Introduction

The growth and maintenance of coral reef communities depend on a dynamic process that involves the accumulation and removal of carbonates over time (^{Goldber 2013}). Carbonate removal largely results from bioerosion, which involves the dissolution and excavation of carbonates by a variety of bioeroding organisms of reef systems, notably sponges (^{Nava and Carballo 2008}), polychaetes (^{Hutchings 2008}), molluscs (^{Kiene and Hutchings 1994}), fish (^{Bruggemann et al. 1996}), and sea urchins (^{Glynn 1988}, ^{Perry et al. 2012}).

The feeding activity of sea urchins has been shown to be relevant in reef systems because of their capacity to control algal communities (^{Carpenter 1986}, ^{Hughes et al. 1987}) and facilitate coral recruitment (^{Sammarco 1980}). Moreover, sea urchins are known to play a role in carbon removal (^{Glynn 1988}, ^{Reaka-Kudla et al. 1996}) and, thus, in the growth and maintenance of reef ecosystems (^{Bellwood et al. 2004}).

In the Tropical Eastern Pacific, studies have been conducted on the bioerosion rate of the sea urchins Toxopneustes roseus, Eucidaris thouarsii, and Diadema mexicanum (^{Glynn 1988}, ^{Reaka-Kudla et al. 1996}, ^{Reyes-Bonilla and Calderón-Aguilera 1999}, ^{Herrera-Escalante et al. 2005}) and on their role in coral reefs of Mexico (Gulf of California, Oaxaca), Costa Rica, Panama, and the Galapagos Islands. Because of its ability to remove cabonates (^{Glynn 1988}, ^{Reaka-Kudla et al. 1996}), its high abundance (^{Glynn and Leyte-Morales 1997}, ^{Herrera-Escalante et al. 2005}, ^{Alvarado et al. 2012}), and gregarious behavior (^{Levitan 1988}), D. mexicanum is considered the most relevant echinoderm in reef ecosystems of the eastern Pacific (^{Alvarado et al. 2015}).

In southwestern Mexico, carbonate removal associated with the grazing activity of D. mexicanum has been estimated to be up to 4 kg/m²/yr in Bahías de Huatulco, Oaxaca (^{Herrera-Escalante et al. 2005}). This estimate, however, can vary considerably depending on the temporal changes in sea urchin abundance (^{Benítez-Villalobos et al. 2008}). Despite previous studies in Ixtapa-Zihuatanejo, Guerrero (^{López-Pérez et al. 2012}), and in Puerto Ángel, Oaxaca (^{Leyte-Morales 1997}, ^{Reyes-Bonilla and Leyte-Morales 1998}), data on carbonate removal by sea urchins are lacking for extensive and important coral sections. On the other hand, the carbonate removal estimated for D. mexicanum at Bahías de Huatulco considered the total carbonates without analyzing the composition of the eroded material (^{Glynn 1988}). The latter is of vital importance because how sea urchins affect the analyzed system will depend on the type of material removed. For example, if D. mexicanum primarily removes calcareous material deposited by algae (magnesium calcite), which is rapidly replaced during algal growth, then the effect on the coralline skeleton will be minimal. Conversely, a high aragonite content in the eroded material indicates that grazing by D. mexicanum is having a damaging effect on the coral structure. Hence, the objective of the present study was to determine the effect of the feeding activity of D. mexicanum on coral reefs and communities in three areas on the Pacific coast of southern Mexico: Ixtapa-Zihuatanejo, Puerto Ángel, and Bahías de Huatulco. In addition to estimating carbonate removal, the composition of the eroded material (magnesium calcite vs aragonite) was examined to more precisely determine the effect of the sea urchin on the coral systems of the region.

Materials and methods

Study area

The study was carried out from September 2008 to August 2009 at 15 sites located in the states of Guerrero and Oaxaca, southwestern Mexico (Fig. 1). The sites were grouped into three areas (extreme coordinates of 15°39'42.76"N and 17°40'49.21"N and 96°07'41.50"W and 101°39'32.43"W): the area of Ixtapa-Zihuatanejo (Guerrero) comprised the sites of Morros del Cerro Colorado, Zacatoso, El Chato, Caleta de Chon, and Manzanillo; the area of Puerto Ángel (Oaxaca) comprised the sites of Mazunte, Estacahuite, La Mina, Playa del Muerto, and La Tijera; and the area of Bahías de Huatulco (Oaxaca) comprised the sites of San Agustín, Riscalillo, Isla Cacaluta, Violin, and La Entrega. In all cases, the fringing coral reefs and/or communities, of varying size, were located in bays or behind rocks that protected them from wave action, at depths not exceeding 12 m. Their substrate characteristics are shown in Table 1. The coral communities in the study area are dominated by Pocillopora spp. (>90%) and to a lesser extent by Pavona spp. (<5%) and Porites spp. (<1%) (^{Glynn and Leyte-Morales 1997}; ^{Reyes-Bonilla and Leyte-Morales 1998}; ^{López-Pérez et al. 2012}, ²⁰¹⁴). For a more detailed description of the species composition, substrate cover, size, depth, and other details of these systems see ^{López-Pérez et al. (2012)} for those in Ixtapa-Zihuatanejo; ^{Reyes-Bonilla and Leyte-Morales (1998)} for those in Puerto Ángel; and ^{Glynn and Leyte-Morales (1997)}, ^{López-Pérez et al. (2014)}, and ^{Santiago-Valentín et al. (2015)} for those in Bahías de Huatulco.

