Marine Biology (1997) 129: 531±539

Ó Springer-Verlag 1997

E. Sala

Fish predators and scavengers of the sea urchin Paracentrotus lividus in protected areas of the north-west Mediterranean Sea

Received: 23 April 1997 / Accepted: 30 May 1997

Abstract Direct observations of predation on 436 individuals of the sea urchin Paracentrotus lividus (Lamarck) were carried out in infralittoral rocky bottoms (between 5 and 20 m deep) in three Mediterranean marine reserves. The predator guild was composed of six ®sh species, the sparids Diplodus sargus and D. vulgaris being the main predators, and the labrid Coris julis a major predator of juvenile sea urchins. Four species attempted but failed to open sea urchins. The scavenger guild was most rich in species, with 17 species observed. Predation was size-dependent; the size of predators increased with increasing size of the sea urchins. The presence of two feeding guilds is suggested, one composed of sparids (Diplodus spp.), able to kill juvenile and adult sea urchins, and the other composed of labrids (mainly C. julis), which feed on juvenile sea urchins. To avoid the extension of overgrazed, barren areas created by P. lividus populations, ®sheries' regulations should focus on major sea-urchin predators, chie¯y D. sargus, D. vulgaris and C. julis.

Introduction Sea urchins are important grazers in most marine benthic sublittoral communities (Lawrence 1975; Lawrence and Sammarco 1982; Schiel and Foster 1986; Verlaque 1987a; McClanahan and Mutere 1994). The abundance and species composition of sea urchin populations

Communicated by A. RodrõÂ guez, Puerto Real E. Sala Departament d'Ecologia, Facultat de Biologia, Universitat de Barcelona, Diagonal 645, E-08028 Barcelona, Spain Present address: (&) Scripps Institution of Oceanography, University of California at San Diego, La Jolla, California 92093-0201, USA

appear to be determined by predation (Duggins 1980; Tegner and Dayton 1981; Tegner and Levin 1983; McClanahan and Muthiga 1989; McClanahan and Sha®r 1990; Sala and Zabala 1996), but such factors as recruitment, diseases and physical factors may also be important. Estimates of predation have come from experiments in which tethered sea urchins were attached to lines and the number of urchins removed was recorded (McClanahan and Muthiga 1989; McClanahan and Sha®r 1990). Sala and Zabala found that ®shes such as Diplodus sargus and other sparids were the main predators of the sea urchin Paracentrotus lividus (Lamarck) in both protected (100% of total predation on experimentally tethered urchins) and unprotected (60%) areas in the northwestern Mediterranean. D. sargus is therefore likely to play a major role in controlling sea urchin abundance, behaviour, and their e€ects on phytobenthic communities (Sala 1996). However, most studies involve experiments where predation is not directly observed, or rely on evidence such as gut-content analysis, which allows evaluation of the importance of predation in natural conditions but does not quantify the importance of a species as predator or scavenger. Knowledge of sea urchin predators and scavengers is poor in the Mediterranean Sea, due to the scarcity of direct observations. Many species of ®shes have been identi®ed as sea urchin consumers (see review by Savy 1987), but no distinction has been drawn between predator and scavenger species ± a critical distinction for management purposes. Only one study, conducted in Kenyan coral reefs, has exhaustively investigated sea urchin predators and scavengers by direct observation (McClanahan 1995a). Such knowledge could be very useful for coastal management, both for ®sheries and conservation, as sea urchins are dominant competitors for food in the absence of predation. Fisheries management could be simpli®ed by focusing ®shery regulations on the predators of sea urchins (McClanahan and Sha®r 1990; McClanahan 1995b). Therefore, marine ecosystem management achieved via predator control necessitates the identi®cation of sea urchin predators

532

and that these be distinguished from scavengers (McClanahan 1995a). In this paper, I present observations of ®sh predation on the common sea urchin Paracentrotus lividus collected over a 7 mo period in three protected areas in the Mediterranean Sea.

