Resource Use by a Mediterranean Rocky Slope Fish Assemblage

P.S.Z.N.: Marine Ecology, 18 (1): 5 1 4 6 (1997) 0 1997 Blackwell Wissenschafts-Verlag, Berlin ISSN 0173-9565

Accepted: February 19, 1996

Resource Use by a Mediterranean Rocky Slope Fish Assemblage M. FASOLA', L. CANOVA', F. FOSCHI', 0. NOVELLI' & M. BRESSAN~

' Dipartimento Biologia Animale, Pz. Botta 9, 1-27100 Pavia, Italy. Dipartimento Biologia, Via Trieste 75,1-35121 Padova, Italy.

With 8 figures and 1 table

Key words: Fish, Adriatic Sea, niche, assemblage, food, depth distribution, habitat.

Abstract. Resource partitioning has been intensely studied in fish assemblages of tropical reefs, whereas surprisingly little research has been conducted on the benthic Mediterranean. Here we describe the composition of a fish assemblage of the rocky slopes at Tremiti Islands, southern Adriatic Sea, the vertical distribution at depths from 0 to 42 m, habitat choice and food use of the different species, and we identify the resources along which species are ecologically segregated. Species composition and fish density showed high similarities with the few other assemblages that have been censused on the rocky coasts of the Mediterranean. We observed a wide overdispersion among species in their use of food and habitat, while very little segregation occurred at the level of depth distribution. No coincident use of food, habitat or active periods occurred among fish species, and no trophic guilds exist in the three- dimensional niche.

Problem

Benthic euphotic zones have high primary production and a wide structural hetero- geneity. These factors diversify microhabitats and promote rich animal communities, which are particularly suitable for studies of community structure and resource partitioning. The shallow benthic Mediterranean includes three main biocoenoses: sand flats, seagrass beds and rocks. The latter can be further divided into rocky bottoms with variously sized blocks and into rocky slopes, where the concretions of algae and of other organisms with calcareous skeletons create a high structural complexity .

Fish communities of tropical coral reefs have been intensely studied (review by SALE, 1991), whereas surprisingly little research has been conducted on the benthic Mediterranean, apart from some analyses of the fish communities of seagrass beds in Southern France (BELL & HARMELIN-VIVIEN, 1982; HARMELIN-VIVIEN, 1982; FRANCOUR, 1991; HARMELIN-VIVIEN & FRANCOUR, 1992) and from some comparisons between protected and unprotected zones (BELL, 1983; HARMELIN, 1987; GARCIA-RUBIES & ZABALA, 1990; FRANCOUR, 1994; DLJFOUR et al., 1995). Par- ticularly little studied have been the fish assemblages of the rocky slopes, for which

U. S. Copyright Clearance Center Code Statement: 01 73-9565/97/1801-005 1$14.00/0

52 FASOLA, CANOVA, FOSCHI, NOVELLI & BRESSAN

only few species lists (SARA, 1971; SELOUDRE, 1984) and community analyses (HARMELIN, 1990) exist.

Resource partitioning has been widely studied in fish assemblages (review by Ross, 1986), but again few studies have concerned the Mediterranean. These studies were mainly limited to single families, describing feeding competition among Gobi- idae and Blenniidae in Southern France (ZANDER, 1982; ZANDER & BERG, 1984), the diet of Scorpaenidae in Southern France (KHOURY, 1984; BELL & HARMELIN- VIVIEN, 1983; HARMELIN-VIVIEN et al., 1989) and of Blenniidae in the N. Adriatic (KOPPEL, 1988; KOTRSCHAL et al., 1991), and habitat partitioning by Blenniidae (ILLICH & KOTRSCHAL, 1990), by Sparidae (HARMELIN-VIVIEN et al., 1995) and by Blenniidae, Tripterygiidae and Gobiidae (MACPHERSON, 1994; SYMS, 1995). Studies of food partitioning were often restricted to peculiar biotopes: tidal pools (GIBSON, 1968), a coastal brackish biotope (FERRARI & CHIEREGATO, 1981; FERRARI & ROSSI, 1983). A comprehensive analysis of resource use by the fish assemblage of a Mediterranean rocky slope has not been undertaken.

We studied the fish assemblage of the rocky slopes at Tremiti Islands, southern Adriatic Sea. In this paper we describe the assemblage composition, we analyse the vertical distribution, habitat choice and food use of the different species, and we identify the resources along which species are ecologically segregated.

Material and Methods

1. Study area

Data were collected from 22 to 28 September 1991 and from 7 to 14 September 1992 at two study sites in the small Tremiti archipelago, Southern Adriatic: on the eastern coast of Capraia Island (42"08'10 N - 15"3030 E) and on the southern coast of S. Domino Island (42"06'20 N - 15"28'40 E). Since 1989, the Tremiti have been surrounded by a Marine Nature Reserve within which, however, access and fishing are restricted only for some zones. Our study sites were in the 'partial reserve' zone where access and sport fishing are allowed, but they were accessible only by boat and they were relatively undisturbed. These sites may be considered typical in terms of disturbance and fishing pressure, of most unprotected but relatively inaccessible Mediterranean rocky slopes.

