Predation upon Diadema aff . antillarum in barren grounds in the Canary Islands

Experimental studies were carried out to determine the effects of predation on populations of the sea urchin Diadema aff. antillarum in barren grounds at the Canary Islands. The studied urchin populations were dominated by small to medium sized individuals (24-38 mm) and were variable in space. Tethering experiments showed that predation rates on D. aff. antillarum were very low and no differences were found between sites. Predation was found to be most intense on juveniles (<20 mm) and on 20-30 mm sized adults, the size range at which most individuals cease to exhibit cryptic behaviour. Urchins with test diameter >40 mm were not preyed upon whatsoever. We have experimentally demonstrated that there is an absolute predator ‘escape size’ of around 40 mm for D. aff. antillarum individuals in barren grounds. Predation rates obtained for juveniles show that a sufficient number may escape predation and sustain the adult population, maintaining the urchin barren habitat. Recruitment and topographic complexity, rather than predation, seem to determine the structure of urchin populations in barren grounds. We conclude that predation in fished barren grounds of the Canarian Archipelago is not of sufficient magnitude to substantially alter dense urchin populations and cause community-level effects.

encrusting coralline algae (Lawrence, 1975;Mann, 1982;Himmelman and Lavergne, 1985;Vadas and Elner, 1992).Therefore, they have been suggested to be involved in mediating transitions between alternate stable states (Knowlton, 1992;Knowlton, 2004), which occur when more than one type of community can stably persist in a single environmental regime (see review in Beisner et al., 2003).These transitions between alternative stable states are usually sudden and difficult to reverse (Knowlton, 1992).
Predation is one of the strongest biological processes affecting community structure and ecosystem organisation (Hariston et al., 1960;Duffy and Hay, 2001;McClanahan, 1998;Guidetti et al., 2005), and it may exert an important influence on the distribution and abundance of prey organisms (Paine, 1966;Levitan and Genovese, 1989;McClanahan and Muthiga, 1989;McClanahan, 1998).However, particularly in the marine environment, the influence of predators can be difficult to quantify (Aronson and Heck Jr, 1995;Aronson et al., 2001).Predators can have indirect impacts on community organisation, especially when their prey interacts strongly with other species in the community (Power, 1992;Duffy and Hay, 2001).Therefore, their effects may extend beyond the prey consumed and reach an entire ecosystem throughout the so-called 'trophic cascades' (Paine, 1980;Menge, 1995;Witman and Dayton, 2001;Duffy and Hay, 2001;Shears and Babcock, 2002).Removal of top predator populations or severe reductions in their abundance is known to trigger trophic cascades (Steneck, 1998;Pace et al., 1999;Pinnegar et al., 2000;Dulvy et al., 2004;Pinnegar and Polunin, 2004).
The large expansion of sea urchin populations and the reduction of algal forests is linked to the increase in fishing pressure on natures 'keystone predators', as reported for sea urchins at various lat-itudes (Carpenter, 1981;McClanahan and Muthiga, 1988;McClanahan and Muthiga, 1989;McClanahan and Shafir, 1990;Sala and Zabala, 1996;Sala, 1997, Pinnegar et al., 2000;McClanahan, 2000;Tuya et al., 2004b;Tuya et al., 2005a,b).However, other factors such as topography and substrate complexity (McClanahan, 1994;McClanahan et al., 1999;Tomas et al., 2004;Hernández, 2006); recruitment (Underwood and Fairweather, 1989;Hereu et al., 2004); pollution and disease; and the variability of oceanographic events may also be important (see review in Pinnegar et al., 2000).The increased prevalence of urchin-dominated barrens throughout the Canary Islands could also be considered as one symptom of long-standing and intense use of the littoral and fishing resources (Aguilera et al., 1994;Tuya et al., 2004b).Along the eastern Atlantic coast, predator removal has been linked to the subsequent creation of barren grounds as an 'alternate stable state' (Tuya et al., 2004b(Tuya et al., , 2005b)), although there is still little empirical evidence of natural reversals from the 'barren state' back to the macroalgal-dominated state (Tuya et al., 2005b).