Study sites and methods Observations were made while SCUBA-diving in the Scandola Nature Reserve (42°20¢N; 8°35¢E: Corsica, France), and the Cabrera National Park (39°09¢N; 2°55¢E) and the Medes Islands Protected Area (42°02¢N; 3°13¢E) (both in Spain), between April and October 1996 (Table 1). Protected areas were chosen as study locations mainly because the high density and large mean size of predatory ®shes (e.g. Garcõ a-Rubies and Zabala 1990; Francour 1991) in these areas facilitate observations of predation. Scandola is a reserve with two levels of protection: (1) a small central area, where all ®shing is prohibited (72 ha); and (2) a larger peripheral area where local ®shermen are authorized to ®sh (928 ha). Observations were made in both zones. Recreational ®shing has been prohibited at Cabrera Island since 1991, but local ®shermen are authorized to ®sh within the National Park. The Medes Islands include a central area of reinforced protection (93 ha) wherein all ®shing is prohibited, and a peripheral area where angling and professional ®shing are allowed within certain limitations. Observations were carried out only in the central protected area. These three protected areas are located at some distance from each other, and hence display di€erences in their physical environment and in the frequency and species composition of sea urchin predators. Therefore, the observations conducted in each protected area are presented separately. Juvenile (hereafter sea urchins with a test diameter without spines of <1 cm) and adult Paracentrotus lividus (Lamarck) (£ 6 cm) were collected by SCUBA-diving from rocky substrates near the study sites. Juveniles were carefully collected from shelters (crevices and beneath boulders) with pincers, and kept in a plastic case. Adults were kept in a mesh bag. No damaged urchins were used for observations. Urchins were measured with a calliper and grouped into size classes of 1 cm. At a variety of sites in bottoms containing big boulders within the protected areas (Table 1), a small number of urchins (4 to 5 juveniles or 1 to 2 adults) were placed on the substrate while the observer hid behind a big boulder or ¯oated at a distance of 5 to 10 m from the baited area (McClanahan 1995a). In total, a similar number of urchins from each size class was o€ered to ®shes. Observations of predation on juveniles were carried out at 5 to 10 m depth, and those on adults at 5 to 20 m depth. Preliminary tests showed that the largest sea urchins (>5 cm diam), which were seldom attacked by ®shes, were attacked on the oral side when turned upside down. Thus, facilitating access to the oral side of sea urchins could bias observational study. As the aim of this study was to observe the predators of sea urchins under natural conditions, experimental urchins were therefore placed in a ``natural'' position, with the oral side on the substratum. No signi®cant night-time predatory ®shes are known (Savy 1987; Sala 1996); therefore ®eld observations were made during Table 1 Description of protected areas and sites, and number of predation observations [Whole protected area area of limited ®shing restrictions; Central protected area extent of area in which all ®shing is prohibited; No. of experiments/site number (range) of non-opened sea urchins (Paracentrotus lividus) o€ered to ®sh at each site]

Year protection established Whole protected area Central protected area No. of study sites Depth range of observations No. of experiments/site No. of predation observations

daylight. The observer recorded the size of the experimental urchin and the number of individuals of each predator species. Observations were categorized as follows (McClanahan 1995a): (1) predator guild ± species that break open the sea urchin test (body wall); (2) attempted predator group ± species that bite the test but fail to open it; (3) scavenger guild ± species that bite an already opened urchin. In addition, the observer estimated the size of the predator (total length) to the nearest 2 cm with the aid of a plastic ruler placed in the baited area, and recorded the time required to eat the urchin in its entirety to the nearest 5 s. When predators carried the carcasses to crevices or away from the observer's ®eld of vision, ingestion time was not recorded. Terminal-male wrasses (Coris julis and Thalassoma pavo) were distinguished from females by their colouration (Corbera et al. 1996). In one set of experiments I opened sea urchins and conducted a census of the scavengers that bit the already-opened tests to determine any di€erences between this and the ``natural'' scavenger guild. Data analyses include calculations of average body size of ®sh, ingestion time, and guild species-diversity (Simpson's index: P D ˆ 1 ÿ pi2 , where pi = number of individuals of species i divided by total number of individuals in the guild; D = 0 is the smallest and D ' 1 is the greatest possible diversity) for each of the three categories. Predatory species were also grouped into taxonomic family categories for body length and ingestion-time analyses. Comparisons of the size of predators and ingestion time among sea urchin size classes were carried out by one-way ANOVA.

