Sargassum blooms in the Caribbean alter the trophic structure of the sea urchin Diadema antillarum

The arrival of large masses of drifting Sargassum since 2011 has caused changes in the natural dynamics of Caribbean coastal ecosystems. In the summer of 2015, unprecedented and massive mats of S. fluitans and S. natans have been observed throughout the Mexican Caribbean including exceptional accumulations ashore. This study uses stable isotopes to assess the impact of Sargassum blooms on the trophic dynamics of the Diadema antillarum sea urchin, a keystone herbivore on many Caribbean reefs. Bayesian models were used to estimate the variations in the relative proportions of carbon and nitrogen of assimilated algal resources. At three lagoon reef sites, the niche breadth of D. antillarum was analysed and compared under massive influx of drifting Sargassum spp. vs. no influx of Sargassum blooms. The effects of the leachates generated by the decomposition of Sargassum led to hypoxic conditions on these reefs and reduced the taxonomic diversity of macroalgal food sources available to D. antillarum. Our trophic data support the hypothesis that processes of assimilation of carbon and nitrogen were modified under Sargassum effect. Isotopic signatures of macroalgae associated with the reef sites exhibited significantly lower values of δ15N altering the natural herbivory of D. antillarum. The Stable Isotopes Analysis in R (SIAR) indicated that, under the influence of Sargassum blooms, certain algal resources (Dictyota, Halimeda and Udotea) were more assimilated due to a reduction in available algal resources. Despite being an abundant available resource, pelagic Sargassum was a negligible contributor to sea urchin diet. The Stable Isotope Bayesian Ellipses in R (SIBER) analysis displayed differences between sites, and suggests a reduction in trophic niche breadth, particularly in a protected reef lagoon. Our findings reveal that Sargassum blooms caused changes in trophic characteristics of D. antillarum with a negative impact by hypoxic conditions. These dynamics, coupled with the increase in organic matter in an oligotrophic system could lead to reduce coral reef ecosystem function.