Population control exerted by predators may be different at the various stages of the prey life cycle, and certain sea urchin sizes may be particularly vulnerable to predation.Therefore, predation upon recently settled juveniles may be an important factor that limits recruitment and population structure (Tegner and Dayton, 1977;Scheibling and Hamm, 1991;Sala and Zabala, 1996).Several organisms have been identified as predators of D. antillarum in the western Atlantic (Schroeder, 1962;Randall, 1967;Behrents and Wells, 1984;Carpenter, 1984;Levitan and Genovese, 1989), and suggested as predators of this species in the eastern Atlantic (Brito and Falcón, 1990;Brito et al., 2004;Tuya et al., 2004b).
Few experimental or observational studies have been carried out to assess the effect of predation on D. antillarum populations, especially in the eastern Atlantic Ocean (Behrents and Wells, 1984;Carpenter, 1984;Levitan and Genovese, 1989).The evaluation of these predatory populations in the Canary Islands comes from a study that assesses the relationships between Diadema aff.antillarum and potential predator populations (Tuya et al., 2004b).
We hypothesised that (1) predation on D. aff.antillarum should be low or inexistent in areas with a high urchin population and (2) that predation should decrease with the increasing size of individual sea urchins.Also, (3) refuge availability, and therefore substratum rugosity, should determine the level of predation on sea urchins and the density of cryptic versus exposed individuals.The main goal of this study was therefore to evaluate whether urchin barrens of the Canarian Archipelago support any kind of predation pressure on sea urchins and, if evidence of predation is found, to assess its role as a controlling force of D. aff antillarum population structure, determining which sea urchin size class is most susceptible.

Study area
The study was carried out in shallow rocky reefs (4-10 m depth) by means of SCUBA diving from April-July 2005.Four sites were selected in the southeast of Tenerife Island (Canary Islands) in areas where urchin barren habitats were present: Boca Cangrejo, Punta Prieta, Abades and La Jaquita (Fig. 1).

Diadema aff. antillarum population structure
Urchin density and size structure were assessed at each site by randomly placing ten 1m 2 quadrats.Urchins were counted within each quadrat and the test diameter without spines of each individual was measured using vernier callipers (±1 mm).For data analysis purposes, test diameters were categorised into size classes of 4 mm.In addition, it was noted whether individuals were located in a crevice (cryptic position), or were openly grazing the substratum (exposed position without physical protection).This information was used as a measure of the urchin's level of exposure to predators.As an environmental variable, topographic complexity of rocky reefs was estimated using the rope-and-chain method (Luckhurst and Luckhurst, 1978;Kingsford and Battershill, 1998;McClanahan and Shafir, 1990).A flexible tape was pressed along the bottom contour measuring the contour distance of eight 10 m long linear transects laid at each site.The rugosity measure was calculated as the straight-line distance per contour distance; a perfectly flat reef would consequently have a rugosity measure of 1.00.
Data from the four sites were used for comparisons of Diadema aff.antillarum density, size structure and level of exposure.Differences in density and mean test diameter between sampling sites were analysed by 1way ANOVAs and Student-Newman-Keuls (SNK) a posteriori tests.Differences in size of exposed/cryptic urchins between sites were also examined using 1way ANOVA and SNK tests.Before ANOVA analyses, the assumptions of normality and homoscedasticity were tested by Kolmogorov-Smirnov and Levene tests.When assumptions were not met and no transformation rendered variances homogeneous (Underwood, 1997), the ANOVA was carried out as it is robust to heterogeneity of variances, particularly for large balanced experiments (Underwood, 1997).The significance level was thus lowered from 0.05 to 0.01 (Underwood, 1981).