Results A total of 436 observations of predation on Paracentrotus lividus were made during the 7 mo study period, and a total of 18 ®sh species that consumed sea urchins were observed (Table 2). In Scandola, 164 observations were made: Diplodus sargus was the most prevalent predator (73% of observations), followed in decreasing order by Coris julis, D. vulgaris and Labrus merula (Fig. 1). Sea urchins >1 cm in diameter were only preyed on by sparids (D. sargus and D. vulgaris), whereas all four observed predators killed juvenile sea urchins, the labrid C. julis being their main predator (55% of observations; Table 3). The attempted predator group was composed of three species in Scandola, D. sargus making by far the largest number of unsuccessful attacks (80% of observations; Fig. 1). Only sparids attempted predation on sea urchins of >1 cm diam (Table 3). In Scandola, three species of scavengers were observed feeding on sea urchin carcasses opened by predatory ®sh (Fig. 1). With the exception of D. vulgaris, the ranking of scavengers was the same as that of predators, with D. sargus being Scandola Nature Reserve

Cabrera National Park

Medes Islands protected area

1975

1991

1983

1000 ha 72 ha 5 5±20 m 23±60 164

8164 ha 0 ha 4 6±15 m 39±75 117

418 ha 93 ha 4 5±15 m 50±100 155

533 Table 2 Fish species investigated in this study Serranidae Serranus cabrilla (Linnaeus, 1758) Serranus scriba (Linnaeus, 1758) Sparidae Diplodus annularis (Linnaeus, 1758) Diplodus cervinus (Lowe, 1838) Diplodus sargus (Linnaeus, 1758) Diplodus vulgaris (E.G. Saint-Hilaire, 1817) Oblada melanura (Linnaeus, 1758) Sparus aurata (Linnaeus, 1758) Spondyliosoma cantharus (Linnaeus, 1758) Pomacentridae Chromis chromis (Linnaeus, 1758) Labridae Coris julis (Linnaeus, 1758) Labrus merula (Linnaeus, 1758) Symphodus melanocercus (Risso, 1810) Symphodus ocellatus (ForsskaÊl, 1775) Symphodus roissali (Risso, 1810) Symphodus tinca (Linnaeus, 1758) Thalassoma pavo (Linnaeus, 1758) Blenniidae Parablennius rouxi (Cocco, 1833)

Fig. 1 Fish species which a killed and ate (Predator guild), b attempted but failed to break test (Attempted predator group), and c bit already opened carcass (Scavenger guild) of Paracentrotus lividus in Scandola Nature Reserve (Corsica, France). Total number of observations, species, and diversity (Simpson's index) are presented

the most important scavenger, followed by C. julis and L. merula. In Cabrera, 117 observations of predation were made, whereby Diplodus vulgaris was the main predator (62% of observations), followed by D. sargus, and the labrids Coris julis and Thalassoma pavo (Fig. 2). As observed in Scandola, only sparids preyed on sea urchins of >1 cm diam (Table 3). Furthermore, D. vulgaris was the most important predator of juveniles (74% of observations). The attempted predator group was composed and structured in the same way as the predator guild, with the same species and ranking order (Fig. 2). Most unsuccessful attacks on adult sea urchins were carried out by the sparids (Table 3). The scavenger guild was the most diverse in Cabrera, with ®ve species observed (Fig. 2). T. pavo and D. vulgaris were the main scavenger ®shes. The Medes exhibited a pattern similar to that of Scandola, but distinct from that of Cabrera. Although ®ve ®shes were observed preying on Paracentrotus lividus (making a total of 155 successful attacks), the diversity of the predator guild was by far the lowest due to the prevalence of Diplodus sargus as predator (92% of observations; Fig. 3). As at the other sites, only sparids preyed on adult sea urchins (except for one observation of Labrus merula; Table 3). Here also, the sparid D. sargus was the species with the highest number of unsuccessful attacks (Fig. 3). The scavenger guild also showed the lowest diversity, due to the predominance of D. sargus (99% of observations; Fig. 3). Urchins of >5 cm diam were also o€ered to ®shes. The latter investigated them for a few seconds before ignoring them. Diplodus sargus utilized a number of techniques to ingest sea urchins. When sea urchins were very small (<1 cm), the ®sh swallowed urchins whole using suction. When the ®sh was not able to ingest the whole sea urchin, it broke the urchin with a strong bite. In the case of small urchins (<3 cm), the ®sh was able to break the test with a few bites, and then rapidly ingest the urchin. With larger sea urchins, ®sh had diculty biting eciently because of the large spines. In this case, the ®sh bit and broke the spines, turned the urchin upside down, and ®nally bit and broke the test. D. sargus were gregarious when preying on Paracentrotus lividus, with several individuals trying to break open the test at the same time. However, when feeding on sea urchins, D. sargus displayed a highly agressive behaviour towards other ®sh species, and consumed the urchins relatively rapidly; thus, scavenger species had diculty feeding on opened carcasses. The labrids Coris julis and Thalassoma pavo swallowed small juvenile sea urchins whole. The diculty of ingesting juveniles using suction increased with increasing urchin size. When a sea urchin was too big to be swallowed in its entirety but small enough to be taken with the mouth, the ®sh carried the urchin o€ and searched for a barren area. There, the ®sh opened the urchin by hitting it against bare rock. Once the urchin

