Marine Biology

, Volume 149, Issue 5, pp 1059–1070

Population structure of an exploited benthic cnidarian: the case study of red coral (Corallium rubrum L.)

Authors

    • Alfred Wegener Institute for Polar and Marine Research
  • Sergio Rossi
    • Institut de Ciències del Mar (CSIC)
  • Josep-Maria Gili
    • Institut de Ciències del Mar (CSIC)
  • Wolf Arntz
    • Alfred Wegener Institute for Polar and Marine Research
Research Article

DOI: 10.1007/s00227-006-0302-8

Cite this article as:
Tsounis, G., Rossi, S., Gili, J. et al. Mar Biol (2006) 149: 1059. doi:10.1007/s00227-006-0302-8

Abstract

Octocorals are an important part of many ecosystems as they add three-dimensional complexity to the benthos and thereby increase biodiversity. The Mediterranean red coral (Corallium rubrum, L. 1758) is a longevous octocoral that is harvested commercially, yet natural and anthropogenic influences on its population size structure are little understood. This study found that some harvested red coral populations had a significantly different size structure when compared to populations at the nearby Marine Protected Area (MPA) of Medas Islands at the Spanish Costa Brava (NW Mediterranean). Eighty-nine percent of the red corals in the harvested Costa Brava area are less than 10 years old and 96% of all colonies have not yet grown more than second-order branches. The size/age distribution of the harvested population is notably skewed towards younger and smaller colonies. Thus, although red coral is still abundant, its population structure is strongly distorted by harvesting. The results confirm that MPAs are useful to distinguish between anthropogenic and natural influences on population structure. However, 14 years of protection appears to be an insufficient recovery time for a longevous octocoral population such as red coral.

Introduction

Gorgonians are characteristic and conspicuous members of many benthic communities in tropical, temperate and polar ecosystems (e.g., Kinzie 1973; Coma et al. 1995; Arntz et al. 1999). They play an important role in marine ecosystems as they add three-dimensional complexity to the habitat and consequently increase biodiversity (Dayton et al. 1974; Jones et al. 1994; Thrush and Dayton 2002).

Previous demographic studies have been a valuable tool to determine the state of octocoral populations (e.g., Grigg 1977; Weinbauer and Velimirov 1996; García-Rodríguez and Massó 1986a; Bak and Meesters 1998). As size/age distribution reflects the combined effects of mortality, recruitment and growth, it records the life history of a population, especially in the case of sessile longevous species such as red coral (Santangelo et al. 1993a). The size/age structure of a healthy population in steady-state recruitment is structured by a monotonic curve with a negative exponent. A severe lack of old individuals is an indicator of high mortality, either due to natural causes (Grange and Singleton 1988) or due to harvesting (Grigg 1976). These factors manifest themselves as a shift of the size/age distribution towards young colonies, deviating from a steady-state distribution, as commonly observed in harvested species (Santangelo et al. 2004).

The Mediterranean red coral (Corallium rubrum L. 1758) (Gorgonacea, Octocorallia) is a sessile cnidarian whose polyps form arborescent colonies, which can reach a height of 50 cm (Garrabou and Harmelin 2002). Red coral is a slow growing species (García-Rodríguez and Massó 1986a; Santangelo et al. 1993a; Garrabou and Harmelin 2002; Bramanti 2005; Marschal et al. 2004), with a life span of up to 100 years (Riedl 1983). However, red coral populations all over the Mediterranean are dominated by young, small colonies (García-Rodríguez and Massó 1986a; Abbiati et al. 1991, 1992; Cattaneo-Vietti et al. 1993; Santangelo et al. 2004). Red coral has a long history of exploitation (Tescione 1973), and during the last decades it has become evident that it is now an overexploited natural resource (FAO 1984, 1988; Santangelo and Abbiati 2001; Tsounis 2005). In order to examine the effect of harvesting, this study compared the size structure of several populations at the Costa Brava, including one in a protected marine park in the same area. A problematic aspect in such comparisons is that red coral populations in other parts of the Mediterranean have been shown to differ in structure (Cattaneo-Vietti et al. 1993). However, by comparing nearby populations, it can be expected that environmental differences are kept small, resulting in minor interhabitat differences of population structure (Desbryéres et al. 1973; Grémare et al. 1998). We hypothesize that differences in colony diameter and height between a protected population and harvested populations must be significantly higher than differences among various harvested populations. The analyzed factors are expected to show a homogeneous size structure due to harvesting pressure, as inter-habitat environmental and biological differences should be masked by the considerable fishery pressure.