2. Sample collection and analysis

We censused the fish assemblage of the rocky slopes along 32 transects, each running from the bottom shelf to the surface. The maximum depth of transects ranged from 14 to 42 m (average 26 m) depending on the depth of the bottom shelf; their length ranged from 23 to 65 m (average 42 m), and their slope (depth/length) ranged from 0.39 to 1.0 (average 0.64). At each transect, a weighted rope was laid transversally from the surface to the bottom of the slope, either during the evening preceding the count, or around 9 a.m.. Rope placing seemed to have only a momentary influence on the distribution of few fish. From 10 to 1 1 a.m., one SCUBA diver ascended along the rope at an approximate speed of 1 m . min- I and counted all fish (omitting juveniles < 3 cm) within a 3.5 m strip (estimated from the width of the diver's arms stretched at both sides of the rope) and within a 2 m water column from the substrate. Cavities within this strip were explored using lamps. Bekg aware of the difficulties in visual fish censusing (HARMELM-VIVIEN er al., 1985), we believe that the only systematic bias in our counts resulted from overlooking some fish hiding in the cavities. Most fish were not influenced by the diver's presence, at least at the normal observation range, about 3 m. Some species (especially Serranidae) tended to approach, while others (Sparidae, the largest Labridae) tended to avoid the diver, who tried

Fish assemblage of a Mediterranean rocky slope 53 to take into account these biasing behaviours and to keep a correct record of the fish that were originally within the 3.5 x 2 m transect. We counted a total of 3985 individuals.

The three divers had several years' experience in the identification of marine fish during diving. Moreover they had become acquainted with the fish of the study area during previous training in September 1990, that was continued until they were able to identify all the fish they detected, The captured species (see below) coincided with those identified during the censuses (with the obvious exception of some rare species that escaped either capture or detection). This confirms that the visual identifications were correct and reasonably comprehensive. We checked the differences in census performance among the three divers by analysing the results of all the counts, which were performed at the same time and along adjacent transects. No significant difference was recorded in the number of fish countedem-' of transect by different divers (ANOVA test). However, one diver identified slightly more rare species than the other two (ANOVA test on the number of species identified - m-' of transect, FZz8 = 9.84; P = 0.02). For species identification we used the handbooks by SOLJAN (1975) and by F~SCHER ef al. (1987); systematics follows FRED] & MAURIN (1987) and RELINI & PICCINETTI (1991).

We categorized the habitat within 25 cm around the point where each fish was first spotted, as regards type (rock, sand, or water column, i.e. > 2 5 cm from substrate), surface shape (flat, cavity, bump) defined as the dominant shape within the 25 cm radius, and cover (sponges, soft algae, calcareous algae, bare) again defined as dominant cover. Some of the 'cavities' were simply concave-shaped rocks and therefore could have an algal cover. Habitat availability was sampled by recording the same habitat categories at regularly spaced points (every 1 m along the transect under each of the lead weights on

Fig. 1. Morphology and cover of the rocky slopes along the census transects

the transect rope). Figure 1 describes the average characteristics of the rocky slopes along our transects. Bumps and cavities occurred mostly at intermediate depths. The sand flats at maximum depths correspond to the bottom of the slopes, and those at intermediate depths correspond to some flat surfaces among the rocks and to the bottom of those transects whose depth was limited to 20-25 m. The water current along these slopes was constant but slow. Water temperature averaged 23.7"C at 0-10 m, 22.6"C at 11-20 m, 19.3"C at 21-30 m, and 19°C at 3 0 4 2 m.

The divers usually performed concurrent counts along three transects spaced 30 m apart. After each counting session, divers snorkelling at depths fro 0 to 20 m speared and collected fish samples (322 individuals) in the area withm the three transects. In order to reduce the possible influence of spearing on the fish in future transects, the next day's transects were placed > 100 m away from the previous ones. We verified our assumption that spearing did not influence fish presence in subsequent transects by comparing the density and species richness for the 13 transects counted on the first day and in a new area (average density 0.81 fish standard deviation 0.37; average richness 28 species, SD 6.6) with those for the 19 transects conducted on the following days and in the same area (average density 0.93 fish.m-2, SD 0.56, average richness 30 species, SD 14.0). Neither the variances nor the means of the two groups of transects differed significantly (for densities, F = 0.91, NS and t = 0.95, NS; for richness, F = 0.90, NS and t = 0.94, NS). Additional surveys and capture sessions were performed from 22:OO to 01:OO in order to sample nocturnal species. Each fish was immediately injected with formaldehyde 10% solution into the abdominal cavity; later we measured its standard length and removed the digestive tract. The content of the stomach and oesophagus was sorted under a dissection microscope and each food item was identified. The other parts of the digestive tract were not examined, in order to minimize the effects of differential digestion rate of food types (BERG, 1979). For each food item we


measured the three dimensions (precision 0.1 mm); we also calculated its volume, assuming a par- allelepiped shape, in order to express the diet as the volumetric proportion of each food type.

In order to select an appropriate sample size threshold for diet description, we plotted the cumulative

0

8 4 0

2 c

1 5 9 13 17 21 25 29 33 37

no. digestive tracts Fig. 2. Cumulative number of food types in relation to sample size in selected fish species.

number of food types for increasing sample sizes (Fig. 2); for each group of species, grouped on the basis of their diet richness from lowest (2) to highest values (9), we plotted only the species with the largest sample size. The total diet richness was reached with only 2 samples in the most stenophagous species, and with 15 samples in the most euryphagous one (Fig. 2). Three-fourths of the food types, generally the most frequent ones, were found in the first 5 samples. We chose to describe the diet only for species with sample size 3 5 , as a compromise between the need to include as many species as possible and to obtain a complete diet description.