534 Table 3 Paracentrotus lividus. Number of predation events on sea urchins of di€erent test sizes in the three protected areas

Area Guild/group Species Scandola Predator guild Diplodus sargus Diplodus vulgaris Coris julis Labrus merula Attempted predator group Diplodus sargus Diplodus vulgaris Coris julis Cabrera Predator guild Diplodus sargus Diplodus vulgaris Coris julis Thalassoma pavo Attempted predator group Diplodus sargus Diplodus vulgaris Coris julis Thalassoma pavo Medes Predator guild Diplodus sargus Coris julis Diplodus vulgaris Labrus merula Sparus aurata Attempted predator group Diplodus sargus Coris julis Diplodus vulgaris

test had been broken, the ®sh ejected the pieces of test and swallowed them one after the other. When sea urchin size was above the breakable limit relative to ®sh size, all predatory species behaved similarly: they carefully examined the sea urchins before trying to bite the test. If the test size was such that they would be unable to break it, most predators abandoned the prey. Terminal-male and female Coris julis were observed preying on juvenile sea urchins in the same proportions, but females displayed considerably more frequent unsuccessful attacks, probably as a result of their smaller size (Fig. 4). Terminal-male C. julis carried out 50% of the total C. julis attacks (successful plus failed). Thalassoma pavo females were observed killing juvenile sea urchins twice as frequently as males (Fig. 4). Only females carried out unsuccessful predatory attacks, also probably as a result of their smaller size. The size of Sparidae predators increased with increasing sea urchin size (Fig. 5; one-way ANOVA: df = 4, F = 82.5, p < 0.001). The length of Sparidae preying on juvenile sea urchins was greater than that of Labridae preying on the same size group (Fig. 5; Student's t-test: t = )8.3, df = 159, p < 0.001). For sparids, the time it took to eat a whole urchin di€ered

Sea urchin diameter <1 cm

1±2 cm

2±3 cm

3±4 cm

4±5 cm

15 8 29 1

49 5 0 0

55 2 0 0

0 0 0 0

0 0 7

17 5 0

67 10 0

2 0 0

1 38 9 3

18 35 0 0

12 0 0 0

1 0 0 0

0 3 3 0

6 19 0 1

7 7 0 0

47 5 5 0 0

32 0 0 1 0

35 0 0 0 0

18 0 0 0 0

11 0 0 0 1

0 1 1

8 0 0

32 0 0

41 0 0

57 0 0

signi®cantly between sea urchin size-classes (Fig. 5; oneway ANOVA: df = 4, F = 28.0, p < 0.001). The ingestion time for sparid ®shes increased with increasing sea urchin size up to 3 cm diam; thereafter, no signi®cant di€erences were observed with increasing sea urchin size. The average ingestion time (all sea urchin sizes combined) was 34 s, 5 s for juveniles, and <90 s for large urchins. Diplodus sargus took only a few seconds to break a sea-urchin test, but spent a signi®cantly longer time consuming the whole urchin because of the time spent in pursuit of carcass pieces. Labrids took much more time (nine times) than sparids to ingest a whole juvenile sea urchin (Fig. 5; Student's t-test: t = 6.8, df = 157, p < 0.001). Scavenging of sea-urchin carcasses opened by the observer di€ered greatly from that after predation attacks by ®shes (Fig. 6). In Scandola and Cabrera, the scavenger guild on the carcasses opened by the observer was richer in species (and diversity) than that on sea urchins broken open by predatory ®shes: the number of scavenger species rose to 14 in Scandola and to 7 in Cabrera; in the Medes Islands, however, the scavengers were the same species observed after predation by ®shes.