Material and methods

Study site

The study was carried out at the Spanish Costa Brava where six populations (defined as a distinct group of colonies after Lincoln et al. 1998) in six different sites, separated by at least 1 km (Fig. 1), were surveyed between January 2002 and September 2003. Each site was sampled at haphazardly positioned stations (Table 1) within the coralligenous hard-bottom sublittoral (Sará 1969; Gili and Ros 1984), with SCUBA diving. At all sites except the Marine Protected Area (MPA) of the Medas Islands (Fig. 1) the populations are subject to commercial exploitation, whereas the three sites at Cap de Creus are within the main harvesting zone.
https://static-content.springer.com/image/art%3A10.1007%2Fs00227-006-0302-8/MediaObjects/227_2006_302_Fig1_HTML.gif
Fig. 1

Map of the study site at the Costa Brava (NE Spain), indicating the Marine Protected Area of Medas Islands

Table 1

Number of sampling stations, depth ranges and number of colonies examined at each of the six sites, as well as the number of replicates used for the statistical tests

Sites

Coordinates of the center of the sampling areaa

No. of sampling stations

Depth range (m)

No. of sampling dives

No. of measured colonies

No. of colonies comparedb

Northern Cap de Creus

42°20′1″N, 3°16′30″ E

8

15–49

14

1,677

695/78

Eastern Cap de Creus

42°16′5″N, 3°16′50″E

12

7–40

27

1,722

695/78

Southern Cap de Creus

42°13′45″N, 3°13′1″E

5

11–45

17

1,632

695/78

Medas Islands

43°2′30″N, 13′30″E

7

15–48

13

819

695/78

Coast of Montgri

42°5′56″N, 3°12′51″E

4

13–32

7

695

695/78

Coast of Begur

42°1′N, 3°14′55″E

4

18–32

9

1,051

695/78

aSampling stations were distributed evenly across coralligenous habitats of the sampling site

bWhile a high number of colonies were counted in order to best represent the demographic structure, subsampling for the statistical analysis was necessary in order to balance the ANCOVA/ANOVA models. The ANOVA model analyzed a depth range of 32–40 m depth, thus resulting in 78 data points (replicates) per population

Colony abundance and distribution

Coral abundance was estimated at three sites: the Coast of Begur and Montgri, as well as the Medas Islands (all of them are outside the main harvesting area). This was done by laying a measuring tape along the selected isobath, using it as a transect line. A total of 20 transects at the three sites (4–6 per site), laid between depths of 20–50 m, were used. The length of each transect was 30–100 m, depending on the dive time and/or extension of the coralligenous hard substrate. A zone of 2 m on each side of the transect line was surveyed by SCUBA diving, resulting in 200 m2 per 50 m of transect length. All coral patches within the survey site were counted and measured. Each red coral patch usually consisted of several densely growing colonies that inhabited a suitable microhabitat like a crevice or an overhanging wall (True 1970; Gili and Ros 1984; Riedl 1983).