Three resource categories were distinguished: depth, microhabitat and food type. For each category, indices of resource use similarity between all species-pairs were computed using the Proportional Similarity Index: 1-0.5 ZlpXi-pyil (SCHOENER, 1970), where p is the proportional use of resource state i by species x and y; the values of this index range of 0 when no resources are shared, to 1 when identical resources are used. As recommended by MLELLER and ALTENBERC (1985), the significance of the similarity indices ( i e . , whether a similarity value was significantly higher than a value corresponding to random utilization) was tested by a bootstrap procedure: 100 randomized matrices of resource state use were generated for each resource category, using JoERN & LAWLOR'S (1980) randomization algorithm 3 ( i e . , the actually observed frequencies of use, including zero frequencies, were randomly assigned to each fish species). Random similarities were calculated for these 100 matrices, and cumulative distributions were constructed. The observed similarity values that were larger or smaller than the 5% tail values of the cumulative distribution were considered to be of significantly high similarity.

Habitat use was analyzed using IVLEV'S (1961) electivity index, which assumes 0 value when use and availability of the habitat type occur in equal proportions, the value - 1 when the habitat is rejected, and which approaches + 1 when the habitat type is used exclusively.

1. Assemblage composition

We recorded 39 species from 13 families of fish (Table 1). All the abundant species were recorded both during the visual counts and among the captured specimens, but 8 were only observed and 4 other species were only captured. Since it was not possible to estimate densities for the species that were only captured, 4 species are listed in Table 1 without density values; they were all rare.

The assemblage was taxonomically dominated by Labridae (1 0 species), Sparidae

Fish assemblage of a Mediterranean rocky slope 55

Table I . Fish assemblage structure. Average density values were estimated only from transect counts (individuals > 3 cm long, counted within a 3.5-m-wide strip on rock surface and within a 2 m water column), while lengths were measured only from captured specimens. The density given for some families refers to the few fish that were identified only at family level.

density ; standard length

Congridae Conger conger

Gadidae Phycis phycis

Scorpaenidae Scorpaena porcus Scorpaena notata

Serranidae Serranus cabrilla Serranus scriba

Apogonidae Apogon imberbis

Mullidae Mullus surmuletus

Sparidae Diplodus annularis Diplodus sargus Diplodus puntazzo Diplodus vulgaris Boops boops Oblada mulanura Sarpa salpa Spondyliosoma cantharus

Pomacentridae Chromis chromis

Labridae Labrus bimaculatus Lubrus merula Coris julis Symphodus mediterraneus Symphodus tinca Symphodus ocellatus Symphodus rostratus Symphodus roissali Symphodus doderleini Thalassoma pavo

Cobiidae Gobius paganellus Gobius cobitis Gobius geniporus Gobius bucchichi Gobius luteus

Mugilidae Liza aurata Oedalechilus labeo

Lipophrys canevae Parablennius gattorugine Parablennius rouxi Parablennius zvonimiri

Tripterygiidae Tripterygion tripteronotus

Blenniidae

0.03

0.03 0.03 0.18 0. I6

0.84 0.47

0.03

33

34

30-36

34-34

13 11

9-24 8-15

14 16

8

10-18 12-25

0.09

0.03 0.29

0.40 7.12 1.69 1.95

16 12-23

13

14 17 13 15 10 13

13-14

11-22 7-23 8-1 6 9-19 7-1 5

51.74 0.03 0.03 0.08

15.66 1.53 0.63 0.03 0.08 0.09 0.08 0.03 0.1 1

9 7-10

27 14 I I 16

8-19 8-17

10-26

12 8-1 5

10 16 13 6

0.03 0.24 0.13 0.92 0.53 1.69

8-16 66

12 13

9-13

0.05 0.03 0.11 3.45 0.03

13 6

7-16 6-6

1 .oo 5 5-6


(8 species), Gobiidae ( 5 ) and Blenniidae (4), and the other 9 families were represented by only 1-2 species. The total fish density was 91.7 fish. 100 m-2. The assemblage was composed of a few common and many infrequent species. Three dominant species alone constituted 8 1 YO of the individuals (Chromis chromis with frequency 57 %, Coris julis 17 % and Boops boops 8 YO). Seven other species (Oblada melanura, Sarpa salpa, Symphodus mediterraneus, Gobius luteus, Liza aurata, Par- ablennius rouxi, Tripterygion tripteronotus) were relatively common, with frequencies from 1 to 4%, and when added to the three dominant ones, these 10 species accounted for 95 % of the assemblage. Seven species were seen or captured only once: Spondyliosma cantharus, Labrus bimaculatus, Symphodus ocellatus, Gobius paganellus, Gobius cobitis, Lipophrys canevae and Parablennius zvonimiri.

We will restrict the treatment of depth distribution and habitat to the 20 species with densities > 0. I fish * 100 m-’, for which we obtained at least 10 habitat records. Food types will be described for 20 species (those represented by 2 5 digestive tract samples), 17 of which coincide with the species analysed for depth and habitat.

2. Depth distribution

Some preferences for certain depths seem to occur, despite the limited range of depths within our study slopes (Fig. 3). However, when the overlap indices for

86.42 =B 0.6 15 2 8

2 4 6 0

8-10 1- 2Ma sM2

0.6 1.6 26

Lirs aumta 2 s 10

0.1 oa 0.8 1 s 6 7 0.6 1 l d 1

20 w 1m 1Y) 10 50

1 2 5 4 8 6 7

0.1 Od 05 2 6 l o t s 5 7

density (fish. lo0 m4 ) Fig. 3. Fish distribution in relation to depth.

depth distribution were calculated and then subjected to a cluster analysis (UMPGA method), we were unable to find any clearly interpretable cluster of species preferring particular depths. A visual interpretation of the depth distributions in Fig. 3 suggests that many species were distributed mostly within the


upper 20 m (Scorpaena porcus, Scorpaena no ta fa , Serranus scriba, Oblada melanura, Sarpa salpa, Symphodus tinca, Gobius geniporus, Gobius bucchichi, Liza aura ta, Parablennius gattorugine, Parablennius rouxi). Few species used mainly the lower levels, deeper than 20 m (Serranus cabrilla, Gobius luteus). Other species used the entire range of depths, but with higher densities at the intermediate levels; they include the two dominant species of the assemblage (Chromis chromis, Coris julis) and some other species (Diplodus vulgaris, Symphodus mediterraneus, Tripterygion tripteronotus) whose distribution, however, appears irregular, probably due to a small sample size.