535

Fig. 2 Fish species preying/scavenging on Paracentrotus lividus in Cabrera National Park (Balearic Islands, Spain). Further details as in legend to Fig. 1

Discussion The number of predators observed in this study (six species) is considerably smaller than the number of species described as consumers of Paracentrotus lividus from stomach-contents analysis (18 species: Savy 1987). Although several ®sh species may ingest sea urchins, only a few are able to break sea-urchin tests. I suggest that these predatory species (chie¯y the two Diplodus species) play a key role in structuring the Mediterranean rocky infralittoral. Another study found that a decrease in these predatory species and others may be a major cause for the high number of sea urchins in some infralittoral sites (Sala and Zabala 1996). The present study may have failed to observe other predatory species, especially the predators of juvenile sea urchins. Nonetheless, because of the frequency of observations and their high density, the observed predators appear to be the main predatory ®shes. Another explanation of the low diversity of the predator guild may be the aggressive behaviour of D. sargus when feeding: this behaviour may exclude competitors. This is especially true in

Fig. 3 Fish species preying/scavenging on Paracentrotus lividus in Medes Islands Protected Area (Catalonia, Spain). Further details as in legend to Fig. 1

Medes, where D. sargus make up to 56% of the total ®sh wet weight (GarcõÂ a-Rubies 1997). The predatory role of other species, able to forage on large sea urchins such as Sparus aurata, could not be determined because they appeared to be shy in the presence of SCUBA divers. Some variation in the prevalence of sea urchin predators in the protected areas was observed. For example, Diplodus sargus was the main predatory species of large sea urchins in Scandola and Medes, whereas in Cabrera it was D. vulgaris. The frequency of observations may be related to the density of ®shes: D. vulgaris is much more abundant than D. sargus in Cabrera (GarcõÂ a-Rubies 1993), while in Scandola (Francour 1991) and Medes (GarcõÂ a-Rubies 1997) the pattern is reversed. Predators of juvenile sea urchins showed a similar pattern. Thalassoma pavo was a predator only in Cabrera, and densities of this species were very low in the other protected areas (Francour 1991; GarcõÂ a-Rubies 1997). Therefore, species composition may vary between sites, and switches in predatory species may occur. The proportion of attacks of terminal-male Coris julis (50% of total C. julis attacks) was strikingly high relative to the proportion of terminal-males in most Mediterra-

536

Fig. 4 Coris julis and Thalassoma pavo. Frequency (combined data from the three protected areas) of successful and attempted (but unsuccessful) predation of Paracentrotus lividus by terminal-male and female wrasses

Fig. 6 Fish species which bit Paracentrotus lividus carcass previously opened by observer in Scandola Nature Reserve (Corsica, France), Cabrera National Park (Balearic Islands, Spain), and Medes Islands Protected Area (Catalonia, Spain). Total number of observations, species, and diversity (Simpson's index) are presented

Fig. 5 Sparidae and Labridae. Total length (mean ‹ SE) and time required to kill and consume juvenile and adult Paracentrotus lividus (£ 5 cm diam) by Sparidae (mostly Diplodus sargus and D. vulgaris) and Labridae (mostly Coris julis and a few Thalassoma pavo and Labrus merula) (combined data from the three protected areas)

nean populations [e.g. 12.3% in the Medes Islands Protected Area (GarcõÂ a-Rubies 1997) and 9.5% in the Carry-le-Rouet Reserve, France (Harmelin et al. 1995)]. This suggests that terminal-males are the dominant consumers within C. julis populations ± independent of their higher ability to prey on sea urchins due to their larger size. These results support the observations of social dominance of terminal-males carried out by Harmelin et al., who observed that they are among the ®rst to be caught by angling when ®shing at a particular spot, resulting in a decrease of male numbers in zones subjected to heavy angling pressure. The predator guild and the attempted predator group were composed mostly of the same species. The attempted predator group was composed of either ®shes which (1) took part in a gregarious attack but failed to break open the test (such as Diplodus sargus), or (2) small ®sh such as Coris julis which attempted to bite a large test. Furthermore, successive observations could result in a single individual ®sh being classi®ed in all three guild categories used in this study.