Colony abundance and patchiness varied across a wide range (shallow patches, i.e., <30 m depth, and deep ones, i.e., >30 m depth, typically differed markedly), requiring two different methods to estimate colony abundance. In shallow water (i.e., <30 m depth) red coral competes with fast growing organisms (e.g., algae, bryozoans, hydrozoans) and grows on semi-dark overhangs or in crevices forming small, dispersed patches with high colony abundances, thus requiring a 20 cm × 20 cm quadrate to count colonies efficiently. This size has successfully been used in other studies (Weinberg 1979; Santangelo et al. 1993a), and in preliminary tests for this study it was found to be a good balance between an efficient working pace and minimum error (see also: Benedetti-Cecchi et al. 1996). The number of 20 cm × 20 cm quadrates counted per transect area depended on the size of the patches encountered and on time constraints. The quadrates were haphazardly positioned within the patches.

At some sites (typically >30 m depth) colony abundance was much lower, while patches extended over huge banks. Those populations could not be sampled with a 20 cm × 20 cm quadrate; thus, a 1 m × 1 m quadrate was used, which was positioned in 1 m increments on the transect line, which was placed across the largest possible distance through the bank. This dual approach was necessary to estimate the coral abundance with optimal precision and efficiency in either case of distribution pattern. The total area surveyed was 2,660 m2. A regression analysis of colony abundance and total patch area per survey area against the independent factor depth was done.

Biometric comparison of the six populations

The biometry of red coral was studied by photographic sampling. This method allowed us to obtain an extensive set of data in a non-destructive way. The highest possible number of colonies were photographed by haphazardly sampling each patch. As the photo sampling was not used to quantify colony abundance, and in order to obtain a maximum number of photographs from each patch, the photo sampling was not restricted to the survey area along the survey line, when the patch dimension exceeded it. This allowed us to measure a total of 7,600 colonies at the Costa Brava (Table 1). A 3.1 megapixel digital camera (Sony Cybershot DSC S75) was used in a waterproof plastic housing (Mangrove 2000, Aditech) with illumination provided by a 12 V 35 W divers light. For each photo a small ruler was placed for size reference next to the coral, in a way that it was oriented parallel to its longitudinal axis. Usually one colony was counted per photograph, but sometimes several colonies were in the same relative axis with the ruler, allowing us to measure 1–4 colonies per photograph.

The photos were processed with Adobe Photoshop™ 5.5 software. In each photo the scale was calibrated using the ruler as a size reference, which allowed a precision of 1 mm. This way the basal diameter and height of coral colonies could be estimated in each photo. A colony was defined by its stem, and the diameter of a colony as the basal diameter of the stem 5 mm above the substrate. Colony height was defined as the maximum distance between the base of the stem and the tips of the farthest branches. Although this method provides only limited precision, it was found to be suitable for the scope of this study (see below). The branches were counted after they were classified into a branching order (1st, 2nd, 3rd and 4th order) following the classification system used to describe branching patterns in gorgonians (Brazeau and Lasker 1988; Mitchell et al. 1993; Coma et al. 1995), as this system allows a comparison of the quantity of newly grown apical branches.

The broad range of depths sampled in order to obtain a representative view of the population structures called for an ANCOVA approach, testing the effect of both factors, “population” (one population per site, N=6) and “depth” on either colony diameter or height. This initial ANCOVA approach found an interaction of the factors “populations” and “depth” (see below), so that no post hoc test could be applied to see whether the variance was higher between the protected population and the five harvested ones, or mainly among the five harvested ones.

Thus to test the hypothesis that the population at the Medas Islands (MPA) is structured differently, we applied balanced one-way ANOVA models testing for effects on the factor “population” (6 levels) on either colony diameter or height. This ANOVA was restricted to data from a narrow depth range of 32–40 m (where the majority of coral fishing takes place in the study area, see Tsounis 2005). This eliminated variance due to depth, and allowed the use of a Tukey post hoc test for the factor “population”. The ANOVA model was balanced by randomly subsampling (using a random number generator) 78 colonies from 32 to 40 m depth, and the before mentioned ANCOVA model was balanced by randomly subsampling 695 colonies at each site from all depths (Table 1). These numbers were set by the smallest sample size per site.