The effect of the urchin position (cryptic or exposed) on individual size was tested by means of the Mann-Whitney U non-parametric test and differences in D. aff.antillarum size distribution were analysed by frequency analysis (χ 2 ).All analyses were performed using the SPSS 12.0 statistics package.

Predation experiment
The level of predation was tested at the study sites by means of a tethering experiment.This technique, which is suitable for sedentary benthic organisms (Aronson et al., 2001), has been used to test predation intensity on sea urchins in tropical ecosystems (McClanahan and Muthiga, 1989;McClanahan and Shafir, 1990;McClanahan, 1998;McClanahan, 1999;McClanahan et al., 1999), in temperate systems of the Mediterranean Sea (Sala andZabala, 1996, Guidetti, 2006) and of the Pacific Ocean (Shears and Babcock, 2002).With this technique, D. aff.antillarum individuals of four different size classes that included juveniles (Class 1: test diameter <20 mm) and adults (Class 2: 20-30 mm, Class 3: 30-40 mm, Class 4: 40-50 mm), were tethered to lines fixed at the substratum (McClanahan and Muthiga, 1989).
As it is difficult to handle Diadema aff.antillarum due to its morphological characteristics, a modified tagging technique was employed, which was formerly used by Olsson and Newton (1977) for the sea urchin Strongylocentrotus franciscanus.The method previously tested and successfully applied in situ to Diadema aff.antillarum (Clemente et al., 2007), consisted in using external tags which were anchored through holes drilled in the urchin tests.This simple technique was not suitable for juvenile D. aff.antillarum (size class 1), because the majority of tests were broken upon piercing (Clemente et al., 2007).Therefore, juvenile tests were perforated through the oral-aboral sections with a hypodermic needle (0.53x88 mm) and threaded with nylon monofilaments (0.25 mm) (McClanahan and Muthiga, 1989), which required removing these individuals temporarily from their habitat.
Ten tagged individuals of each size class were attached at 1 m intervals along 11 m transect lines laid over shallow rocky reefs at depths between 4 and 8 m.A total of four transects and 40 urchins were placed at the same time per site.Each individual was threaded with 40 cm of nylon monofilament, which allowed urchins to move in an area of approximately 0.785 m 2 and usually find holes or crevices to occupy in the substrate.The experiments were visited every 24 hours over 5 days to determine the number of individuals that died during each daily interval and to classify the condition of the carcass.
Examining the condition of the carcass provides crude information about the type of predator that fed on the sea urchin (McClanahan and Muthiga, 1989;Shears and Babcock;2002).Carcasses were classified and were typically found to be: (1) gone -if urchin body could not be found but the nylon tether was still present, which may be due to fish predators such as sparids and labrids that often consume urchins whole (McClanahan, 1995), but the source of predation was unknown; (2) broken -if urchin body was present but broken, which is often attributable to fish predators such as balistids that methodically break open the carcass and leave part of the test (McClanahan, 1995); or (3) with intact test but patches of freshly stripped spines -attributable to predation by starfish Coscinasterias or Marthasterias (Shears and Babcock, 2002).Daily monitoring of the individuals enabled urchins that appeared to be dying as a result of the piercing procedure to be identified.Individuals damaged by the procedure were characterised by intact, bleached tests with spines missing around the hole through which they were pierced (Clemente et al., 2007).Adult sea urchins that lost their tag in the period 1-4 hours before the beginning of the predation experiment were also detected.In order to minimise these effects, these individuals were replaced with new ones and removed from the data analysis.
Survival rate was calculated for each individual urchin; defined as the number of days each D. aff.antillarum individual survived in the experiment.Predation rate was calculated as the total length of the experiment (5 days) minus the survival rate in days.Finally, a relative predation intensity index was calculated for each site and size class dividing predation rate by the length of the experiment (IP= (5-S)/5).The index produces a value between 0 and 1, where 0 corresponds to no sea urchin eaten over the whole experiment, and 1 to all individuals eaten during the first experimental day.