537

The number of ``natural'' scavengers (i.e. the scavengers that bite a sea urchin carcass already opened by a ®sh) was lower than that of scavengers which bit a carcass broken by the observer. This could be due to the facts that (1) Diplodus sargus, which are gregarious and are important scavengers, display an aggressive behaviour that prevents other scavenger species approaching the baited area, and (2) small labrid predators of juvenile sea urchins ingest these in their entirety. The ®rst hypothesis is supported by observations of scavengers on sea urchins broken by the observer: once the observer had broken the sea urchin test, the scavenger species in the vicinity hastened to feed on the opened carcass before the arrival of the bigger D. sargus. Therefore, the number of potential scavenger species may be larger than the number of scavengers observed in this study. Nevertheless, predatory species appear to constitute the most important scavenger species also. Sala and Zabala (1996) suggested that Diplodus sargus is a key predatory species in the northwestern Mediterranean rocky infralittoral. The results of the present study suggest the presence of two important feeding guilds, one composed of sparids (D. sargus and D. vulgaris), which can prey on several sea urchin sizeclasses, even large adults, and the other composed of labrids, which only prey on juveniles. The latter guild is composed mainly of Coris julis, a highly abundant species in most infralittoral habitats, although other labrids such as Labrus merula, Symphodus spp. and Thalassoma pavo also prey on recruits (Quignard 1966; Lejeune 1985; and this study). Although the relative biomass of sea urchin recruits ingested by these ®shes is small (e.g. 3% by weight of the C. julis diet: Sala 1996), the e€ect of predation on Paracentrotus lividus population dynamics may be of some importance. The interaction between prey populations and consumers may involve but a small transfer of energy, but could nevertheless be central to prey dynamics when consumers feed on population bottlenecks (Polis and Strong 1996). The results obtained in this study may have been biased somewhat by the experimental procedure. Sea urchins, especially the smallest, were o€ered outside their natural shelters, thus increasing their susceptibility to be eaten by ®shes. Nevertheless, small sea urchins are usually consumed by littoral ®shes (Khoury 1987; Sala 1996). Natural predation on these small sea urchins implies that nurseries exist in more accessible habitats, such as dense algal communities in which ®shes forage. On the other hand, the circadian activity of both predators and prey may allow encounters at particular times, probably at twilight hours, when small sea urchins begin to move from shelters to graze in more exposed locations (Kempf 1962) and the maximum predatory activity of ®shes is reported to occur (Bell and HarmelinVivien 1983; Khoury 1987). During the daytime, sea urchins are usually ®rmly attached to the surface of rocks by their podia. The experimental procedure could have increased the ease with which the sea urchins were turned upside down,

since many were attacked within the ®rst minute of being placed on the substrate. Another possible bias is the fact that the Medes reserve is intensively frequented by sportdivers, who often feed the ®shes (including breaking-open sea urchins), and may have some e€ect on the behaviour of dominant ®shes (mainly Diplodus sargus) by exaggerating their aggressiveness and by conditioning them (diver = food). The depth at which the experiments were carried out may also have a€ected predator and scavenger frequency. Although the experiments were carried out between 5 and 20 m, most observations were in shallow water (between 5 and 10 m), where Thalassoma pavo and small Diplodus spp. are most frequently found. Had my observations been conducted in deeper waters, perhaps fewer members of these species would have been present and possibly other species would have been more frequently observed preying on sea urchins. Although the observer was hidden or swam at a prudent distance from the baited area, his presence may have biased the frequency of observations of shy species. In addition, observations were restricted to the ®sh present within the areas of attraction. In a previous study in the Medes Islands, Sala and Zabala (1996) found that small Paracentrotus lividus usually sheltered, whereas large sea urchins (>4 cm) were usually exposed to predators. In the present study, Diplodus sargus made but a low proportion of successful attacks on sea urchins of >4 cm. These observations suggest that P. lividus of 4 to 5 cm escape ®sh predation. Of major concern in the management of Mediterranean littoral ecosystems, are sea urchin population increases that can radically change the benthic community structure by overgrazing (Verlaque 1987b, 1996). Local population outbreaks may arise from factors other than ®shing, such as pollution by organic matter (e.g. Harmelin et al. 1981; Pancucci et al. 1992). Nevertheless, regardless of the causes of population increases, ecosystem managers must bear in mind that the regulation of sea urchin populations may require the recovery of populations of both large and small predatory ®shes. Thus, the widespread practice in the Mediterranean of prohibiting spear®shing but not other kinds of ®shing such as angling in some protected areas may be insucient to control sea urchin populations. Angling causes a decrease in the mean size of labrids (which are not ®shed by spear but generally by angling) within unprotected ®shing areas (Garcõ a-Rubies and Zabala 1990; Harmelin et al. 1995; Garcõ a-Rubies 1997). Harmelin et al. established two ®sh-target species-groups, related to the ®shing method, as indicators of the ``reserve e€ect'' (i.e. the changes in ®sh populations after prohibition of ®shing): (1) Type A target-species, comprising Diplodus spp., particularly threatened by spear®shing; and (2) Type B target-species, comprising Coris julis, particularly a€ected by angling. The present study has demonstrated a positive relationship between ®sh size and ability to prey on sea urchins. Thus, the ongoing reduction in the mean size of ®shes in ®shing areas is