The colony diameter data showed comparable variances (O’Brien’s test), but showed normal distribution (Shapiro–Wilk test, P=0.1), only after logarithmic transformation. As the variances of the colony height data were not homogenous (O’Brien’s test), they were transformed using the following equation:
$$ {X}\ifmmode{'}\else$'$\fi = \log 10{\left( {X + \frac{3} {8}} \right)} $$
(Zar 1996).

After transformation the data met the assumptions for parametric testing as described above and the data were tested in the same way as the diameter data.

In addition to the ANCOVA and the Tukey test, a Kolmogorov–Smirnov test was applied to compare the colony diameter and height frequency distributions of the six populations in order to test the shapes of the curves for differences. The relative quantity of branches of various orders was plotted in order to evaluate the developmental stages of the colonies.

Results

Colony abundance and distribution

The regression analysis showed a moderate relationship between depth and colony abundance within patches (R2=0.525; P<0.05) (Fig. 2). Total patch area per transect area showed a weaker relationship with depth (y=0.0003e0.0588x+0.0047, where y = colony abundance (colonies m−2) and x = depth (m); R2 = 0.1467). The average colony abundance within patches was 127.1±118.6 colonies m−2 (mean ± SD), the patch size was 0.428±0.741 m2, and the patch abundance was 0.063±0.05 patches m−2. Total red coral abundance (m−2) on coralligenous hard substrate (20–50 m depth) was estimated at 3.42±4.39 colonies m−2. Patch size did not significantly differ between populations and depths.
https://static-content.springer.com/image/art%3A10.1007%2Fs00227-006-0302-8/MediaObjects/227_2006_302_Fig2_HTML.gif
Fig. 2

Red coral colony abundance (within patches) as a function of depth. N=20 Transects (426 quadrates), at 14 Stations (4–6 per site) in the three sites: Medas Islands, Coast of Mongri and Begur

Biometric comparison of the six populations

Colony diameter and height differed significantly among populations and also among depths (Table 2). Depth and population interactively affected colony diameter and height (Table 2).
Table 2

ANCOVA of colony diameter and height among populations and depths (depth range: 7–48 m)

Source of variation

Colony diameter

Colony height

df

SS

F

P

df

SS

F

P

Population

5

61.6

82.1

0.001

5

9.93

83.4

0.001

Depth

1

2.72

18.1

0.001

1

0.43

18.0

0.001

Population × depth

5

3.95

5.30

0.001

5

0.63

5.21

0.001

The six populations examined were: Cap de Creus North; East; South; Medas Islands; Coast of Montgri; Coast of Begur. Six hundred and ninety-five replicate colonies were used per population (so the total N=4170)

One factorial ANOVA in the depth range with the heaviest harvesting pressure (32–40 m, see Tsounis 2005) revealed significant differences in colony diameter and height between the populations (Table 3). A subsequent Tukey test showed that colony diameter and height are significantly larger at the Medas Islands than at all other populations, although there were also significant differences in colony diameter between the other populations (Table 4). However, the difference in mean diameter between the Medas Islands (6.9±2.4 mm) and the other populations is very large (Table 4).
Table 3

ANOVA of colony diameters and height among six red coral populations at the Costa Brava, situated at 32–40 m depth

Source of variation

Colony diameter

Colony height

df

SS

F

P

df

SS

F

P

Population

5

8.01

12.6

0.001

5

2.86

9.64

0.001

The smallest sample size at 32–40 m set the number of 78 replicate colonies used for each of the six levels of the factor “population” (so the total N=468)

Table 4

Tukey test results (α=0.05) (following the ANOVA) comparing populations for differences in colony diameter and height

Level

Colony diameter

Colony height

Groupsa

Least square mean

Basal diameter (mm) (mean ± SD)

Groupsa

Least square mean

Colony height (mm) (mean ± SD)

Medas Islands

A

  

1.72

6.9±2.4

A

 

1.56

41.9±25.6

Eastern CC

 

B

 