Complementary depth distributions were shown by the Serranidae Serranus cabrilla-S. scriba, and by Gobius luteus compared with the other two Gobiidae, but not by any of the Scorpaenidae nor of the Sparidae, Labridae and Blenniidae.

3. Habitat

The frequency with which each species was observed in the 14 habitat types (Fig. 4) differed significantly from habitat availability estimated from the sample points (x2 tests, P c O.Ol), except for Symphodus tinca, Liza aurata and Parablennius gattorugine. For these three species, however, the difference was not significant probably because their data set was small. Therefore, each fish species used a specific micro-habitat.

We calculated habitat overlap indices between each species pair. A cluster analysis of these overlaps (Fig. 5) identifies 6 clusters with overlaps higher than 0.48, the threshold for significant overlap generated by the bootstrap procedure based on 100 random habitat similarities. These significant clusters identify 6 clearly distinct categories of habitat use that can also be discerned in Fig. 4. Five species were clustered together by their frequent use of the water column (those defined as ‘water column specialists’ and ‘mainly water column and algae’ in Fig. 5) , but they are better divided into two categories (3 and 4 in the next list), so that the habitat use categories amount to the following 7.

1) generalists: 6 species used all, or most, habitat categories, especially the rocky habitats and less often water column and sand; these species include the assemblage- dominant Coris julis, plus Symphodus mediterraneus, Symphodus tinca, Para- blennius rouxi, Gobius luteus and Tripterygion tripteronotus. Symphodus tinca seem to use a narrower range of habitats than the other species, but this may be due to the small sample size for this species. A seventh species, Parablennius gattorugine, was joined to the generalists by the clustering algorithm, although it uses only rock bumps and flats covered by algae (Fig. 4) and is better categorised as an ‘exposed rock specialist’ (category 6).

2) species using mainly the water column and cavities: Chromis chromis, another assemblage-dominant species, and Serranus cabrilla were also generalists, but they used mainly the water column and some cavities.

3) species using mainly substrates covered by algae and the water column: Boops boops, Oblada melanura and Diplodus vulgaris used the water column and also the rocks, bumps and flats covered by soft algae. These species fed on benthic or encrusting organisms; therefore, their presence near rock surfaces was relevant albeit less frequent than in the water column.

58 FASOLA, CANOVA, FOSCHI, NOVELLI & BRFSSAN

1

0

-1 1

0

-1 1

0

-1 1

0

-1 1

0

-1

1

0

-1 1

0

-1 1

0

-1 1

0

-1 1

0

-1

I s e m n u s cabrilla I

Diplodus sargus I

k Boopsboops I

I Coris julis I

I , Symphodustinca I

Gobius bucchichi

Liza aurata I

Parablennius rouxi w

Scotpaena notate

Serranus scriba

klhllf3 Diplodus vulgaris

I

Oblada melanum

I Chmnn’s chroms I

I SvmDhodus meditemneus I

Gobius luteus w

I Parablennius g8ttONQhel

Fig. 4. Use of the 14 habitat categories. The values of the electivity index are positive for selection and negative for avoidance; its value is 0 when use is proportional to availability, - 1 for complete rejection, and it approaches + 1 for exclusive use of a given habitat.

Fish assemblage of a Mediterranean rocky slope fl I

I

I r

59

generalists

mainly water column and cavity

water column specialists - mainly water column and algae i

0.1 0.3 0.6 0.7 0.B ovdap

Fig. 5. Dendrogram of habitat overlap (UPGMA method) and characterization of the 6 significant clusters. The dotted line indicates the threshold for significantly similar overlaps.

4) water column specialists: only Liza aurata and Diplodus sargus were exclusively observed within the water column. This specialization is confirmed for Liza by its diet of planktonic organisms (see below the chapter on food).

5 ) rock cavity specialists: Scorpaena porcus, Scorpaena notata and Serranus scriba mainly used rock cavities, less often rock bumps.

6) exposed rock specialists: Sarpa salpa had a sharp preference for rock bumps and flats, mainly those covered by algae but also bare. Parablennius gattorugine also used rock bumps and flats with algal cover (Fig. 4).

7) sand specialists: Gobius geniporus and Gobius bucchichi mainly or exclusively used sand at the bottom of the slopes or on the flats among the rocks; Mullus surmuletus, not shown in Fig. 4 because of the scarce data, is another sand specialist.

4. Food

The initial 40 taxonomic types identified in the digestive tracts were reduced to 10 ecological categories (Fig. 6 ) in order to simplify the comparison across the entire assemblage. ‘Sessile invertebrates’ included Ascidia, Hydrozoa, Serpulida and sponges; ‘mollusks’ were 80% Gastropoda, 20% Bivalvia and a few Poly- placophora; ‘benthic invertebrates’ included mainly Polychaeta; ‘benthic crustaceans’ mainly Isopoda, Decapoda and Pantopoda; ‘planktonic invertebrates’ were mainly Appendicularia; ‘planktonic crustaceans’ Cladocera, Copepoda and Amphipoda; ‘echinoderms’ Ophiura and Echinoidea. ‘Other’ included fish eggs, wood fragments and a few unidentified food items.