INTRODUCTION
The arrival of massive amounts of pelagic Sargassum spp. has caused changes in the natural benthic dynamics of Caribbean coastal ecosystems for the last nine years (Gower, Young & King, 2013;Schell, Goodwin & Siuda, 2015). Pelagic Sargassum is a complex of two species, namely S. fluitans and S. natans (Oyesiku & Egunyomi, 2014). Since 2011, extensive masses of Sargassum appeared in unusual ways in oceanic waters off northern Brazil (De Széchy et al., 2012;Sissini et al., 2017), along the West Indies and Caribbean coasts (Gower, Young & King, 2013) from Trinidad to the Dominican Republic (Rodríguez-Martínez, van Tussenbroek & Jordán-Dahlgren, 2016;van Tussenbroek et al., 2017), and along the west African coast from Sierra Leone to Ghana (Smetacek & Zingone, 2013). Wang et al. (2019) recorded that for June 2018, wet biomass reached more than 20 million tons in the Caribbean Sea and Central Atlantic Ocean.
The Mexican Caribbean shores faced atypical massive mats of pelagic Sargassum in the summer of 2015 (van Tussenbroek et al., 2017;Cuevas, Uribe-Martínez & Liceaga-Correa, 2018;Arellano-Verdejo, Lazcano-Hernandez & Cabanillas-Terán, 2019). There was a subsequent decrease during 2016 and 2017, but for most of 2018 and thus far in 2019 influx has increased again. Several studies revealed that these massive mats of Sargassum have a new possible distribution source different from the historic North Atlantic Recirculation Region (NARR) known as ''The Sargasso Sea'' (Schell, Goodwin & Siuda, 2015). Instead, the most probable origin of the massive influx on the Caribbean shores is the North Equatorial Recirculation Region (NERR) (Johnson et al., 2013;Schell, Goodwin & Siuda, 2015). High oceanic temperatures and nutrient inputs (Franks, Johnson & Ko, 2016;Wang et al., 2018), among other oceanographic coupled patterns such as changes of surface currents, are the most probable causes of this new region of Sargassum flourishment (Johnson et al., 2013;Gower, Young & King, 2013;Sissini et al., 2017). A recent study by Wang et al. (2019) revealed that increases of pelagic Sargassum are driven by upwelling off West Africa during the boreal winter and by Amazon River discharge during the spring and summer. The authors state that recurrent blooms in the Caribbean Sea and tropical Atlantic are likely, highlighting the importance for understanding their effects on existing ecosystems for future planning.
Changes in habitat structure can directly influence trophic dynamics (Hunter & Price, 1992;Sweatman, Layman & Fourqurean, 2017) and have been shown to cause synergistic effects on coral reefs (Smetacek & Zingone, 2013). For example, harmful macroalgae blooms have been recognized as drivers of degradation in coral reef habitats (Lapointe et al., 2005). This has effects on the diversity of reef biota (Bauman et al., 2010;Louime, Fortune & Gervais, 2017) like variations in the sea urchin populations (Lapointe et al., 2010). The carbon and nitrogen flow by macroalgae blooms likely has adverse effects at different scales. Such disturbances from Sargassum, coupled with pre-existing threats on coral reefs, add to the drivers of Anthropocene reef degradation (Alvarez-Filip et al., 2011;Cramer et al., 2012).
The massive decomposition of Sargassum has negative impacts not only on tourism and local fisheries, but on nearshore ecosystems (Solarin et al., 2014;Louime, Fortune & Gervais, 2017). However, few studies assess the trophic impact of Sargassum blooms on benthic communities. Pelagic Sargassum and their attached epiphytic algae can contribute new organic matter to these communities (Rooker, Turner & Holt, 2006;Wang et al., 2018). Therefore, we consider whether or not these new sources of nitrogen and carbon act in a detrimental manner on the trophic chain of benthic communities. The beaching and decomposing of massive Sargassum mats produce hypoxia in near-shore coral reef communities (Rodríguez-Martínez et al., 2019). This effect coupled with high hydrogen sulfide and ammonium concentrations have been shown to cause faunal mortality in the Mexican Caribbean (Rodríguez-Martínez et al., 2019). As a consequence, the coastal environment becomes even more sensitive to degradation agents. To assess these issues, we included measurements of dissolved oxygen in our study.
Evaluating consumers and resources through a trophic approach by tracking the relationships between consumers and prey provides relevant information on the trophic structure and dynamics of a benthic community (Minagawa & Wada, 1984;Vanderklift, Kendrick & Smit, 2006;Behmer & Joern, 2008). Stable isotopes of carbon (δ 13 C) and nitrogen (δ 15 N) have been used in marine ecosystems to determine the feeding habits of species (Peterson & Fry, 1987), nutrient migrations within food webs, trophic position of organisms and their contribution at all trophic levels (Vander Zanden & Rasmussen, 1996). It is also possible to trace the origin and transformation of the ingested organic matter and to detect changes in the trophic positions of organisms that coexist in the same habitat (Hobson, 1999;Vanderklift, Kendrick & Smit, 2006;Rodríguez-Barreras et al., 2016).
Stable carbon and nitrogen isotope ratios provide time-integrated information regarding feeding relationships and energy flow through food webs (DeNiro & Epstein, 1981;Peterson & Fry, 1987;Vander Zanden & Rasmussen, 2001). Moreover, stable isotopes can be used to study the trophic niche breadth of a species (Bearhop et al., 2004;Parnell et al., 2010;Phillips et al., 2014). This is directly influenced by consumers and resource input, providing a quantitative assessment of trophic conditions (Newsome et al., 2007;Boecklen et al., 2011). Stable isotope analyses are useful for assessing the health of ecosystems because it is possible to associate the consumers trophodynamics and niche breadth with habitat disturbances (Layman et al., 2007b;Hamaoka et al., 2010). It is also possible to detect changes in the trophic spectrum from anthropogenic impacts or unusual conditions that cause shifts in ecosystems (Wing et al., 2008;Prado, Alcoverro & Romero, 2010;Tomas, Box & Terrados, 2011). In light of the massive arrival of pelagic macroalgae, sea urchin herbivory is a good model to understand variability in the benthic trophic chain, as sea urchins are considered generalist consumers with a plastic feeding habit (Lawrence, 1975;Vanderklift, Kendrick & Smit, 2006). Echinoids have the capability to modify the community structure through foraging behaviour (Carpenter, 1986;Hay & Fenical, 1988;Sala et al., 1998;Eklöf et al., 2008). Thus, the relative position of δ 13 C vs. δ 15 N echinoids displayed in a bi-plot can give insights about organism responses to niche shifts, diet variability and habitat modification (Layman et al., 2007a;Layman et al., 2007b;Layman et al., 2012;Sweatman, Layman & Fourqurean, 2017).
The effect of Sargassum and their leachates on the diet of D. antillarum can improve our understanding on the impact on trophic ecology of one of the most important sea urchins of the Mexican Caribbean. The main reason to focus this study on D. antillarum is that this species is and was the major shallow-hard-bottom grazer in our study sites (Jorgensen, Espinoza-Ávalos & Bahena-Basave, 2008;Jordán-Garza et al., 2008). One of the most dramatic events in the Caribbean resulted from the pathogen-driven reduction in the populations of D. antillarum (Lessios et al., 1984) with detrimental ecological consequences like coral-algal phase-shifts. The southern part of Quintana Roo is not an exception encompassing with the effects of the abrupt coastal development and watershed pollution as key drivers along the Costa Maya (Arias-González et al., 2017).
The overarching aim of this study was to determine variations in the relative proportions of carbon and nitrogen of assimilated algal resources and the niche breadth of D. antillarum under massive influx of drifting Sargassum spp. vs. no influx of Sargassum at back reefs. We also aimed to determine whether pelagic Sargassum was a substantial source of energy for D. antillarum. To do this, we compared δ 15 N and δ 13 C values of D. antillarum with and without influx of Sargassum to track changes in this species trophic ecology (diet, trophic position and niche breadth). Ultimately, we tested the hypothesis that an influx in Sargassum in coastal ecosystem creates a significant change in the available algal sources and a shift in the trophic structure.