1.50

4.6±1.7

 

B

1.40

32.1±16.9

Northern CC

 

B

C

1.48

5.7±2.4

 

B

1.34

36.6±20.3

Coast of Montgri

 

B

C

1.39

4.9±2.1

 

B

1.34

25.5±14.2

Southern CC

 

B

C

1.36

5.3±2.1

 

B

1.34

26.8±15.2

Coast of Begur

  

C

1.33

4.9±1.5

 

B

1.32

29.5±16.5

The smallest sample size at 32–40 m set the number of 78 replicate colonies used for each of the six levels of the factor “population” (so the total N=468)

aGroups marked with different letters are significantly different from each other

The frequency distributions of the basal diameter and colony height were similar among the three populations at Cap de Creus and the two coastal populations at Montgri and Begur, all showing a positively skewed curve (Figs. 3, 4). The Medas Islands population showed a strikingly different frequency distribution for basal diameter, being bell shaped, with an elevated number of large colonies (Fig. 3).
https://static-content.springer.com/image/art%3A10.1007%2Fs00227-006-0302-8/MediaObjects/227_2006_302_Fig3_HTML.gif
Fig. 3

Frequency distribution of the basal diameter of red coral colonies at the Costa Brava (NE Spain)

https://static-content.springer.com/image/art%3A10.1007%2Fs00227-006-0302-8/MediaObjects/227_2006_302_Fig4_HTML.gif
Fig. 4

Frequency distribution of the colony height of red coral colonies at the Costa Brava (NE Spain)

A comparison of basal diameter frequency distribution, using the Kolmogorov–Smirnov test revealed that the distribution off the Medas Islands was significantly different (α=0.05) from that of all the other populations, except for the one off eastern Cap de Creus, while all the other populations did not differ significantly from each other. Similarly, the distribution of colony height was different among all of the examined populations except for those between at the Medas Islands and the opposing coast of Montgri (α=0.05).

The distribution of branch orders showed a lack of characteristic arborescent morphology all along the Costa Brava (including the MPA), as most of the colonies consisted of only one or two branches (Fig. 5). Seventy-one percent of all colonies sampled had only first-order branches, and 96% had not yet grown tertiary branches (Fig. 5), indicating that they were all at young developmental stages.
https://static-content.springer.com/image/art%3A10.1007%2Fs00227-006-0302-8/MediaObjects/227_2006_302_Fig5_HTML.gif
Fig. 5

Frequency distribution of the number of branches of red coral colonies at the Costa Brava (NE Spain). N=7600

Discussion

Colony abundance and distribution

Although depth affected colony diameter and height, only a weak correlation occurred between depth and population abundance. This may be due to natural small-scale variation in environmental conditions, resulting in the considerable patchiness and high variance in population abundance that we observed. Among several factors contributing to a suitable micro-habitat, space competition with other fast growing organisms that may produce a trophic shadow (Kim and Lasker 1997) could explain the discontinuous distribution of red coral patches in shallow zones. It is interesting that the regression analysis seems to confirm our field observations that in shallow water (18–25 m), corals grow in dense patches in cracks, crevices and overhangs, while below 35 m the colonies appeared to grow to larger size in less dense abundances outside of the typical habitats such as crevices. Further research may confirm if this is due to reduced competition from algae.