Differences in survival rate of D. aff.antillarum individuals between size classes and sites were analysed using a 2-way ANOVA and Student-Newman-Keuls (SNK) a posteriori test.The factor 'size' (4 levels) was treated as a fixed effect and the factor 'site' (4 levels) as a random effect.The statistical package GMAV5 for windows was used under the specifications of Underwood et al., 2002.
There was significant variation in urchin size between sites (F= 171.720, p<0.001) (Fig. 2).The SNK test shows that the smallest urchin sizes were recorded at La Jaquita (29.70±0.35mm) and Abades (30.46±0.45mm) sites (p=0.329), which were significantly different from those at Boca Cangrejo (41.30±0.58mm) (p<0.01) and from the largest sized urchins recorded at Punta Prieta (44.78±1.21mm) (p<0.01) (Fig. 2).Likewise, pooled data showed significant variation in size of exposed and cryptic urchins (U=24771.50;p<0.001); test diameters of individuals that remained cryptic were lower (28.74±0.52 mm) than those of urchins categorised as exposed (37.84±0.41mm) (Fig. 3).The relationship between Diadema aff.antillarum size and density was found to be significant and negative (r= -0.949, p<0.05).Populations of Diadema aff.antillarum at the studied sites were unimodal; very few urchins had test diameters below 25 mm and those that did generally remained in cryptic positions (Fig. 4).The size-frequency distribution showed significant variations among the 4 sampling sites (χ 2 =587.220,p<0.001) (Fig. 4).The general pattern in terms of sea urchin exposure to predators was that the smallest individuals were cryptic, while exposed urchins usually belonged to larger size classes (Fig. 5).

Predation experiment
From carcass observations, it was easy to distinguish between death caused by tethering and death caused by predation; therefore, individuals dying from the tagging procedure were able to be removed from the data analysis.Only 1.87% of the experimental individuals in all sites died from the handling and tethering procedure.
We registered predation events on Diadema aff.antillarum at all studied sites but only in sea urchin size classes 1-3, as none of the urchins bigger than 40 mm test diameter were consumed (Fig. 6).
Predation intensity was highest for size class 1 (0.19±0.04) and no predation occurred for size class 4. In the latter all individuals remained alive until the end of the experimental period (Fig. 6).Comparison of relative predation intensity indices by 2-way ANOVA shows that 'size' was a significant factor (F= 22.17, df= 3, p<0.001).The SNK test differentiates (p<0.01) between urchins belonging to size classes 1 (<20 mm) and 2 (20-30 mm) (p=0.515), which had the highest predation rates, and the less predated size classes 3 and 4 (p=0.396)(Fig. 6).
The fate of the juvenile urchins (10-20 mm) which were preyed upon was unknown as the tests were completely removed from the tethers (Table 1).This could have resulted from predation by sparid fishes, such as Diplodus cervinus and D. sargus, which are relatively abundant in barren grounds and which completely engulf the urchin, or from predation by invertebrate predators such as Coscinasteras tenuispina, the clearly dominant asteroid at the studied sites, which breaks up or removes small urchins.Of the adult individuals found dead, 66.67% of the 20-30 mm size class and 25% of the 30-40 mm size class were gone, which may be due to fish predation; 33.33% and 75% respectively were present with test intact and patches of stripped spines, a state attributable to C. tenuispina predation (Table 1).The relative predation intensity index ranged within sites between 0.07±0.03 in Punta Prieta and 0.13±0.03 in La Jaquita (Fig. 6).However, the analysis shows no significant effect of the factor 'site' on urchin survival (F= 1.01, df=3, p=0.389) (Fig. 6), and the overall mean predation intensity obtained at barren grounds was 0.09±0.02.