The food types of those fish species for which sample size was adequate are shown in Fig. 6 as the proportion of the numerical item frequency and as proportion of the volume. Each fish species strictly consumed a few particular food types. In terms of volume, > 90% of the diet was composed by only one food type for 7 fish species, and by only two types for the other 13 species.

A cluster analysis of the overlaps in food type (Fig. 7) identified 5 small clusters


d

Fig. 6. Diet as the proportion of the numeric frequency of the items (white bars) and of the volume (black bars) of each food type. The sample size (n of digestive tracts) is given after the name of each species.

I r- -- Sernnua s u i h lk~hlcaustacems

I benthiCCP&&XWM

I Scorp.uu n&ll I benthic austaeeans I

mollusk I I , - Iufb I I

I I

"""s benthic invettebrafes I , - %g$g%@foJWhU Mulka aumulalscn IplmMcaustacean

I I LT* ChlwIhChrwnh

I WhP-

sLnphh*

P.Rbknnka~nClUglne

I I -up. I [ w h w m - - ~ i lmollu*

I I I , , D ,

ai as 0.7 I0 Ovzap

Fig. 7. Dendrogram of food type overlap (UPGMA method) and characterization of the 6 significant clusters. The dotted line indicates the threshold for significantly similar overlaps.

of only 2-4 species with overlaps > 0.68, the threshold for significant overlap that was generated by the bootstrap with 100 random food similarities. The remaining 8 fish species overlapped less than the significance threshold and did not form any


cluster. The same stable foods (the most abundant foods accounting for 90% of diet volume) are shared by only a few species, those which form the significant clusters of Fig. 7. Fish plus benthic crustaceans were staple foods for Scorpaena porcus and Serranus scriba; benthic crustaceans for Scorpaena notata, Oblada melanura, Serranus cabrilla and Symphodus tinca; benthic crustaceans plus mollusks for Coris julis and Gobius geniporus; benthic invertebrates plus planktonic crustaceans for Symphodus mediterraneus and Tripterygion tripteronotus; mollusks for Diplodus vulgaris and Gobius bucchichi. All the other species had differing staple foods or combinations of foods: benthic invertebrates plus benthic crustaceans for Mullus surmuletus; echinoderms for Diploduspuntazzo; planktonic crustaceans plus algae for Boops boops; algae plus benthic crustaceans for Sarpa salpa; benthic plus planktonic crustaceans for Chromis chromis; echinoderms plus algae for Symphodus roissali; planktonic crustaceans and invertebrates for Liza aurata; sessile invertebrates for Parablennius gattorugine.

Benthic crustaceans were the most frequent staple food (10 times), while the other categories formed the staple food at approximately equal frequencies: planktonic crustaceans 4 times, algae, mollusks and benthic invertebrates 3 times, echinoderms and fish 2 times, and sessile invertebrates and planktonic invertebrates once.

5. Overlap patterns

The overlaps between all 20 fish species were higher for depth (average overlap within the assemblage = 0.43) than for habitat (0.32), and w r e lowest for food

depth 0 s ,

0.1 0.3 0.5 0.7 0.0

habitat t

0.5 1 O . l h , I 1. I -

0.1 0.5 0.5 0.7 0.0

food 0 . 8 ,

overlap

Fig. 8. Frequency of the overlaps among the 20 fish species with regard to their depth distribution, habitat and food.

(0.19). The frequency distribution of the overlaps (Fig. 8) does not deviate from normality for depth (test of the difference from normal distribution, x2 tests, NS), while it is skewed towards low values for habitat (x’ = 18.1, P c 0.005) and highly skewed towards values lower than 0.1 for food (x’ = 68.5, P < 0.0001).

The species grouped in the same cluster of habitat use (Fig. 5 ) were segregated into different clusters of food type (Fig. 7), and no fish species coincided in their use of both food and habitat.

Discussion

1. Similarities among Mediterranean rocky coast assemblages

The fish assemblage in our study area showed high similarities with the few others that have been censused on the rocky coasts of the Mediterranean, mainly on


boulder bottoms of Southern France (BELL, 1983; HARMELIN & BELLAN-SANTINI, 1985; HARMELIN, 1987,1990; FRANCOUR, 1991). In all these assemblages the dominant families were Labridae, Sparidae, Gobiidae and Blenniidae. The species found elsewhere coincide from two-thirds to three-fourths with those in our area. The richness in our area (39 species) falls within the narrow range, from 35 to 45 species, found in the studies cited above. All these assemblages were composed of many scarce and by a few dominant species, the latter always including Chromis chromis and Coris julis. The densities of each dominant species in our area were very similar to the densities of the same species in rocky areas of Southern France (HARMELIN, 1987). The total density of the entire assemblage (92 fish. 100 m-2) was comparable to those found in other areas with rocky slopes: 163 and 220 fish. 100 m-2 (HAR- MELIN, 1987), 66 and 134 fish. 100 m-2 (HARMELIN, 1990), 48, 54, 103 and 116 fish. 100 md2 (BELL, 1983), 151,157 and 195 fish. 100 m-2 (HARMELIN & BELLAN- SANTINI, 1985). All the above parameters therefore indicate a high similarity among the fish communities of the rocky slopes, at least for the northern-central Medi- terranean.