An opposite depth gradient in colony abundance has been found in a population in the Ligurian Sea, Italy, where red coral grows in extremely dense patches (200–1,300 colonies m−2), with an increasing abundance from 25 to 35 m (Cattaneo-Vietti et al. 1993; Santangelo et al. 1993a). Although the studied depth range was not very large and the number of italian patches tested small, the present study demonstrates a substantial difference between the NW Italian and the NE Spanish populations. Other data from the western Mediterranean are similar to the results of this study: 55 colonies m−2 at a depth of 40 m in Palma de Mallorca and 20 colonies m−2 at a depth of 60 m along the Costa Brava (FAO 1984). In Corsica, 90–100 colonies m−2 have been recorded (FAO 1984). Different techniques and spatial scales contribute to the observed variation and make comparisons difficult. The scraping technique, for example, detects the smallest colonies (except recruits) more reliably than do some other methods (Santangelo et al. 1993a). Nevertheless, colony abundance clearly is higher in the Ligurian Sea compared to other regions (Table 5). This leads to the conclusion that red coral population structure can vary widely among geographic areas, and environmental conditions (Table 5). Coral abundance is relatively high in all regions, even if harvested, which is a result of the relatively high recruitment rate of red coral (Grigg 1989; Santangelo and Abbiati 2001; Bramanti et al. 2003, 2005, 2006). To better understand red coral demography, it is essential to obtain more data on settlement and recruitment (Bramanti et al. 2005). It has been demonstrated recently that red coral recruitment patterns in Calafuria (Italy) differ from those at the Medas Islands (Spain) and Isola d’Elba (Italy); recruitment at the Medas Islands is lower and juvenile mortality higher (Bramanti et al. 2006). Reproductive features and population abundance may be key to understanding the recent development of red coral populations in these places (Tsounis et al. 2006). Another recently analyzed factor that may affect the demographic structure of red coral populations are mass mortality events. Bramanti et al. (2005) showed that the recruitment and mortality patterns of shallow red coral populations were affected by a mass mortality event detected off the French and Italian NW Mediterranean coasts in 1999 (Garrabou et al. 2001). In red coral populations of Provence (France), shallow and small colonies were affected by partial or total colony mortality (Garrabou et al. 2001), which may shape future demographic structure in that area.
Table 5

Comparison of colony abundances and maximum sizes among shallow water gorgonian species

Species

Abundancea (ind./m2)

Numberb

Area sampled (m2)

Depth (m)

Max. sizec (cm)

Location

Source

Pseudopterogorgia acerosa

0.52

33

64

0.5–2

100

Florida

Opresco (1973)

Pseudopterogorgia americana

0.02

1

64

0.5–2

86

Florida

Opresco (1973)

Pterogorgia citrina

4.1

262

64

0.5–2

46

Florida

Opresco (1973)

Pterogorgia anceps

0.08

5

64

0.5–2

47

Florida

Opresco (1973)

Paramuricea meandrina

0.18–1.43

42–54

>200

Hawaii

Grigg (1976)

Muricea californica

1.8–5.9

15–17

60

California

Grigg (1975)

Muricea californica

0.45–7.9

210

12–20

60

California

Grigg (1977)

Muricea fructicosa

0.1–2.35

210

12–20

40

California

Grigg (1977)

Muricea californicad

0.0005–0.1

6,050

12–20

60

California

Grigg (1977)

Muricea fructicosad

0–0.01

6,050

12–20

40

California

Grigg (1977)

Antipathes aperta

25

2,988

15–25

400

New Zealand

Grange and Singleton (1988)

Antipatharian spp.

0.02–1.5

200

17–27

Caribbean

Sànchez (1999)

Lophogorgia ceratophyta

6.8±3.0

123

40

19–22

100

Ligurian Sea

Mistri (1995)

Paramuricea clavata

56±23

17–29

150

Costa Brava

Coma et al. (1994)

Eunicella cavolini

162±15

625

25

50

Corsica (Channel)

Weinbauer and Velimirov (1996)

Eunicella cavolini

51±23

625

16–28

50

Corsica (Wall)

Weinbauer and Velimirov (1996)

Eunicella cavolini

68±21

“Some m2

10–15

50

Corsica (Boulders)

Weinbauer and Velimirov (1996)

Corallium rubrum

30

0.04

31–36

50

Ligurian Sea

Santangelo et al. (1993a)

Corallium rubrum

200–1,300

50

Portofino

Cattaneo-Vietti et al. (1993)

Corallium rubrum

5247

4,408

0.84

25–40

50

Ligurian Sea

Bramanti (2003)