DISCUSSION
The studied sea urchin populations were variable in space in terms of density and size structure.However, the spatial variation seen in urchin density and size in barren ground habitats did not seem to be influenced by predation, as predation pressure was found to be low throughout with no differences found between sites.McClanahan (1998) found similar results in tropical populations of Echinometra mathaei in which the echinoid was the most abundant and dominant species.At the lowest levels of predation E. mathaei individuals often show signs of food limitation and consequently it is thought that their populations are regulated by food resource availability and intra-and inter-specific competition for these resources (McClanahan and Kurtis, 1991;McClanahan, 1998).Not only this species but many echinoids are able to continue to survive with low levels of food (Lawrence, 1975;Ebert, 1980;Black, 1984;Levitan, 1991).Field experiments with Diadema antillarum have shown that this species has the ability to reduce and adjust skeletal body size and metabolic costs as population density fluctuates (Levitan, 1988;Hernández et al., 2006b).The density and size results obtained here for D. aff.antillarum at different sites in the Canary Islands are further evidence of urchin's capability to continue to survive when density increases by shrinking in size, as seen in La Jaquita and Abades.There is also evidence of urchins increasing in size under decreased density, noticed at Punta Prieta and Boca Cangrejo.Moreover, density seems to be higher in sites where reefs have higher topographic complexity, probably as a result of there being more space available to support more dense populations, according to Hernández (2006), who found that the number of recruits positively correlates with rocky reef complexity, possibly enhancing higher number of small individuals.
The size structure of Diadema aff.antillarum populations was clearly unimodal in habitats with high densities which conform barren grounds, typi-cal of fished sites with low levels of size-specific predation (Andrew and Choat, 1982;Shears and Babcock, 2002).Populations were dominated by small to medium sized individuals (24-38 mm), in concordance with the size class spectrum of Tuya et al., (2004b) (15-55 mm).Moreover, that Diadema aff.antillarum was frequently observed exposed on the substrate may be explained by several factors common to barren grounds, as mentioned by Carpenter (1984) and Alves et al., (2001), such as the high urchin densities and the low predation levels reported here along with low abundance of predators (Tuya et al., 2004b).
Juveniles (individuals up to 20 mm test diameter) were always observed in crevices at barren grounds, as has been previously found for Diadema antillarum in the Caribbean (Bak, 1985;Hunte and Younglao, 1988) and for D. aff.antillarum in the Canary Islands (Hernández, 2006).Our tethering experiment results suggest that this is a predator-avoidance response by the most susceptible size class; a strategy which is also common in other echinoids in the presence of predators (Ogden et al., 1973;Tegner and Dayton, 1977;Carpenter, 1984;Hunte and Younglao, 1988;Levitan and Genovese, 1989;McClanahan and Kurtis, 1991;Sala and Zabala, 1996;Tomas et al., 2004).As demonstrated for other species and systems, the presence of shelter can reduce the amount of mortality caused by predation (McClanahan and Shafir, 1990;Hixon and Beets, 1993;Andrew, 1993).In this sense, habitat complexity, in terms of substrate rugosity and availability of spatial refuges, is an important factor determining juvenile escape from predation (Hereu et al., 2005).
Urchins of 20-30 mm diameter were often observed in open areas of the rocky sublittoral.This exposed behaviour pattern in small sized sea urchins indirectly suggests that D. aff.antillarum may regularly escape from predation after having achieved approximately 20 mm in test diameter.Based on the extent of exposure displayed by urchins sized between 20-30 mm, it is thought that overall predation upon them is very low.In addition, 'hyperabundance' of Diadema aff.antillarum, as we found in Abades and La Jaquita, seems to induce a physical stress caused by the saturation of refuges, so that these small sizes are found out of refuges.
Urchins over 40 mm test diameter were not preyed upon at all and predation was very low on individuals sized between 30-40 mm test diameter.Consequently, we have experimentally demonstrated the existence of a predator 'escape size' (sensu Sala, 1997) of around 40 mm test diameter for D. aff.antillarum individuals in barren ground habitats beyond which small fishes cannot effectively predate on sea urchins.These results therefore support the hypothesis that there is a lack of top predators specialised in feeding on such large sized urchins in barren grounds (Tuya et al., 2004b).This is in concordance with the low abundance and small sizes of potential predatory fishes currently found in overexploited barren grounds of the Canarian Archipelago (Falcón et al., 1996;Tuya et al., 2004b).Alternatively, the abundance of these predators is so low that the characteristics of this experimental design did not allow us to detect their effect on the urchin populations investigated.