2. Resource use

The generality of the habitat preferences we observed cannot be evaluated because comparable information is lacking, although the preferences of some species were similar to those described by MACPHERSON (1994). In the case of food, the types preferred by fish in other Mediterranean rocky coasts show some similarities with our results. Both in our sample and in tidal pools (GIBSON, 1968), the most abundant food items were crustaceans, followed by algae, mollusks and benthic invertebrates. Specific preferences for certain food types seem to be maintained across the Med- iterranean by some species, e.g., for fish by Scorpaena porcus, for crustaceans by Scorpaena notata and Serranus cabrilla (BELL & HARMELIN-VIVIEN, 1983, and results from this study). On the other hand, differences in food preference among areas exist for Parablennius gattorugine, Tripterygion tripteronotus, Gobius bucchichi (GIBSON, 1968, this study), and for Diplodus vulgaris, Symphodus mediterraneus and Syrnphodus tinca (BELL & HARMELIN-VIVIEN, 1983; this study). Food preferences therefore, seem to have a certain degree of generality, albeit accompanied by some flexibility.

The factors determining assemblage structure and resource partitioning among coexisting animal species remain unclear, and the suggested role of competition has generated much field research and controversial interpretations. It is now generally accepted that competition greatly affects niche partitioning within animal communities (reviews by CONNELL, 1983; SCHOENER, 1983; GUREVITCH et al., 1992), but abiotic factors can also be effective (DUNSON & TRAVIS, 1991). Fish communities in reef environments are structured by several interacting processes, among which physical disturbances, recruitment limitations, resource availability of space, shelter and food, interspecific competition and predation have been hypothesized (JONES, 1991; HIXON, 1991; SALE, 1991). The factors structuring coastal fish communities are still being actively investigated, and recent field studies and experiments of assemblage manipulations have concluded that competition is not important in some cases (DELBEEK & WILLIAMS, 1987; NORTON, 1991), while


in other cases both competitive and predatory effects were found (ROBERTSON & GAINES, 1986; TONN et al., 1986; BRABAND & FAAFENG, 1993).

Competitive effects within an assemblage will produce resource partitioning or ‘niche diversification’, i.e., an overdispersion of resource use with consistently low overlap among species. Resource partitioning in marine fish communities was reviewed by Ross (1986). A general pattern is that trophic separation is more frequent than habitat separation, temporal separation being the least common; in most habitats fish assemblages show approximately equal separation among spatial and trophic dimensions, but temperate marine reef assemblages are exceptional in showing greater importance of trophic separation. This pattern contrasts with resource partitioning in terrestrial systems, due mainly to habitat and less to food (SCHOENER, 1974). In one of the very few studies on resource partitioning of Mediterranean fish, KOPPEL (1988) found that Blennius canevae and B. incognitus compete for space, and that competition is reduced by differing activity rhythms and by differences in territoriality. Another study on substratum utilization by Mediterranean Blenniidae, Tripterygiidae and Gobiidae (MACPHERSON, 1994) suggested that species displayed clear preferences for specific substratum types, but substratum occupancy has no adverse effect on the other species.

In our fish assemblage, we observed a wide overdispersion in the use of food and habitat; the species overlaps along these resources were highly skewed towards low values (Fig. 8), therefore the species were more dispersed than random along these resource axes. Segregation was wider for food, for which the interspecific overlap was least. On the other hand, very little segregatirn occurred at the level of macrohabitat (depth distribution), for which the overlaps were random (their distribution did not deviate from normality). HARMELIN (1990) also found that fish assemblages on shallow (1&15) and deep (2547) rocky bottoms were similar. The diurnal or nocturnal activity rhythms added another dimension to the segregation among species. Most species in our assemblage are diurnal, but the Scorpaenidae are active both day and night and the Congridae, Gadidae and Apogonidae are active only at night (KI-IOURY, 1984; HARMELIN, 1987). These activity rhythms were confirmed by our observation of different species during diurnal censuses and night surveys; moreover, diurnal species had full digestive tracts only when captured by day, the reverse being true for nocturnal species.

These overdispersion patterns are broadly consistent with the hypothesis of competitively induced niche diversification. However, the fish assemblage we studied was taxonomically diverse, and our results cannot distinguish the effects of phylo- genetic constraints and of competition on niche segregation. Moreover, resource segregation of the type observed in our study is only weak evidence for competitive effects. Niche shifts and complementary overlap between niche dimensions would be stronger evidence, but were not found in our results. Conclusive evidence of competition can be provided only by experimentally induced niche shifts and density changes, attributable to the manipulation of competitor densities.

No fish species coincided in their staple food, habitat category and activity period. This overdispersion of resource use splits the assemblage into segregate species, so that no foraging guilds (sensu ROOT, 1967) can be identified. The resource use groups that have been identified among marine fish (spatial categories, HARMELIN, 1987, or foraging guilds, EBELING & HIXON, 1991) are broad categories within which further segregation occurs in the multidimensional niche space.

64 FASOLA, CANOVA, FOSCHI, NOVELLI & BRE.SSAN

Summary

On the rocky slopes of Tremiti Islands, S. Adriatic, at depths from 0 to 42 m, the fish assemblage was composed of 39 species from 13 families, mainly Labridae, Sparidae, Gobiidae and Blenniidae. Total density averaged 92 fish- 100 m-*, with three species alone constituting 81 YO of the individuals. Species were assorted into 7 habitat use types: generalists, species mainly using the water column and cavities, species mainly using the water column and algae, water column specialists, rock cavity specialists, exposed rock specialists and sand specialists. Each species consumed a small number of food types. Benthic crustaceans were the most frequent staple food, while planktonic crustaceans, algae, mollusks, echinoderms, other benthic invertebrates, fish, sessile invertebrates and planktonic invertebrates served as staple food at equal frequencies. Fish species were ecologically segregated mainly through different diets (average assemblage overlap = 0.19) and through microhabitat types (overlap = 0.32), whereas depth distributions showed higher overlap (0.43). Additional segregation was due to diurnal or nocturnal activity. No fish species coincided in their staple food, habitat category and activity period. This overdispersion of resource use splits the assemblage into segregate species, so that no foraging guilds can be identified.