Corallium rubrum

127±118

2,451

2,660

18–45

50

Costa Brava

This study

aAbundance per square meter within patches

bNumber of colonies counted

cLargest height that individuals were observed to reach

dIn this study, density was estimated by dividing the number of colonies seen during a dive by the area covered swimming

Biometric comparison of the six populations

Our sampling arrangement across a wide geographic range (the stations spanning across 18 km from north to south, Fig. 1) may help to understand how harvesting pressure affects red coral: The small differences in diameter and height between the five legally harvested sites along the Costa Brava (Begur, Montgrí, Cap de Creus South, East and North) indicate a high homogeneity of the Costa Brava population. In fact, throughout the Mediterranean, population structure and diameters of red coral are similar (Table 6). Red coral appears to be ubiquitous within its known distribution zone, and has a relative homogeneous demographic pattern.
Table 6

Comparison of Corallium rubrum population structure among geographic regions

Site

Source

Growth ratea (basal diameter) (mm year−1)

Mean basal diameter (mm)

Colony height (mm)

Population size structureb

Marseille, France

Garrabou and Harmelin (2002)

0.24±0.05

6.4±0.5

69.3±12

95% are <7 mm

Livorno, Italy

Santangelo et al. (1993a)

0.91

3.9

40

95% are <3.64 mm

Cap de Creus, Spain

García-Rodríguez and Massó (1986a)

1.32

7.2

61.8

99% are <15 mm

Costa Brava, Spain

Present study

4.8±2.1

27±17.1

98% are <7 mm

43% are <4 mm

Shown are means, or where available, means ± SD

aAnnual increase of the colony’s basal diameter in mm

bPercent of colonies with the given basal diameter in each population

The interactive effects of depth and site on colony diameter and height, as detected by ANCOVA, suggest that colonies in deeper habitats may be larger/older than those in shallow areas. This probably contributed to the standard deviation of the diameter and height data from the Medas Islands population, and can be explained by the fact that some of the deeper banks in the Medas Islands are particularly protected from human impact. They are situated at weather-exposed, remote and deep areas, which are not easily accessible to recreational divers and poachers. The more easily accessible zones, on the other hand, are exposed to a considerable human impact, as some of the 70,000 recreational divers that frequent the islands (Hereu et al. 2002) likely collect coral souvenirs, or damage colonies by poor diving technique. This may also explain why a previous study of red coral at the Medas Islands found almost no colonies with basal diameters >7 mm between 18 and 40 m depth (Linares et al. 2000). In this regard it must also be mentioned that poaching is a severe problem in both, MPAs and in harvested areas (see Hereu et al. 2002).

These first mesoscale data on red coral demography allow the comparison of six populations along the northwestern Spanish coast. The positively skewed size distribution shows a dominance of small colonies within all populations, which indicates a low survival of the larger size classes (Grigg 1976; Bak and Meesters 1998). In all populations there are virtually no colonies exceeding a basal diameter of 30 mm. However, in the Medas Islands population there is a higher percentage of medium-sized colonies, which are the result of its 14 years of protection. Inter-habitat differences in trophic conditions may explain minor differences in size distribution: rich food supply can lead to higher growth rates in gorgonians that enable young colonies to escape into larger size classes where survival is higher. Conversely, for old colonies, a better food supply does not necessarily lead to higher survival, and only affects growth rate (Yoshioka 1994). Differences in food supply thus may explain some of the differences in the population structure between the sites. However, due to the extreme difference between the Medas Islands and the other populations (Table 4), the trophic advantages are concluded to be a minor reason for the observed differences in size. Furthermore, although a noteworthy environmental characteristic at the Medas Islands is organic material input from the nearby river “Ter”, the thriving red coral populations are situated on the exposed northwestern side of the Medas Islands, where sedimentation induced by the river may be low (Rossi et al. 2003; Rossi and Gili 2005).