Figura 1: Ubicación de los sitios de muestreo en la costa de Guerrero y Oaxaca, sudoeste de México. 

Table 1: Density and size (test diameter) of the sea urchin Diadema mexicamum, reef substrate characteristics, and carbonate budget parameters for each sampling site and area. Mean value ± standard error. 

Results

Spatial analysis of the sea urchin density data revealed significant differences between areas (H = 13.4, P = 0.001). Puerto Ángel had a significantly higher density of sea urchins (Q = 2.394) than Bahías de Huatulco and Ixtapa-Zihuatanejo (Table 1). At a local scale, there were no significant differences among the sites located in Ixtapa-Zihuatanejo (H = 7.1, P = 0.13) and Puerto Ángel (H = 8.5, P = 0.07). While significant differences were observed for Bahías de Huatulco (H = 17.4, P = 0.003), the a posteriori test was unable to discern among which sites these differences occurred.

Regarding sea urchin size, the analysis revealed significant differences (H = 139.5, P = 0.00) in test diameters between areas (Q = 2.394). The test diameter of the individuals was significantly higher in Puerto Ángel, followed by Ixtapa-Zihuatanejo, and lastly Bahías de Huatulco (Table 1). At a smaller spatial scale, the diameter differed among the sites in Ixtapa-Zihuatanejo (H = 25.21, P = 0.00) and Bahías de Huatulco (H = 11.68, P = 0.01). In Ixtapa-Zihuatanejo, the diameter of the sea urchins collected from Manzanillo differed significantly (Q = 2.807) from those found at Zacatoso, El Chato, and Morros del Cerro Colorado (Table 1). In the case of Bahías de Huatulco, the a posteriori test was unable to identify significant differences between the diameters of the sea urchins from the different sites.

According to the principal coordinates analysis, the ordination of the study sites and areas based on substrate characteristics can be described by the first two components (PCO1 and PCO2), which contributed 96.6% of the total variation (PCO1, 74.3%; PCO2, 22.3%) (Fig. 2a). The ordination of the sites shows a gradient running from the Bahías de Huatulco area (left side of the plot), through the Ixtapa-Zihuatanejo area, to the Puerto Ángel area (right side of the plot). This gradient is mainly related to spatial variation in the percentage of coral cover (left portion of the ordination) and rock cover (right portion of the ordination). That is, coral cover is relevant in the Bahías de Huatulco area but loses relevancy towards the Puerto Ángel area where rock cover increases. The other substrate characteristics were not relevant in the ordination.

Figure 2: Ordination of sampling sites and areas by principal coordinates analysis: (a) ordination of sampling sites and areas based on substrate characteristics, (b) ordination of sampling sites and areas based on substrate characteristics and their relation to sea urchin size, (c) ordination of sampling sites and areas based on substrate characteristics and their relation to sea urchin density. I-Z = Ixtapa-Zihuatanejo, Guerrero; PA = Puerto Ángel, Oaxaca; HUX = Bahías de Huatulco, Oaxaca. Bubbles refer to relative sea urchin size and density at each sampling site. Arrow direction refers to the importance of corals and rocks in the ordination. 

Analysis of the data revealed that there was a significant relation between the scores of the first component (PCO1) and the size of D. mexicanum (r ² = 0.79, P = 0.00) (Fig. 2b), as well as between the scores and density of sea urchins (r ² = 0.28, P = 0.03) (Fig. 2c). The latter indicates that 79% of the spatial variation in sea urchin size and 28% of the spatial variation in sea urchin density may be explained by the characteristics of the substrate at each site. That is, the size and density of the organisms is larger at sites with greater rock cover (e.g., Puerto Ángel area) and scant live coral cover. The density of sea urchins is inversely and significantly correlated with coral cover (Spearman r = -0.59, n = 15, P < 0.05) and with the abundance of algae (Spearman r = -0.53, n =15, P < 0.05), but directly correlated with the presence of rocks (Spearman r = 0.53, n = 15, P < 0.05). The diameter of the sea urchins is inversely and significantly correlated with coral cover (Spearman r = -0.82, n = 15, P <0.05) and algal cover (Spearman r = -0.81, n = 15, P < 0.05), and directly correlated with the presence of rocks (Spearman r = 0.78, n = 15, P < 0.05).