Acknowledgements

We thank MARES@ for providing the excellent SCUBA equipment we used during our research, as well as M. L. HARMELIN-VIVIEN, A. CRIVELLI and B. BRITTON for their advice.

References

BELL, J. D., 1983: Effects of depth and marine reserve fishing restrictions on the structure of a rocky reef fish assemblage in the North-Westem Mediterranean Sea. J. Appl. Ecol., 2 0 357-369.

-- & M. L. HARMELIN-VIVIEN, 1982: Fish fauna of French Mediterranean Posidoniu oceunica seagrass meadows. 1. Community structure. Tethys, 10 337-347.

-- & M. L. HARMELM-VIVIEN, 1983: Fish fauna of French Mediterranean Posidonia oceanica seagrass meadows. 2. Feeding habits. Tethys, 11: 1-14,

BERG, J., 1979: Discussion of methods of investigating the food of fishes, with reference to a preliminary study of prey of Gobiusculusflauescens (Gobiidae). Mar. Biol., S O 253-273.

BRABAND, A. & B. FAAFENG, 1993: Habitat shift in roach (Rurilus rurilus) induced by pikeperch (Slizosredion lucioperca) introduction - Predation risk versus pelagic behaviour. Oecologia, 9 5 38-46.

CONNELL, J. H. 1983: On the prevalence and relative importance of interspecific competition: evidence from field experiments. Am. Nat., 122 661-696.

DELBEEK, J. C. & D. D. WILLIAMS, 1987: Food resource partitioning between sympatric populations of brackish water sticklebacks. J. Anim. Ecol. 5 6 949-967.

DUFOUR, V., J.-Y. JOUVENEL & R. GALZIN, 1995: Study of a Mediterranean reef fish assemblage. Comparisons of population distributions between depths in protected and unprotected areas over one decade. Aquat. Living Resour., 8 17-25.

DUNSON. W. A. & J. TRAVIS. 1991: The role of abiotic factors in community organization. Am. Nat., - - 138 1067-1 09 I .

EBELING, A. W. & M. A. HIXON, 1991: Tropical and temperate reef fishes: comparison of community structures. In: P. F. SALE (Ed.), The ecology of fishes on coral reefs. Academic Press, San Diego: -. 509-563.

FERRARI, I. & A. R. CHIEREGATO, 1981: Feeding habitats of juvenile stages of Spurus aurafus L.,

Fish assemblage of a Mediterranean rocky slope 65 Dicentrarchus labrax L. and Mugilidae in a brackish embayment of the Po river delta. Aquaculture,

-- & R. ROSSI, 1983: Regime alimentare di Atherina boyeri RLSSO in una laguna del delta del Po. Nova Thalassia, 6 275-280.

FISCHER, W., M. SCHNEIDER & M.-L. BAUCHOT (Eds.), 1987: Fiches F A 0 &identification des e s e s pour les besoins de la pkhe. Mediterranee et Mer Noire. FAO, Roma; 1529 pp.

FRANCOUR, P., 1991: The effect of protection level on a coastal fish community at Scandola, Corsica. Terre Vie, 46: 65-8 1.

--, 1994: Pluriannual analysis of the reserve effect on ichthyofauna in the Scandola natural reserve (Corsica, Northwestern Mediterranean). Oceanol. Acta, 17: 309-3 17.

FRED], G. & C. MAIJRM, 1987: Les poissons dans la banque de donntes “Mtdifaune”, application a I’ttude des characttristiques de la faune ichthyologique mtditerrankenne. Cybium, 11: 218-341.

GARCIA-RUBIES, A. & M. ZABALA, 1990: Effect of total fishing prohibition on the rocky fish assemblages of Medes Island marine reserve. Sci. Mar., 54: 317-328.

GIBSON, R. M., 1968: The food and feeding relationship of litoral fish in the Banyuls region. Vie Milieu, 19 447-456.

GUREVITCH, J., L. MORROW, A. WALLACE & J. S. WALSH, 1992: A meta-analysis of competition in field experiments. Am. Nat., 140: 539-572.

HARMELIN, J. G., 1987: Structure et variabilitt de I’ichtyofaune d’une zone rocheuse prottgke en Mtditerranke (Parc national de Port-Cros, France). P.S.Z.N.1: Marine Ecoloyg, 8 263-284.

--, 1990: Ichtyofaune des fonds rocheux de Muiterrante: structure de peuplement du coralligtne de I’ile de Port-Cros (Pam national, France). Mesogte, 5 0 23-30.

-- et D. BELLAN-SANTINI, 1985: Modtles naturek pour les rtcifs artificiek en Muiterrante. toll. fr.- japon. Odanographie, Marseille, 6: 85-92.

HARMELM-VIVIEN, M., 1982: Ichtyofaune des herbiers de posidonies du Parc National de Port-Cros: I. Composition et variations spatio-temporelles. Trav. Sci. Parc. Nat. Port-Cros, 8: 69-92.

-- & P. FRANCOUR, 1992: Trawling or visual censuses? Methodological bias in the assessment of fish populations in seagrass beds. P.S.Z.N.1: Marine Ecology, 13: 41-51.