Study sites
We determined the stable isotopes of carbon and nitrogen for D. antillarum at three reef lagoons (Mahahual, Xahuayxol, and Xcalak) with different distances from the beach to the reef crest (Fig. 1). The main strategy implemented by local authorities at some beaches with the massive arrival of macroalgae included the removal and disposal of Sargassum in the highest part of the beach or in places determined ex profeso. This contributed to a continuous accumulation of Sargassum masses on the beach. However, the Sargassum removal was not quantified and the information regarding removal included here is only preliminary.
Mahahual (18 • 42 16.96 N 87 • 42.619 W) is located in the northern part of the Mesoamerican Barrier Reef System (MBRS) in the state of Quintana Roo. Mahahual is a former fishing village but during the last two decades has undergone reef degradation due to anthropogenic impact (Martínez-Rendis et al., 2016). It has a narrow reef lagoon (230-450 m). Sargassum management in this locality was active through removing it from the beach and ex situ disposition.
Xahuayxol (18 • 30 21.78 N; 87 • 45 24.84 W) located south of Mahahual, has a larger reef lagoon measuring 300 to 500 m from the beach to the reef crest. Sargassum was not removed from the beach in any systematic way and remained accumulated on the shore. This reef is the northern limit of the marine protected area Parque Nacional Arrecifes de Xcalak (PNAX) and human activities are less salient than in Mahahual (Schmitter-Soto et al., 2018).
Xcalak (18 • 14 7.68 N; 87 • 50 1.46 W), at the southern limit of the Mexican Caribbean, is part of PNAX since 2000. It is also part of the MBRS (Hoffman, 2009). It has a wide reef lagoon (950-1,200 m), and Sargassum was accumulated along the shore in large amounts. There was active but less intense Sargassum management in place at Xcalak, where final disposal was in situ on the highest part of beach. At all sampled sites, the dominant forcing mechanism was reef lagoon circulation from wave action (Mariño-Tapia et al., 2010). In our study area, during the period from June to August has the wave orbital velocity over the threshold of motion (Maldonado-Sánchez et al., 2019), indicating active circulation in the reef lagoons.

Collecting and processing data
This study covers two periods: Under Sargassum effect (USE) during the months of July-August 2015 and without Sargassum effect (WSE) in July-August 2016. USE sampling for stable isotope analysis included drifting Sargassum (mixture of S. fluitans and S. natans), turf associated pelagic Sargassum, benthic macroalgae, local turf and 19 individuals of D. antillarum. WSE sampling included benthic macroalgae, local turf and 15 individuals of D. antillarum (see sampling details ST1). Samples sizes were based on previous studies to obtain sufficient data for statistical analysis (Rodríguez, 2003;Tomas et al., 2006;Wing et al., 2008;Rodríguez-Barreras et al., 2016). The sampling sites were at coastal lagoons in the back reef zone (section c, Fig. 2), zone with no visible presence of Sargassum leachates (van Tussenbroek et al., 2017) and where D. antillarum is distributed (Steneck & Lang, 2003;Jorgensen, Espinoza-Ávalos & Bahena-Basave, 2008;Jordán-Garza et al., 2008;Maldonado-Sánchez, 2018). Under Sargassum effect (USE) measurements USE included measurements of dissolved oxygen (mg l −1 ) recorded with a calibrated Multi-parameter water quality checker HORIBA 50 at Mahahual, Xahuayxol and Xcalak. Measurements of dissolved oxygen were made at points distributed in three sections from areas with decomposing Sargassum (section a), leachates (section b -dark brown water-) and reef lagoon areas without Sargassum leachates (section c) (Fig. 2). Pelagic Sargassum spp., turf (benthic turf and the associated turf to pelagic Sargassum) and macroalgae samples were collected in coral reef patches of section c (back reef zone) for each sampling site.

Under and without Sargassum effect (USE and WSE) measurements
We collected algal samples to obtain biomass, and for stable isotope analysis using nine quadrats (50 × 50 cm) per site. Pelagic Sargassum biomass was calculated based on sunken thalli and overlaid on reef substrates inside the quadrats. The quadrats were located randomly within the sea urchin habitat (radius of 15 m from collected echinoids). The substrate inside each quadrat was scrapped, carefully removed, collected in bags, and frozen for later analysis.
Macroalgae were identified according to Littler & Littler (2000). Analyses were performed to genus level. For biomass estimates samples were dried for 48 h in an oven at 60 • C. Samples were weighed with a digital balance (standard error = 0.0001 g). To determine D. antillarum differential algae assimilation considering USE and WSE, algae samples were pooled per site. The sampled echinoids and algal species for this study are not threatened. The collection permit was obtained from the Comisión Nacional de Acuacultura y Pesca (CONAPESCA, PPF/DGOPA-002/17).
The collected individuals of D. antillarum were at the same depth range (1.5-2.5 m) and only individuals greater than 5.0 cm in test diameter were collected to avoid any ontogenic effect. Samples were frozen shortly after collection and processed later at the laboratory. The muscles of Aristotle's lanterns were carefully removed and washed from the stomach contents to estimate algal assimilation by D. antillarum because this tissue offers a time-integrated measure of carbon and nitrogen assimilated sources (Polunin et al., 2001;Ben-David & Schell, 2001;Phillips & Koch, 2002).
Macroalgae and local turf, pelagic Sargassum species (S. fluitans and S. natans), turf associated to pelagic Sargassum, and echinoids muscle samples were rinsed with filtered water, dried at 50 • C during 48 h, grounded to a fine powder and placed in glass vial for isotope analyses. To remove carbonates from some algal species (eg., Halimeda spp. Penicillus spp., etc.), the samples were washed with diluted HCl at 1 N prior to drying to avoid disturbance in the mass spectrometer reading.
A subsample of each algae and muscle (1mg) was taken to evaluate the 13 C/ 12 C and 15 N/ 14 N ratios using a Delta V Plus Mass Spectrometer. Catalyzers silvered cobaltous/cobaltic oxide and chromium oxide were used. Carbon and nitrogen samples were analysed in a dual isotope mode at the Centro Interdisciplinario de Ciencias Marinas from Instituto Politécnico Nacional. Isotope samples were loaded into tin-capsules and placed in a 50-position automated Zero Blank sample carousel on a COSTECH 4020 elemental analyzer. The carbon and nitrogen isotopic results were expressed in standard delta notation relative to Vienna Pee Dee Belemnite (VPDB) and to atmospheric air.
The standard deviations of δ 13 C and δ 15 N replicate analyses were estimated; the precision values were 0.2 for carbon and nitrogen isotope measurements. In addition, we calculated the trophic level (TL) according to Hobson & Welch (1992) for every individual of D. antillarum in each site, expressed as: Where Nm is the mean δ 15 N ratio of each sea urchin, Nb is average basis δ 15 N value of the algal community, and TEF is the given value for the trophic enrichment factor (TEF). We assumed a TEF of 2.4 following Moore & Semmens (2008).