In most cases, the specific predators responsible for attacks on tethered D. aff.antillarum individuals could not be identified by examining the carcass condition.However, a substantial percentage of predation events on adult urchins, especially of those between 30 and 40 mm, were probably carried out by the starfish Coscinasterias tenuispina.Taking into consideration that C. tenuispina is by far the most abundant invertebrate predator in the Canary Islands (Clemente et al., unpublished manuscript), most predation by fish species is probably limited to sea urchins <30 mm.
In general, predation pressure is very low on Diadema aff.antillarum populations in barren grounds (0.09±0.02) compared with results obtained from tethering experiments using other sea urchin species.Sala and Zabala (1996) found a higher predation rate of 0.36 on Paracentrotus lividus in a marine protected area of the Mediterranean Sea, but a similar value to ours of 0.07 at fished sites.McClanahan (1998) reported an average predation index for Echinometra mathaei in coral reefs off southern Kenya of 0.5.In a study on Echinometra viridis in the Caribbean (Belize) the same author found a mean predation index that ranged between 0.19 and 0.51 depending on the location along the patch reefs (McClanahan, 1999).
Taking into consideration the high urchin densities in fished barren grounds, it is likely that Diadema aff.antillarum population structure cannot be modified by the low level of predation reported in this study.This has facilitated the demographic explosion of D. aff.antillarum on the unprotected coasts of the Canary Islands, as shown for other barren grounds (Andrew and Choat, 1982;Andrew and MacDiarmid, 1991).In this sense, Tuya et al., 2004b found that low abundance and biomass of potential top predatory fish were common in well developed urchin-grazed barrens, which appears to be related to high densities of sea urchins and in turn, to low cover of fleshy macroalgae.The consequence is the establishment of 'undesired' alternate stable states (Knowlton, 2004) in which systems shift from complex, highly diverse and productive states to simplified, low diversity, low productive states.This is another case in which anthropogenic disturbance by removal of top predators has caused dramatic shifts in the organisation and structure of the coastal community.The result has been damage to the resilience of the marine system (Myers and Worm, 2003;Hughes et al., 2005) with the subsequent establishment of 'undesired' organisational states (Knowlton, 2004;Hughes et al., 2005).Furthermore, knowledge of these phase shifts has important implications for future management strategies focused on mediating transitions between alternate states, as 'undesired' states may be highly resistant to restoration.
We conclude that predation in barren grounds in fished areas of the Canarian Archipelago is not of sufficient magnitude to substantially alter such dense urchin populations and cause communitylevel effects.Moreover, and considering the high settlement rates obtained by Hernández et al. (2006a), the predation rate registered for juvenile urchins shows that sufficient numbers of juveniles may be escaping predation and sustaining the adult population to maintain the urchin barren habitat.Therefore, we suggest that the recruitment rate and topographic complexity, rather than predation, determine the structure of urchin populations in barren grounds as proposed by Hereu et al., 2004 andHernández 2006.Further observational and experimental approaches should be used to specifically identify D. aff.antillarum predators both in overexploited barren grounds and at sites with high density and well structured fish populations.It is therefore suggested that research should focus on urchin populations in Marine Protected Areas of the Canary Islands, where controlled fishing effort would result in higher abundances of top predators.and J.C.H. benefited from a postgraduate fellowship provided by 'Ministerio de Educación y Ciencia' of Spain within the 'FPU' program.
FIG. 6. -Mean survival rates ± SE of Diadema aff.antillarum obtained within (a) sites and (b) sea urchin size classes with predation experiments.

TABLE 1 .
-Source of predation on tethered urchins, based on the study of the condition of the carcass in predation experiments.