Total carbonate removal was directly proportional to sea urchin size (class 1 = 0.052 g/ind/d, class 2 = 0.202 g/ind/d, class 3 = 0.325 g/ind/d); hence, small individuals removed a smaller amount of carbonates than larger individuals. Of the total carbonate removed by sea urchins, 22% corresponded to coralline aragonite and 78% to magnesium calcite; the latter was also significantly higher than the former (t = -6.21, P = 0.000). With regard to removal by carbonate type and sea urchin size, the same relation is observed as for total carbonate, that is, the removal of both coralline aragonite and magnesium calcite was directly proportional to the size of the individuals.

The rate of carbonate accumulation in coral reefs and communities was significantly different (H = 47.17, P = 0.00) among areas (Q = 2.39). Bahías de Huatulco had the highest rate of accumulation, followed by Ixtapa-Zihuatanejo and Puerto Ángel (Table 1). Significant differences were observed among the Ixtapa-Zihuatanejo sites (H = 14.81, P = 0.005) and among the Puerto Ángel sites (H = 10.42, P = 0.03), but not among the Bahías de Huatulco sites (H = 8.50, P = 0.13). In Puerto Ángel, differences were observed among all the sites (Q = 2.80). In Ixtapa-Zihuatanejo (Q = 2.80), the accumulation rates were significantly higher in Zacatoso and Manzanillo than in El Chato, Morros del Cerro Colorado, and Caleta de Chon (Table 1), while Zacatoso presented higher accumulation than Manzanillo.

The rate of carbonate removal due to the feeding activity of D. mexicanum was significantly different among areas (H = 24.78, P = 0.00). The removal rate was higher in Puerto Ángel (Q = 2.39), followed by Bahías de Huatulco and Ixtapa-Zihuatanejo (Table 1). The removal rate among sites varied significantly in Ixtapa-Zihuatanejo (H = 11.96, P = 0.01) and Bahías de Huatulco (H = 12.17, P = 0.03), but not among the Puerto Ángel sites (H = 8.37, P = 0.07). In Ixtapa-Zihuatanejo (Q = 2.39), there were significant differences among all sites, except between Morros del Cerro Colorado and El Chato. In Bahías de Huatulco (Q = 2.93), La Entrega differed significantly from the other sites, Cacaluta differed from Violín and San Agustín, and Riscalillo differed from San Agustín (Table 1).

Finally, the balance between the accumulation and removal of coralline aragonite showed that only 3.1% of the aragonite deposited was removed in the Puerto Ángel area, where bioerosion was more intense. More modest values were obtained for Bahías de Huatulco, followed by Ixtapa-Zihuatanejo. Hence, in the coral reefs in our study area, net accumulation of coralline aragonite (~96.9%) was much higher than removal (~3.1%) caused by the feeding activity of D. mexicanum in all the areas (Table 1).

Discussion

The results showed that the removal of carbonates associated with the feeding activity of D. mexicanum represents a small fraction (~3.2%) of the total of carbonates deposited by reef builders off the Pacific coast of southern Mexico. This suggests that grazing by this species does not compromise the growth and permanence of coral reefs and communities in Ixtapa-Zihuatanejo in Guerrero and in Puerto Ángel and Bahías de Huatulco in Oaxaca. Our findings are contrary to those obtained in the 1980s by ^{Glynn (1988)} in Panama and by ^{Guzmán and Cortés (1989b)} in Costa Rica, where D. mexicanum was found to actively erode the coral systems. They are, however, in agreement with that reported by ^{Reyes-Bonilla and Calderón-Aguilera (1999)}) for Cabo Pulmo, by ^{Herrera-Escalante et al. (2005)} and ^{Benítez-Villalobos et al. (2008)} for Huatulco, and by ^{Alvarado et al. (2012)} for Isla Cocos, Isla Caño, and Bahía Culebra in Costa Rica, where the carbonates removed by sea urchins represented a small fraction of the carbonates deposited by reef builders.

Contrary to the daily removal values previously recorded for D. mexicanum in the eastern Pacific, those obtained in the present study (0.52-0.324 g/ind/d) were lower than those recorded in Panama (1.98 g/ind/d, ^{Glynn 1988}), Huatulco (2.08 g/ind/d, ^{Herrera-Escalante et al. 2005}), and Costa Rica (0.56-2.25 g/ind/d, ^{Alvarado et al. 2012}). Our daily removal values translate into an annual impact of 0.03 kg/m²/yr on the coral reefs and communities of southwestern Mexico, which represents, when compared with lower annual removal values, only 17% of that recorded in other reefs of the region (Panama, 3.47-10.4 kg/m²/yr, ^{Glynn 1988}; Huatulco, 0.17-4 kg/m²/yr, ^{Herrera-Escalante et al. 2005}, ^{Benítez-Villalobos et al. 2008}; Costa Rica, 0.31-0.75 kg/m²/yr, ^{Alvarado et al. 2012}).