Data analysis
Dissolved oxygen data were summarized to obtain average values (± standard error) by section (sections a, b, c in Fig. 2) and reef lagoons (Mahahual, Xahuayxol, and Xcalak). We evaluated differences among sections and at the reef lagoons (sections a, b, c, in Fig. 2). We plotted raw data of dissolved oxygen as a function of distance to coast to visualize the low to high values gradient related to that distances in every reef lagoon.
The relative contribution of algae to the diet of the sea urchins D. antillarum was estimated with a Bayesian isotopic mixing model ( SIAR Parnell & Jackson, 2013), which included the isotopic signatures, fractionation and variability to estimate the probability distribution of the contribution of the food source to a mixture. This procedure supplied accurate information about the contribution of algal species to the sea urchin tissues, as it provided the proportion for every source and recognized the main sources as important components of the diet (Peterson, 1999;Fry, 2006;Wing et al., 2008) at three different sites, and under and without Sargassum effect. To run the model, the isotopic discrimination factor values used were 2.4 ± 1.6 (mean ± SD) for δ 15 N, and 0.4 ± 1.3 (mean ± SD) for δ 13 C (Minagawa & Wada, 1984;Fry & Sherr, 1989;Moore & Semmens, 2008;Cabanillas-Terán et al., 2016).
The following algal taxa/groups were considered for the mixing models analyses: Caulerpa, Codium, Dictyota, Halimeda, Laurencia, Lobophora, Padina, Penicillus, Sargassum polyceratum, Stypopodium, turf, and Udotea. The sources for the model were selected following the theoretical geometric assumptions of the mixing model according to Phillips et al. (2014) and Rodríguez- Barreras et al. (2015) to ensure reliable resources. Samples of D. antillarum did not require lipid extraction since C:N ratios of Aristotle lantern's muscle were lower than 3.5 (Post et al., 2007).
We performed a comparison USE and WSE between the niche width and overlap for D. antillarum by using Stable Isotope Bayesian Ellipses in R (SIBER) (Jackson et al., 2011) from the SIAR package (Parnell & Jackson, 2013). This procedure performs metrics based on ellipses and provides the standard ellipse corrected area (SEAc) used as the trophic niche breadth and the overlap between ellipses, presuming that values close to 1 exhibit a higher trophic overlap. Models were run with 200,000 iterations and a burn in of 50,000.
Homogeneity and normality of variance were tested by performing a Kolmogorov-Smirnov and a Cochran's test (Zar, 1999). Nitrogen data followed the premises of parametric analysis, but the carbon, dissolved oxygen and biomass data required a power transformation for reaching normality and homogeneity of variance (Box & Cox, 1964). We ran two-way ANOVA to evaluate dissolved oxygen data differences among sections in the reef lagoons and we performed a post hoc comparison using Tukey-HSD test. The functions aov and glm from the Gaussian family were used to test the differences in isotopic ratios of carbon and nitrogen values to compare the effect (WSE and USE) between sites and their interaction. Statistics were performed with α < 0.05 (R Core Team, 1.0.153, 2017).

RESULTS
The dissolved oxygen values USE indicated that the effects of the leachates generated by the decomposition process, together with the organic material carried in their vegetal structures, reduced the values of dissolved oxygen in the reef lagoon water. The decomposing Sargassum area (section a, Fig. 2) showed an average range from 1.01 (S.E. ± 0.30) mg l −1 at Xcalak to 1.88 (S.E. ± 0.37) mg l −1 at Mahahual. The leachates area (section b, Fig. 2) showed an average range from 2.42 (S.E. ± 0.32) mg l −1 at Xahuayxol to 3.66 (S.E. ± 0.42) mg l −1 at Mahahual. The back reef area (section c, Fig. 2) showed an average range from 4.1 (S.E. ± 0.34) mg l −1 at Mahahual to 4.8 (S.E. ± 0.22) mg l −1 at Xcalak. The two-way ANOVA indicated significant differences between reef lagoons (p < 0.05) and sections (p < 0.01); Mahahual was significantly different to Xcalak, but Mahahual and Xcalak were not significantly different to Xahuayxol (Post-hoc HSD of Tukey test, 95% confidence). The three sections at the three reefs were significantly different, except the sections b and c of Mahahual (Post-hoc HSD of Tukey test, 95% confidence). Therefore dissolved oxygen data showed a gradient significantly different between sections. The overall values of dissolved oxygen displayed the lowest concentrations for section a, near the shoreline and higher values beyond the back reef section c (Fig. 3).