538

resulting in a smaller ``escape size'' for sea urchins, which may facilitate sea urchin proliferation. The point at which this escape size is established is of considerable interest to ecological management. I therefore suggest that in order to achieve a successful ®sh predation on the whole size-spectrum of Paracentrotus lividus, ®sheries regulation should not be con®ned to spear®shing, but should be extended to cover all types of ®shing. Fishes are not the only consumers of Paracentrotus lividus in the Mediterranean (Savy 1987), but little attention has been paid to the importance of other sea urchin predators (Dance and Savy 1987). Fishes are probably the most important sea urchin predators during daylight hours, but other organisms may be important predators at night. For instance, the seastar Marthasterias glacialis (L.) displays nocturnal activity and might have a notable impact on P. lividus populations (Dance and Savy 1987). On the other hand, man has proved to be the most ecient sea urchin predator in some Mediterranean countries, mainly France. Conversely to the situation in ®shed areas where sea urchins are not harvested (and their density is high), the impact of human predation (professional and recreational) on large-sized sea urchin stocks could prove very important, and should be considered as an ecient controlling action outside protected areas. Acknowledgements I am grateful to T.R. McClanahan, again a source of inspiration, for his discussion of this paper. I am also indebted to C.F. Boudouresque, J.G. Harmelin, M. HarmelinVivien, M. Verlaque and anonymous referees who improved the manuscript, and to A. Garcõ a-Rubies who kindly provided useful information. I thank E. Ballesteros very much for allowing me to share a ®eld trip to Cabrera and for many other things. Thanks also to the rangers of the Scandola Nature Reserve, F. Arrighi, J.M. Dominici and F. Finelli, who, by providing excellent conditions and a friendly atmosphere, made the work possible in Corsica. This study was possible thanks to the permission of the managers of the Scandola Nature Reserve (Corsica, France), the Cabrera National Park (Balearic Islands, Spain) and the Medes Islands Protected Area (Catalonia, Spain). Research was partially supported by the Parc Naturel ReÂgional de Corse.

References Bell JD, Harmelin-Vivien ML (1983) Fish fauna of French Mediterranean Posidonia oceanica meadows. 2. Feeding habits. TeÂthys 11: 1±14 Corbera J, SabateÂs A, Garcõ a-Rubies A (1996) Peces de mar de la Penõ nsula IbeÂrica. Planeta, Barcelona Dance C, Savy S (1987) Predation on Paracentrotus lividus by Marthasterias glacialis: an in-situ experiment at Port-Cros (France, Mediterranean). Posidonia Newsl 1: 35±41 Duggins DO (1980) Kelp beds and sea-otters: an experimental approach. Ecology 61: 447±453 Francour P (1991) The e€ect of protection level on a coastal ®sh community at Scandola, Corsica. Rev Ecol (Terre Vie) 46: 65± 81 Garcõ a-Rubies A (1993) Distribucio batimeÁtrica dels peixos litorals de substrat rocoÂs a l'illa de Cabrera. In: Alcover JA, Ballesteros E, FornoÂs J (eds) HistoÁria natural de l'arxipeÁlag de Cabrera. CSIC±Moll, Societat HistoÁria Natural Balears, Palma de Mallorca, pp 645±661