--, J. G. HARMELIN, C. CHAUVET, C. DUVAL, R. GALZIN, P. LEJEUNE, G. BARNABB, F. BLANC, R. CHEVALIER, J. DUCLERC & G. LASSERRE, 1985: Evaluation visuelle des peuplements et populations de poissons: mtthodes et problemes. Terre Vie, 4 0 467439.

--, J. G. HARMELM & V. LEBOULLEUX, 1995: Microhabitat requirements for settlement of juvenile sparid fishes on Mediterranean rocky shores. Hydrobiologia, 24: 309-320.

-- R. A. KAIM-MALKA, M. LEWYER & S. S. JACOB-ABRAHAM, 1989: Food partitioning among scor- paenid fishes in Mediterranean seagrass beds. J. Fish Biol., 34: 715-734.

HIXON, M. A,, 1991: Predation as a process structuring coral reef communities. In: P. F. SALE (Ed.), The ecology of fishes on coral reefs. Academic Press, San Diego: 475-508.

ILLICH, I. P. & K. KOTRSCHAL, 1990: Depth distribution and abundance of Northern Adriatic littoral rocky reef Blennioid fishes (Blenniidae and Tryprerigion). P.S.Z.N.1: Marine Ecology, 11: 277-289.

IVLEV, V. S., 1961: Experimental ecology of the feeding of fishes. Yale Univ. Press, New Haven; 302 PP.

JOERN, A. & L. R. LAWLOR, 1980: Food and microhabitat utilization by grasshoppers from arid grasslands: comparisons with neutral models. Ecology, 61: 591-599.

JONES, G. P., 1991: Postrecruitment processes in the ecology of coral reef fish populations: a multi- factorial perspective. In: P. F. SALE (Ed.), The ecology of fishes on coral reefs. Academic Press, San Diego: 294328.

KHOURY, C., 1984: Ethologies alimentaires de quelques espbces de poissons de l’herbier de posidonies du Parc National de Port-Cros. In: C. F. BOUDOURESQUE, A. JEUDY DE GRISSAC & J. OLIVER (Eds.), International Worshop Posidonia oceanica beds. G.I.S. Posidonie, Marseille: 335-347.

KOPPEL, V. H., 1988: Habitat selection and space partitioning among two Mediterranean Blenniid species. P.S.Z.N.1: Marine Ecology, 9: 329-349.

KOTRSCHAL, K., D. G. LINDQUIST & I. P. ILLICH, 1991: Opportunistic feeding in Mediterranean Blennioid fishes (Blenniidae and Tripterygion) as revealed by experimental food provisioning in the field. P.S.Z.N.1: Marine Ecology, 1 2 63-73.

MACPHERSON, E., 1994 Substrate utilisation in a Mediterranean littoral fish community. Mar. Ecol. Prog. Ser., 1 1 4 21 1-218.

MUELLER, L. D. & L. ALTENBERG, 1985: Statistical inference on measures of niche overlap. Ecology, 66: 1204-1210.

NORTON, S. F., 1991: Habitat use and community structure in an assemblage of Cottid fishes. Ecology, 7 2 2181-2192.

RELINI, G. & C. PICCINETTI, 1991: Stato attuale die censimenti ittici nei mari italiani. Ricerche Biologia

ROBERTSON, D. R. & S. 0. GAINES, 1986: Interference competition structures habitat use in a local

2 5 243-257.

Selvaggina, Supplemento, 1 6 29-54.

assemblage of coral reef surgeonfishes. Ecology, 67: 1372-1383.


ROOT, R. B., 1967: The niche exploitation pattern of the blue-gray gnatcatcher. Ecol. Monogr., 37:

Ross, S. T., 1986 Resource partitioning in fish assemblages: review of field studies. Copeia, 1986 352-

SALE, P. F., 1991: Reef fish communities: open nonequilibrial systems. In: P. F. SALE (Ed.), The ecology

SAG, M., 1971: Le peuplement du coralligtne des Pouilles. Rapp. Comm. Int. Mer Mtditerrante 2 0

SCHOENER, T. W., 1970: Non-synchronous spatial overlap of lizards in patchy habitats. Ecology, 51:

-- 1974 Resource partitioning in ecological communities. Science, 185: 27-39. -- 1983: Field experiments on interspecific competition. Am. Nat., 122 240-285. SELOUDRE, P., 1984: Contribution A I’btude de I’ichtyofaune de la Reserve Marine de Banyuls-Cerbtre

par comptage en scaphandre autonome. Mtm. Univ. Sci. techn. Languedoc; 32 pp. SOWAN, T., 1975: I pesci dell’Adriatico. Mondadori, Milano; 522 pp. SYMS, C., 1995: Multi-scale analysis of habitat association in a guild of blennioid fishes. Mar. Ecol.

Prog. Ser., 125 3 1-43. Tom, W. M., C. A. PASZKOWSKI t T. C. MOERMOND, 1986: Competition in Umbra-Perca fish

assemblages: experimental and field evidence. Oecologia, 1986: 126133. ZANDER, C. D., 1982: Feeding ecology of Littoral gobiid and blennioid fishes of the Banyuls area

(Mediterranean Sea). 1 - Main food and trophic dimension of niche and ecotope. Vie Milieu, 3 2 1-10.

-- & J. BERG, 1984: Feeding ecology of littoral gobiid and blennioid fishes of the Banyuls area (Mediterranean Sea). 2 -Prey election and size preference. Vie Milieu, 34: 149-157.

317-350.

388.

of fishes on coral reefs. Academic Press, San Diego: 564-598.

235-237.

408-418.

Resource Use by a Mediterranean Rocky Slope Fish Assemblage

Documents