf, CCA, sand, rubble, and recently dead coral and by site for macroalgae, soft corals, sponges, and sand (Table 3). All categories except for soft corals, sponges, and sand showed a year by site interaction (Table 3). Although scleractinian coral cover declined significantly across the study period, the hard coral category, with one exception in 2012, remained the most highly represented on the transects (Supplementary Table 1). Conversely, the percent cover of macroalgae and turf significantly increased over time (Fig. 4; Table 3). When LME models for the five most abundant scleractinian coral genera (Acropora, Agaricia, Orbicella, Porites, and Siderastrea) were examined, significant decreases in benthic cover between 2010 and 2019 by Agaricia spp. (Df = 1, F = 3.04, p = 0.0056) were partially offset by significant increases in Siderastrea spp. (Df = 1, F = 7.42, p < 0.0001; Fig. 3; Table 4). However, site was significant for Orbicella, Porites, and Siderastrea and all but Porites showed a year by site interaction (Table 4).

Table 3 Benthic cover categories examined by LME models followed by ANOVAs to evaluate year, site, and year by site interactions
Full size table
Fig. 4
figure 4

Mean percent cover for hard corals, macroalgae and turf across the duration of the study. Error bars show standard error

Full size image
Table 4 The five most dominant hard coral genera examined by LME models followed by ANOVAs to evaluate cover in 2010 and 2019, by site, and year by site interactions
Full size table

After applying Bonferroni corrections, pairwise comparisons among years using lsmeans followed by Tukey’s post-hoc tests showed significant differences in hard coral cover between 2011 and 2018 (p = 0.0002), macroalgal cover between 2013 and 2016 (p = 0.0016), sand between 2011 and 2013 (p = 0.0015), and multiple years for turf (including 2011–2019: p < 0.0001), rubble (e.g., 2012–2016: p = 0.0009), and dead coral (e.g., 2010–2017: p = 0.0014 and 2017–2019: p < 0.0001; Supplementary Table 3). When examined using the five most abundant scleractinian genera, significant differences in cover between multiple years were only seen for Siderastrea spp. (e.g., 2010 to 2019: p = 0.0003; Supplementary Table 3). While no significant correlation was found between the increase in macroalgal cover and the decrease in hard coral across the study period (R = − 0.03, p = 0.67), there was a significant negative correlation for the increase in turf cover and decrease in hard coral cover (R = − 0.19, p = 0.0093).

Although most corals were visually identified to the species level for abundance data, visual differentiation of species of Agaricia can be difficult (Bongaerts et al. 2013) and no collections were permitted within the marine park to allow for further microscopic or molecular analyses. As a result, all colonies of Agaricia were simply recorded as Agaricia spp. Using Mann–Whitney U tests on abundance data, no significant differences were observed between 2010 and 2019 for any of the categories. Trends suggesting a gradual reduction in the number of individuals within the five most abundant scleractinian genera over time were noted, as was a slight increase in the abundance of soft corals over time (Table 5; Supplementary Fig. 1).

Table 5 Median abundances of the five most abundant genera of scleractinian corals and the categories soft corals and sponges from 2010 to 2019 analyzed using Mann–Whitney U tests
Full size table

Discussion

Results from this study demonstrated that the composition of the benthic community on shallow, near-shore reefs around Roatán significantly changed from 2010 to 2019. A reduction in scleractinian coral cover and a concomitant increase in macroalgal and turf cover were noted over the study period. This phenomenon, known as a "coral-algal phase shift", has been previously documented on reefs in Belize, Curaçao and Bonaire (Hughes 1994; Nugues and Bak 2008; Jackson et al. 2014; Smith et al. 2016a, b; de Bakker et al. 2017) but has not been previously reported from the Bay Islands. The increased abundance of macroalgae and turf on benthic substrata inhibits growth, fecundity, and survivorship of the scleractinian corals on the reefs (Lewis 1986; Tanner 1995; Hughes et al. 2007), suppresses colonization by coral larval recruits (Kuffner et al. 2006), and increases the prevalence of coral diseases (Nugues et al. 2004). All of these factors work jointly to further decrease the presence of hard coral (Burkepile and Hay 2010).

Although the increased prevalence of both macroalgae and turf algae likely impacted the growth and survival of hard corals, a direct correlation linking the decline in hard coral cover with the increase in macroalgal cover was not found. However, a significant negative correlation was observed between the decrease in hard coral cover and increase in turf cover. This suggests that one of the causes for the shift in benthic composition observed on Roatán’s reefs could be attributed to the ability of algae to outcompete corals for space, but that among the two algal categories, the increased presence of turf algae appears to be more detrimental to coral growth and/or survival. Other benthic categories such as soft corals and sponges remained stable over the study period while significant changes in percent cover of CCA, rubble, sand and dead coral were observed.