Biomass, δ 15 N and δ 13 C of macroalgae
Biomass data for benthic taxa displayed no significant differences between USE and WSE, but significant differences were found among localities (ANOVA, df = 2, F = 8.24, p < 0.0001). Mahahual had the highest mean benthic biomass values (55.2 dry weight m −2 ) followed by Xahuayxol with (38.8 dry weight m −2 ) and Xcalak (16 dry  Under and without Sargassum effect values revealed significant differences in overall benthic algae values of δ 15 N (ANOVA, df = 1, F = 20.27, p < 0.0001). Specifically under Sargassum blooms most of the algae exhibited isotopic signatures with significantly depleted δ 15 N like Dictyota and turf across the lagoon reef sites ( Table 2). The overall macroalgal δ 15 N under Sargassum fluctuated from 0.023 to 2.08 . At Xcalak Caulerpa displayed the highest mean values of nitrogen with 2.02 ± 0.08 . Local Turf USE displayed negative values and overall turf values fluctuated from −0.97 to 0.42 . Xahuayxol displayed the most negative δ 15 N mean value of local turf (−0.51 ± 0.02 ). Without Sargassum effect the mean algal genus δ 15 N fluctuated from 0.06 ± 0.08 with Penicillus at Xcalak, and Xahuayxol displayed the highest mean value of δ 15 N with Caulerpa (5.68 ± 0.01 ) ( Table 2).

Sea urchins
There were significant differences δ 15 N among sites (ANOVA df = 2, F = 6.473, p = 0.005) and the interaction between site*effect (USE and WSE) showed significant differences (ANOVA, df = 2, F = 7.321, p = 0.003). D. antillarum exhibited no differences among sites for δ 13 C values p > 0.05. However, we found significant differences analysing the USE and WSE effect (ANOVA df = 1, F = 5.301, p =0.03). The isotopic ratios of D. antillarum (USE) varied from 3.83 to 6.13 for δ 15 N, while δ 13 C ranged from −9.41 to −13.62 . Mahahual was the site with the highest average values for δ 15 N 5.80 ± 0.30 , while Xcalak displayed the lowest average value 4.38 ± 0.29 . The isotopic ratios of D. antillarum (WSE) ranged from 4.69 to 6.16 for δ 15 N, while δ 13 C fluctuated from −8.83 to −13.42 . We found significant differences for δ 15 N for sea urchins between sites (USE, ANOVA, df = 2, F = 6.47, p < 0.005).-Xcalak showed particularly low values under Sargassum effect (average value 4.38 ± 0.29 versus WSE average value 5.44 ± 0.36 ). Nevertheless, δ 13 C exhibited no significant differences although we noticed a negative trend in the values of δ 13 C under Sargassum effect (USE).

Algal source contributions (SIAR)
Mixing models provided evidence for the contribution of different algal resources for three sites USE and WSE (Table 3). SIAR analysis showed that D. antillarum behaved as an opportunistic grazer under the Sargassum effect, it is important to note that pelagic Sargassum, despite being one of the most abundant available resources, was not the most assimilated resource (Fig. 4). Relatedely, there was a reduction in benthic food sources USE (Fig. 4). Without Sargassum effect D. antillarum consumed, Laurencia, Stypopodium and Udotea (12-15% in average) at Mahahual; Caulerpa, Laurencia, Penicillus, Sargassum and Stypopodium (8-14% in average) at Xahuayxol; and Codium, Lobophora and Padina (13-15% in average) at Xcalak. Nevertheless, those resources were absent in the diet of D. antillarum under Sargassum effect (Table 3). Hence, the species displayed differential resource assimilation and Caulerpa was the most important resource for D. antillarum in Mahahual WSE (up to 37%), followed by Turf (up to 34%) and Halimeda and Udotea (up to 29% for both). USE the most important resource was Halimeda (up to 44%) followed by Caulerpa and Dictyota (both up to 31% of contribution). S. fluitans and S. natan s were no important sources (0-28% and 0-23% respectively), and turf associated to Sargassum blooms was the lesser assimilated resources by D. antillarum from 0 up to 22% (Table 3).
At Xahuayxol WSE D. antillarum showed Caulerpa was the most important resource for D. antillarum (from 2 up to 25%) and for the rest of algae there were very similar algal contribution (from 0 up to 23%). The main macroalgal contributor of USE was Udotea with up to 61%, followed by Halimeda and Lobophora (with up to 35% and 38% respectively) as secondary resources. Sargassum's associated turf showed evidence of low   (Table 3). Turf was the main algal resources for D. antillarum in Xcalak WSE (up to 45%) followed by Caulerpa, Codium and Padina as secondary resources (close to 30% maximum of contribution); contrasting USE the main macroalgal contributors in Xcalak were Penicillus and Caulerpa with up to 39% and 40% respectively. Likewise Dictyota and Sargassum polyceratium ( benthic Sargassum) were secondary resources up to 26% and 29%, respectively. The pelagic components in the other reef lagoons were negligible contributors for D. antillarum diet with just 18-23% of maximum contribution (Table 3, Fig. 4).