Garcõ a-Rubies A (1997) Estudi ecoloÁgic de les poblacions de peixos litorals sobre substrat rocoÂs a la MediterraÁnia Occidental: efecte de la fondaÁria, el substrat, l'estacionalitat i la proteccioÂ. Tesi Doctoral. Universitat de Barcelona, Barcelona Garcõ a-Rubies A, Zabala M (1990) E€ects of total ®shing prohibition on the rocky ®sh assemblages of Medes Islands marine reserve (NW Mediterranean). Scientia mar 54: 317±328 Harmelin JG, Bachet F, Garcõ a F (1995) Mediterranean marine reserves: ®sh indices as tests of protection eciency. Pubbl Staz zool Napoli (I: Mar Ecol) 16: 233±250 Harmelin JG, Bouchon C, Hong JS (1981) Impact de la pollution sur la distribution des eÂchinodermes des substrats durs en Provence (MeÂditerraneÂe nord-occidentale). TeÂthys 10: 13± 36 Kempf (1982) Recherches d'eÂcologie compareÂe sur Paracentrotus lividus (Lmk.) et Arbacia lixula (L.). Recl Trav Stn mar Endoume 25: 47±116 Khoury C (1987) Ichtyofaune des herbiers de posidonies du Parc National de Port-Cros: composition, eÂthologie alimentaire et roÃle dans le reÂseau trophique. TheÁse Doctorat. Universite d'Aix-Marseille II, Marseille Lawrence JM (1975) On the relationships between marine plants and sea urchins. Oceanogr mar Biol A Rev 13: 213±286 Lawrence JM, Sammarco PW (1982) E€ects of feeding in the environment: Echinoidea. In: Jangoux M, Lawrence JM (eds) Echinoderm nutrition. Balkema, Rotterdam, pp 499±519 Lejeune P (1985) Le comportement social des labrideÂs meÂditerraneÂens. Cah EÂthol appl 5: 1±208 McClanahan TR (1995a) Fish predators and scavengers of the sea urchin Echinometra mathaei in Kenyan coral-reef marine parks. Envir Biol Fish 43: 187±193 McClanahan TR (1995b) A coral reef ecosystem-®sheries model: impacts of ®shing intensity and catch selection on reef structure and processes. Ecol Modelling 80: 1±19 McClanahan TR, Mutere JC (1994) Coral and sea urchin assemblage structure and interrelationships in Kenyan reef lagoons. Hydrobiologia 286: 109±124 McClanahan TR, Muthiga NA (1989) Patterns of predation on a sea urchin, Echinometra mathaei (de Blainville), on Kenyan coral reefs. J exp mar Biol Ecol 126: 77±94 McClanahan TR, Sha®r SH (1990) Causes and consequences of sea urchin abundance and diversity in Kenyan coral reef lagoons. Oecologia 83: 362±370 Pancucci MA, Panayotidis P, Zenetos A (1992) Morphological changes in sea urchin populations as a response to environmental stress. In: Aldrich JC (ed) Proc 27th Eur mar Biol Symp 247±257 [Aldrich JC (ed) Japaga, Ashford, Dublin] Polis GA, Strong DR (1996) Food web complexity and community dynamics. Am Nat 147: 813±846 Quignard JP (1966) Recherches sur les Labridae des coÃtes europeÂennes. SysteÂmatique et biologie. Causse et Castelnau, Montpellier Sala E (1996) The role of ®shes in the organization of a Mediterranean sublittoral community. TheÁse Doctorat. Universite d'Aix-Marseille II, Marseille Sala E, Zabala M (1996) Fish predation and the structure of the sea urchin Paracentrotus lividus populations in the NW Mediterranean. Mar Ecol Prog Ser 140: 71±81 Savy S (1987) Les preÂdateurs de Paracentrotus lividus (Echinodermata). In: Boudouresque CF (ed) Colloque International sur Paracentrotus lividus et les oursins comestibles. GIS Posidonie, Marseille, pp 413±423 Schiel DR, Foster MS (1986) The structure of subtidal algal stands in temperate waters. Oceanogr mar Biol A Rev 24: 265±307 Tegner MJ, Dayton PK (1981) Population structure, recruitment and mortality of two sea urchins (Strongylocentrotus franciscanus and S. purpuratus) in kelp forest. Mar Ecol Prog Ser 5: 255±268 Tegner MJ, Levin LA (1983) Spiny lobsters and sea urchins: analysis of a predator±prey interaction. J exp mar Biol Ecol 73: 125±150

539 Verlaque M (1987a) Relations entre Paracentrotus lividus (Lamarck) et le phytobenthos de MeÂditerraneÂe occidentale. In: Boudouresque CF (ed) Colloque international sur Paracentrotus lividus et les oursins comestibles. GIS Posidonie, Marseille, pp 5±36 Verlaque M (1987b) Contribution aÁ l'eÂtude du phytobenthos d'un eÂcosysteÁme photophile thermophile marin en MeÂditerraneÂe oc-

cidentale. TheÁse Doctorat d'EÂtat. Universite d'Aix-Marseille II, Marseille Verlaque M (1996) EÂtude des populations de Paracentrotus lividus de la ReÂserve de Scandola. AnneÂe 1995. Groupe d'InteÂreÃt Scienti®que (GIS), Marseilles (Internal Scienti®c Report to the Parc Naturel ReÂgional de Corse)