Continued development of coastal areas to support the local population as well as to enhance tourism may be one of the leading factors in the benthic transition (Burke and Maidens 2004; Bozec et al. 2008; Burke et al. 2011; Stubler et al. 2015; Pendleton et al. 2016). On the island of Roatán, the amount of tourism has been increasing by approximately an order of magnitude per decade (Doiron and Weissenberger 2014), putting greater stress on the fragile coastal ecosystems such as mangroves and coral reefs that bring tourists to the island (Dorion & Weissenberger 2014; Canty et al. 2018). Correspondingly, the amount of land classified as urban increased over time, from 6.3% in 1985 to 24.6% in 2015 (Helm 2014). The continued development of previously forested land for human use results in greater runoff of sediments (e.g., erosion), nutrients, and pollutants from land-based sources (Carrasco et al. 2013) that can limit the filtration capacity of sponges and other filter-feeding invertebrates (Rogers 1990; Fabricius and Wolanski 2000; Prouty et al. 2008; Bannister et al. 2012; Bell et al. 2015), increase the susceptibility of hard corals and other reef organisms to diseases (Koop et al. 2001; Fabricius 2005; Kaczmarsky and Richardson 2011; D’Angelo and Wiedenmann 2014; Vega Thurber et al. 2014), and create eutrophic conditions that promote the proliferation of algae and algal blooms (Hughes et al. 1999; Anderson et al. 2002; Lapointe 2019). Roatán is also likely impacted by riverine discharges from the mainland Honduran coast, which can easily reach the island via the cyclonic gyre off the Gulf of Honduras and may impact the composition of benthic communities (Harborne et al. 2001; Prouty et al. 2008). To demonstrate the impacts of these activities on near-shore reefs, long term measurements of seawater nutrient concentrations paired with sediment traps are needed.

The changes in benthic cover are likely responsible for the homogenization of species assemblages that occurred from 2010 to 2019. Reef health is influenced by species biodiversity, which confers ecosystem resilience during ecological reorganization (Nyström et al. 2000; Nyström 2006; Camargo et al. 2009). In the absence of any major disturbance (e.g., hurricane, massive disease outbreak) that would rapidly impact the composition the benthic community, significant changes in abundance of benthic categories were not expected over the 10 year study period and were not observed. Instead, the gradual alteration in the presence and abundance of benthic constituents towards a more uniform community may be important for determining optimal management strategies in the future. At present, although biotic homogenization is becoming more common on Caribbean and other reefs (Burman et al. 2012; Mouillot et al. 2014; Graham et al. 2015; Richardson et al. 2018; Estrada-Saldívar et al. 2019), its consequences are not yet fully understood and additional studies are needed to elucidate the long-term repercussions of homogenization on the functional stability of the entire coral reef ecosystem.

Benthic rugosity measurements were not performed in this study, limiting what can be inferred for the architecture of nearshore Roatán reefs. However, the observed decline in branching and structurally complex scleractinian species (e.g., Acropora and branching Porites spp.) across the study period appeared to be partially offset by the increased presence and/or growth of sub-massive coral species (e.g., Siderastrea spp., Porites astreoides). On the reefs of nearby Cayos Cochinos (Honduras), sub-massive scleractinian corals dominate and support high abundance, biomass and richness of fish assemblages compared to locations characterized by fewer sub-massive and more leafy coral species (Cáceres et al. 2020). Similar scenarios have been observed on Caribbean reefs since the 1980’s, where rising seawater temperatures coupled with disease outbreaks led to shifts in the dominance of hard corals from fast growing, branching and structurally complex species to more stress tolerant but less architecturally intricate species (Carpenter 1988; Alvarez-Filip et al. 2009; Alvarez-Filip et al. 2011a, b). Recent studies have also found that some reef-building corals were already declining due to human activities before these additional stressors were documented (Cramer et al. 2020; Cybulski et al. 2020). Coral reefs are one of the most biodiverse ecosystems on the planet but many of the services provided to humans and other species are linked not only to the amount of live coral, but also to the three-dimensional topographic complexity of these habitats (Wilson et al. 2007; Graham 2014; Hoegh-Guldberg et al. 2019). Several studies demonstrated that coral habitats with high structural complexity positively affected reef communities (Messmer et al. 2011; Holbrook et al. 2015), as this element impacts the availability of resources and environmental niches necessary to support diverse and abundant invertebrate and fish communities (Holling 1992; Wilson et al. 2010; Darling et al. 2017). Unfortunately, coral taxa with the highest three-dimensional complexity appear to also be the species that are most susceptible to the wide range of natural and anthropogenic stressors impacting coral reefs (Cramer et al. 2020; Cybulski et al. 2020). Future studies should incorporate benthic rugosity measurements in order to be able to more directly discuss changes in reef architecture.