Trophic Levels
The overall trophic level data for D. antillarum (TL) ranged from 1.97 to 3.22. The species exhibited significant differences among sites (ANOVA df = 2, F = 10.63, p = 0.0004), and exhibited significant differences between WSE and USE (ANOVA, df = 1, F = 17.7, p = 0.0003). Likewise, calculating the interaction between site*effect (USE and WSE) revealed significant differences (ANOVA, df = 2, F = 12.65, p = 0.0001). The highest TL values were reported for Mahahual USE, while the lowest one was recorded in Xahauayxol WSE. At Mahahual, the TL mean value of D. antillarum was 2.35 ± 0.18 WSE and 3.08 ± 0.13 USE; at Xahuayxol, the TL mean value was 2.13 ± 0.30 WSE and 2.49 ± 0.27 USE, and at Xcalak TL mean value was 2.62 ± 0.15 WSE and 2.45 ± 0.12 USE (Table 4).    Table 5 shows data on isotopic niche breadth as measured by the corrected standard ellipse area (SEAc). The Stable Isotope Bayesian Ellipses in R (SIBER) analysis suggested a reduction in trophic niche particularly in Xcalak. This site showed the main difference in the trophic niche breadth with SEAc of 3.48 and 0.14 (WSE and USE respectively). An overlap of isotopic niches between WSE and USE was only found in Xahuayxol (Fig. 5). SEAc was higher USE in this site with 3.57 versus 2.68 SEAc WSE (Fig. 5).

DISCUSSION
Our results provide evidence of the detrimental effect of Sargassum blooms on the physicochemical water properties and ecological processes in near-shore coral reef communities as recently has been identified in our study area (Rodríguez-Martínez, van Tussenbroek & Jordán-Dahlgren, 2016;van Tussenbroek et al., 2017;Cuevas, Uribe-Martínez & Liceaga-Correa, 2018). Particularly, the results provide evidence for the input of external carbon and nitrogen resulting from Sargassum blooms on benthic communities that alter the nutrient inputs and trophic niche for D. antillarum. These findings contribute to the growing recognition of the role of exogenous nutrient enrichment in modifying natural sources in a food web. Hence the organic matter inputs from Sargassum coupled with hypoxia leads to modification of natural algal resources for D. antillarum. Considering the detrimental effects this likely represents a nutrient limitation to sea urchin herbivory. Onshore Sargassum exhibits physical processes of fragmentation, decomposition and remineralization by bacteria, meiofauna and grazers (Colombini & Chelazzi, 2003). The algae-derived organic matter, product of that decomposition, has an effect on in situ oxygen availability (Haas et al., 2010). Sargassum blooms clearly showed a negative impact hypoxic conditions found at the three studied reef lagoons (Fig. 3). This could ultimately drive the success of the communities' nitrogen fixation, evidenced by depleted values of δ 15 N as reported by Dorado et al. (2012) andFrance (1995).
The dissolved oxygen values in the back reefs of our study areas were lower than the standard values for coral reefs dominated by algae (7.9 ± 0.5 mg l −1 ) according to Haas et al. (2010) and values reported by Camacho-Cruz et al. (2019) for Xahuayxol and Mahahual. This supports ideas from Kendrick et al. (2000) and Haas et al. (2010), who argue that benthic communities linked to reef lagoons are very susceptible to environmental degradation. Some benthic algae play an important play in the transfer of energy and can be catalyzers of oxygen dynamics in reefs due to coral reef associated algae-derived organic matter (Wild et al., 2010).

Isotopic variations in the algal resources
We found that the composition of benthic macroalgae assemblages were different under Sargassum and without Sargassum effect. USE showed a reduction in the taxonomic diversity of macroalgal food sources available to D. antillarum and isotope values presented substantially lower δ 15 N values ( Table 2). The fact that there were fewer available algal sources in the USE condition implies that the trophic chain becomes less complex as the interaction of primary consumers with their resources is reduced (Phillips & Gregg, 2003).
Overall δ 13 C values ranged from −21.98 to −5.65 are similar to ranges reported by Fry & Sherr (1984) and Morillo-Velarde, Briones-Fourzán & Álvarez Filip (2018). Those authors reviewed the δ 13 C data of benthic algae, noting that values ranged between −30 and 5 . δ 15 N overall algae values fluctuated from 0.02 to 5.68 . Despite these values agree with the variation reported in other studies like Owens (1987) and France (1995), we found USE very low, ergo according to Lapointe et al. (2005) andFrance et al. (1998). These low 15 N: 14 N ratios can be indicative of macroalgae living in oligotrophic reefs which experience nitrogen fixation (Montoya, Carpenter & Capone, 2002). In the presence of the leachates of decomposing Sargassum, it is possible that anaerobic bacteria gained significance over other benthic groups (Table 2), (Carpenter & Cox, 1974;Rooker, Turner & Holt, 2006), and could be the cause of the low macroalgal isotopic signatures. On the other hand, high values of δ 15 N in macroalgae are linked to land-based N enrichment sources, being a good indicator of anthropogenic nitrogen inputs (Umezawa et al., 2002) such as sewage discharges (Risk et al., 2009;Lapointe et al., 2011). France (1995) reported nitrogen ranges of marine macroalgae from −3 to 18 . The inconsistencies in this pattern with values of δ 15 N close to atmospheric signature of 0% suggest a fixation of nitrogen. Dorado et al. (2012) associated the depleted values of δ 15 N with nitrogen fixation and its impact on the trophic position of consumers. So, temporal difference between values in this study WSE and USE might be explained by the influence of organic input derived from floating Sargassum dragged components. We considered that it is likely that the Sargassum effect modifies organic matter dynamics. These modifications stem from changes in the oxygen levels, which were consistently reflected in the low δ 15 N values we recorded of for the primary producers.