Annual carbonate removal by D. mexicanum in coral reefs of southwestern Mexico is closely linked to the feeding activity of this species (^{Glynn 1988}) and its variation depending on sea urchin size (^{Herrera-Escalante et al. 2005}). Carbonate removal is also associated with sea urchin density levels, the spatial distribution patterns (i.e., clustering) of the individuals, and their spatial distribution in the different reef zones (^{Hawkins 1981}, ^{Bak 1994}, ^{Woodley 1999}). Our estimated value for total carbonate content in the feces of D. mexicanum as a function of size falls within the values reported for other other Diadema spp. found in the Caribbean and Pacific Ocean (^{Perry et al. 2012}). However, when the type of carbonate removed is considered, the diet of D. mexicanum includes only ~6-22% of coralline aragonite and the rest corresponds to magnesium calcite. Our results differ strongly from those obtained by ^{Glynn (1988)} in Panama, where D. mexicanum removed a greater proportion (~75%) of coralline aragonite. The difference between values can be attributed to different methodologies, though part of the variation could be the result of differences in the type of substrate on which the species grazes or to spatial differences in its feeding habits. ^{Glynn (1988)} observed that bioerosion by D. mexicanum was two-fold higher on dead coral than on live coral. The effect of grazing by D. mexicanum off southwestern Mexico may be closely linked to the state of conservation of the coral reefs studied. Hence, reefs with large areas of dead coral experience high levels of aragonite removal and vice versa. Moreover, the combined effect of enhanced aragonite removal related to the state of conservation, increased sea urchin density, and the gregarious behavior of D. mexicanum (^{Levitan 1988}) may exacerbate bioerosion of the reef structure. The findings of studies carried out in Costa Rica, Panama, and the Galapagos Islands after the El Niño 1982-1983 event (^{Glynn 1988}, ^{Guzmán and Cortés 1992}, ^{Eakin 1996}, ^{Reaka-Kudla et al. 1996}, ^{Alvarado et al. 2012}) show that impoverishment of the state of conservation of the coral community promotes erosional processes. In southwestern Mexico, coral reefs and communities in the area of Acapulco in Guerrero (^{López-Pérez et al. 2012}), and in Puerto Escondido (López-Pérez, unpublished data), Puerto Ángel (^{Reyes-Bonilla and Leyte-Morales 1998}), and Bahías de Huatulco in Oaxaca (^{Glynn and Leyte-Morales 1997}, ^{López-Pérez and Hernández-Ballesteros 2004}), may have experienced rapid bioerosion after the deterioration of the systems as a result, primarily, of human activities, though data are lacking to confirm this.

In general, the density of D. mexicanum in the coral reefs and communities of southwestern Mexico does not differ from, or is slightly below, the values reported for other reef systems in the Tropical Eastern Pacific (^{Alvarado and Fernández 2005}, ^{Herrera-Escalante et al. 2005}, ^{Benítez-Villalobos et al. 2008}, ^{Alvarado et al. 2012}). In our study area, significant differences were observed in the density of D. mexicanum at different spatial scales. The spatial distribution of sea urchin density and size may be explained by the variation in substrate characteristics. In particular, the density and size of D. mexinanum were greater at those sites with lower coral and algal cover but greater rock presence. These results may indicate the potential present and future effects of bioerosion by D. mexicanum on coral reefs of southwestern Mexico. The highest levels of bioerosion caused by sea urchin grazing were recorded in Puerto Ángel (Oaxaca) and the lowest in Ixtapa-Zihuatanejo (Guerrero). If the present levels of use and state of conservation of the coral systems off the Pacific coast of southern Mexico persist, bioerosion is likely to increase in the area of Puerto Ángel, especially at sites such as La Tijera and Playa del Muerto where large patches (>30%) of dead coral are observed.

Finally, it is important to take into consideration that the present study addresses only the potential effects of the feeding activity of D. mexicanum on coral reefs and communities but does not address the bioerosion associated with the mechanical abrasion of sea urchin spines (^{Herrera-Escalante et al. 2005}). Moreover, the data presented here do not consider the bioerosion caused by sponges (^{Nava and Carballo 2008}; ^{Carballo et al. 2008}, ²⁰¹³), fish (^{Ong and Holland 2010}), molluscs, and other macro and micro bioeroders (^{Perry et al. 2012}) that also contribute to the removal of carbonates from reef ecosystems.