Species within the genera Acropora and Orbicella represent the primary contributors to building the Caribbean reef framework (Jordan-Dahlgren and Rodriguez-Martinez 2003; Perry et al. 2013; Kuffner and Toth 2016) but the presence of these taxa has dramatically declined over the last three decades. Several recent studies of coral larval recruitment in the Caribbean found that the majority of coral recruits belonged to brooding genera (e.g., Agaricia, Porites) while recruits from broadcast spawning genera (e.g., Acropora, Orbicella) were rarely observed (Arnold et al. 2010; Urvoix et al. 2012; Brandt et al. 2019). This suggests that future reef communities may be dominated by weedier, brooding hard coral species and result in a further reduction of reef framework in the Caribbean. This transition to dominance of non-framework-building corals has been observed for many reefs within the Mesoamerican Reef System (González-Barrios and Álvarez-Filip 2018) but more data from long-term studies are needed to more fully comprehend the implications of these alterations.

The decline in framework-building corals has been largely attributed to severe bleaching events and disease outbreaks (Harvell et al. 2007; Randall and Van Woesik 2017; Van Woesik and Randall 2017), including the current stony coral tissue loss disease outbreak (Alvarez-Filip et al. 2019; Meyer et al. 2019; Muller et al. 2020; Rosales et al. 2020) that has now reached the reefs surrounding Roatán (Precht 2021). Although coral bleaching has been recorded in the Caribbean since 1911 (Rowlands et al. 2008) and severe episodes were documented in years characterized by anomalously high water temperatures (Eakin et al. 2010), the average monthly seawater temperatures remained stable in Roatán during the study period (Supplementary Table 4). This suggests that the changes observed in benthic cover were not directly attributed to the consequences of bleaching.

Information on the incidence and extent of coral diseases in the region are incomplete (Kramer et al. 2000), with relatively few of the over 40 described syndromes (Bruckner 2016) reported from the reefs around Roatán including white band disease, yellow band disease, black band disease (BBD), and white plague type II (Kramer et al. 2000; Riegl et al. 2009; Kroll et al. 2018). Although BBD was noted infrequently affecting various genera of hard corals during this study, we observed a different infection that shared a similar gross appearance. This report, to our knowledge, represents the first documentation of Caribbean ciliate infection in Roatán (Fig. 5) although its presence was observed at other Caribbean locations beginning in 2004 (Cróquer et al. 2006). Visual inspection of the dark band typical of this syndrome on the affected A. palmata coral revealed ciliates with prominent peristomial wings, presumably belonging to the genus Halofolliculina, clustered between the visibly healthy coral tissue and the bare skeleton. Continued increases in sea surface temperatures combined with observations of new diseases on these reefs suggests that further declines in coral cover may be anticipated.

Fig. 5
figure 5

a, b Acropora palmata affected by Caribbean ciliate infection in Roatán showing the typical dark band and spotted appearance of this disease. c Higher magnification of halofolliculinid ciliates on the coral skeleton

Full size image

The sites used in this study are located in a marine protected area (MPA) that was established in 2010 to promote sustainable fishing practices, protect migratory species, and control invasive species (Carrasco et al. 2013). Unfortunately, as evidenced from this study, the establishment of an MPA does not guarantee success in conserving reef ecosystems or achieving the stated objectives. It is likely that anthropogenic impacts on shallow reef environments take many years to become apparent, suggesting that early protection and long-term mitigation efforts are needed for a higher probability of successful outcomes. Many of the anthropogenic factors that have been shown to negatively affect coral reefs (e.g., eutrophication, sedimentation, ocean warming, sea level rise, intensity and frequency of storms; e.g., Mumby et al. 2007; Smith et al. 2008; Pandolfi 2010; Perry et al. 2013; Jackson et al. 2014; Muñiz-Castillo et al. 2019; Cáceres et al. 2020; França et al. 2020) cannot be easily mitigated and are outside the reach of MPA management strategies. Until MPAs are staffed and funded at levels that permit effective enforcement and monitoring and governmental agencies work collectively to address the larger issues that threaten reef health, nearshore reefs are likely to face continued gradual declines and phase shifts.