Status of Diadema antillarum in the Mexican Caribbean
It is important to note that we focused our study on the most abundant species at the three localities and the most important shallow-bottom herbivore on Caribbean reefs (Carpenter, 1981;Hughes, 1994;Aronson & Precht, 2006;Kissling et al., 2014). For the Mexican Caribbean, there has been considerable variation in D. antillarum population data. Jordán-Garza et al. (2008) showed a high presence of D. antillarum with densities of more than 7 ind m −2 in several areas, including our study area. Jorgensen, Espinoza-Ávalos & Bahena-Basave (2008) reported densities of 12.6 ind m −2 after hurricane Dean. According to Maldonado-Sánchez (2018) population density of D. antillarum displayed <1 ind m −2 for five different habitats of the Parque Nacional Arrecifes de Xcalak (PNAX) reef lagoon (back reef, seagrasses, sandy bottoms and reef patches) and the fore reef. The back reef exhibited the highest abundance with an average of 0.5 ind m −2 . However for Mahahual, we registered an average density of 0.6 ind m −2 (N Cabanillas-Terán, pers. obs., 2017), because of the broad variability exhibited in D. antillarum populations from the back reef.

Trophic parameters of D. antillarum
Our results support the evidence that Sargassum blooms impacted δ 15 N differentially among sites, as the ratios of δ 15 N and δ 13 C are determined by their resources (Phillips & Gregg, 2003). It was conspicuous that D. antillarum showed higher δ 15 N values USE at Mahahual.
Although some available resources (e.g., Dictyota and turf) were present in both conditions (WSE and USE), measuring the contribution of algae to the sea urchin tissues can display key information about how consumers assimilate habitat resources and this could reveal information on the degree of disturbance (Layman et al., 2007b). Therefore, it is possible that the ecological role of D. antillarum was different in each site and could be explained by the variation in the number of available resources and a differential assimilation (Table 3). The higher δ 15 N values USE in the muscle of D. antillarum were a result of the synergistic effect determined by resource availability and disturbance condition.
Pelagic sources may provide new sources of food and the possible nitrogen fixation carried out by turf attached to pelagic Sargassum undoubtedly brought a new source of organic matter to basal trophic levels (Rooker, Turner & Holt, 2006). However, those sources were not major contributors for D. antillarum and appear to avoid the invasive pelagic macroalgae. This is consistent with the feeding ecology by marine generalist herbivores (Boudouresque & Verlaque, 2001) and such feeding response is in line with evidence from other sea urchin species in the face of other invasive resources. The experiments carried out by Tomas, Box & Terrados (2011) provide evidence that some seaweed invaders were strongly avoided by Paracentrotus lividus and therefore escape enemy control by reducing herbivore preference.
The mean trophic level for D. antillarum exhibited at Mahahual was 2.35 ± 0.18 WSE up to 3.08 ± 0.13 USE. Hence, WSE supported the idea that this species occupies an herbivorous position. However USE D. antillarum revealed that the species can occupy different trophic niches when faced with resource limitation. Under Sargassum blooms, D. antillarum displayed a position more in line with omnivorous conditions, suggesting trophic level indicative of herbivorous behaviour tending towards omnivory, according to Vander Zanden & Rasmussen (1999). These authors stated that primary consumers have a trophic position of 2.0 (strictly herbivorous); but if organisms assimilate primary consumers, they are considered to be a trophic level of 3.0. The results for Mahahual are consistent with Andrew (1989) who argued that sea urchins could take advantage of ecosystem changes through omnivory if variation exists in the availability of resources. Our results suggest that D. antillarum behave as a facultative omnivore depending on patterns of nutrient availability. δ 15 N signatures for D. antillarum in Mahahual suggest a different carbon source USE. These signatures are also likely the result of anthropogenic nitrogen inputs, as this site has a high eutrophication, being an area with elevated touristic demand (Martínez-Rendis et al., 2016;Arias-González et al., 2017). Furthermore, possible nitrogen fixation by anaerobic bacteria as an important factor in the variation of available sources of food.
Regarding the TL values exhibited for D. antillarum in Mahahual USE 3.08 ± 0.13 versus 2.35 ± 0.18 for WSE would place D. antillarum in an omnivorous position tending towards carnivory. Similar values were obtained from Mediterranean sea urchins as a strategy to avoid exclusion by sympatric species (Wangensteen et al., 2011). However, we cannot state that D. antillarum is carnivorous in Mahahual. This would require a more complete temporal study, and an adjustment of a new δ 15 N baseline for primary producers, considering that 15 N/ 14 N ratios can vary spatially and temporally (Jennings et al., 1997;Vanderklift, Kendrick & Smit, 2006).
The results for Xahuayxol showed also a trend towards higher δ 15 N. However by analyzing the condition of D. antillarum in Xahuayxol no significant differences were observed. We can assume that this locality was least changed in its foraging behavior position against the nutrients modification and the species occupied a lower trophic level WSE. Meanwhile, Xcalak displayed the opposite trend compared to Mahahual and Xahuayxol and USE D. antillarum trophic level was lower than WSE. Our results suggest that for Xcalak the effect of Sargassum blooms completely modified and reduced the possibility for finding available resources, displaying a trophic level around 2.5 between the two scenarios of Sargassum blooms. This corresponds to a predominantly herbivorous to omnivorous condition. Moreover this was confirmed with the isotopic niche breadth data where a reduced niche was observed for Xcalak (Fig. 3).
The rank found for D. antillarum in this study is consistent with the study conducted by Rodríguez- Barreras et al. (2015) in Puerto Rico where microinvertebrates were used as source of organic matter by the sea urchin. Finally, TL values support the premise that echinoids are able to modify their foraging behaviour depending on the availability of resources (Randall, Schroeder & Starck, 1964;Muthiga & McClanahan, 2007), and in this case under Sargassum blooms condition was not only determined by macroalgae availability, but for unusual conditions that caused a shift in the ecosystem (Cabanillas-Terán et al., 2016).

Isotopic niche breadth
The ellipses provide integrated information on the relationship between the availability of sources and the niche width. The results of Mahahual indicated that in USE. D. antillarum consumes different carbon and nitrogen sources (Fig. 4).
Several studies (Lawrence, 1975;Carpenter, 1981;Sammarco, 1982;Hay & Fenical, 1988) noted that echinoids have the ability to adapt their foraging behavior depending on algae availability as well as their population density and site characteristics (Bak, Carpay & De Ruyter Van Steveninck, 1984;Bak, 1994;Alvarado et al., 2016). We observed at Mahahual that USE D. antillarum exhibited a broader trophic niche than WSE. Despite the limited resources this could lead to trophic overlap and stronger habitat degradation. SIAR results showed a resource shift and this could be explained in terms of omnivory as stated by France et al. (1998) ''omnivory is a prevalent attribute of aquatic food webs''.
The trophic niche of Xahuayxol reflects that there was no difference in the use of carbon and nitrogen sources. It is noteworthy that for the case of Xcalak, the resulting isotopic niche of D. antillarum was significantly smaller under Sargassum effect. This is consistent with the metric that associates smaller niche amplitude with disturbed ecosystems (Layman et al., 2007b).

Limitations of the study
To assess the effect of differential management of Sargassum and to effectively evaluate the effect of disposal management, quantitative information on beach disposal would be necessary.
From our results, it is clear that algae communities were modified due to Sargassum. However, due to the structuring role of sea urchins, and, considering that algae respond to temporal variability naturally, it would be necessary to study changing gradients at different time scales. Such a temporal study would provide more conclusive information about the effect of Sargassum spp. on benthic communities.
It is necessary to strengthen the sampling effort to evaluate current population status. A more comprehensive discussion would need to include the interactions with other herbivorous/omnivorous species, that coexist at each site and whether, or how they carry out resource partitioning.
The metrics used in this study allowed us to evaluate the variation of the isotopic signatures that formed the trophic spectrum of D. antillarum under two different scenarios. Metric values based on an instantaneous characterization of a single food web provide a limited view of the food web. Therefore, to evaluate the trophic structure and consequently its functional structure, the most promising evaluations would have to include a comparison of multiple gradients, and, to examine the same food web on a longer temporal perspective.
The deposited biomass regarding to S. fluitans and S. natans did not include a measurement of the total arrived Sargassum blooms. However, our results established a baseline for the amounts that were more available for the echinoids that inhabit the back section of the Caribbean shallow reefs.
It would be challenging to evaluate the ecological role of other coexisting species (Echinometra viridis, E. lucunter and Eucidaris tribuloides), and to include samples of micro-invertebrates. However, this could offer new clues to the connectivity between sympatric species, including trophic loops and successional states of algal communities (Camus, Daroch & Opazo, 2008) within the benthic communities of coral reefs.
• Miguel-Ángel Ruiz-Zárate and Alejandro Vega-Zepeda performed the experiments, analyzed the data, contributed reagents/materials/analysis tools, authored or reviewed drafts of the paper, approved the final draft.
• Alberto Sanchez-Gonzalez analyzed the data, contributed reagents/materials/analysis tools, authored or reviewed drafts of the paper, approved the final draft.

Field Study Permissions
The following information was supplied relating to field study approvals (i.e., approving body and any reference numbers): The collection permit (PPF/DGOPA-002/17) was obtained from the Comisión Nacional de Acuacultura y Pesca (CONAPESCA).