Natural mortality in C. rubrum is low, as it is a longevous species with life spans of up to 100 years (Riedl 1983; García-Rodríguez and Massó 1986c; Garrabou and Harmelin 2002). Recent data confirm that red corals have low mortality, as 60% of the recruits on experimental settling panels survived more than 22 years (Garrabou and Harmelin 2002). Natural mortality is not only associated with competition but also base detachment due to a variety of causes, such as damage due to waves (Grigg 1976), mechanical failure of weak substrate, or parasite-induced weakness of the base (Branch 1984; Yoshioka and Yoshioka 1991; Cattaneo-Vietti et al. 1993; Corriero et al. 1997). However, natural mortality is still low compared to mortality from harvesting (G. Tsounis, unpublished data).

Size/age structure

Historic data on coral height documents the former existence of enormous colonies more than 50 cm in height (Tescione 1973; Garrabou and Harmelin 2002). Despite this growth potential, the present study shows that 91% of colonies in the Costa Brava population are smaller than 5 cm in height, and none are larger than about 30 cm within our sampled depth range. This illustrates the level of harvesting pressure, as its selectivity should be towards large colonies. Furthermore, only 19 years ago the population structure in the same area was less skewed towards a basal diameter of 7 mm than today, showing a harvested but older population off Cap de Creus (García-Rodríguez and Massó 1986a). Nineteen years ago only 15% of the population showed a basal diameter <5 mm, whereas now it is 65%. Even accounting for a sampling bias toward larger colonies due to subsampling from professional harvests in the original study, the population structure was noticeably different from data in the present study. The study of García-Rodríguez and Massó (1986c) pointed out the importance of developing improved red coral fishery management at the Costa Brava, but harvesting pressure has continued with little changes until today.

Size is the major determinant of first reproduction, reproductive output and survival in gorgonians (Yoshioka 1994; Coma et al. 1995). Colony height frequency distributions are useful as they may be compared easily with historical records or with observations from fishermen and divers. However, due to larger variability, height is not the best suited parameter for determining the age of a colony (Santangelo et al. 1993a). Nevertheless it is important to stress that colonies of a mean height of 2–3 cm, consisting of just one or two branches, are unlikely to be able to ensure the reproductive potential needed for the survival of a population. Future work may confirm initial indications that the reproductive potential of such distorted populations appears to be severely diminished (see Tsounis et al. 2006). A variable relationship between height and age may be related partially to the fact that red coral is an exploited species: A coral colony can be deprived of a large part of its height, e.g. by partial harvesting, whereas its basal diameter is much less prone to mechanical influence. The basal diameter of red coral colonies, on the other hand, is highly correlated with age and has been used in several studies to calculate the age of gorgonians (Grigg 1976; García-Rodríguez and Massó 1986b; Santangelo et al. 1993a; Bramanti et al. 2005; Marschal et al. 2004).

In order to establish precise age distributions, it will be necessary to incorporate growth studies on the same populations, which was not attempted in this study. Recent longtime experimental studies using recruitment panels found low increases in basal diameter of 0.24±0.05 mm per year (Garrabou and Harmelin 2002), 0.62 mm per year (Bramanti et al. 2005), and 0.35 mm per year (Marschal et al. 2004), indicating that the higher values estimated by growth ring analysis in the past have overestimated growth due to non-validated sclerochronology methods (Garrabou and Harmelin 2002; G. Santangelo, personal communication). Further differences stem from interhabitat variation, which affects growth rate, as hydrodynamic conditions influence food availability to suspension feeders (Sebens et al. 1997). Thus, it is likely that 0.24 mm describes growth in habitats of reduced water movement (Garrabou and Harmelin 2002). The Costa Brava, however, is a relatively eutrophic and weather-exposed part of the Mediterranean (Rossi et al. 2003; Rossi and Gili 2005), and coral divers in this region who harvested the same coral patches during the last 30 years estimate diametral growth to be up to 1 mm per year (A. Pluja, personal communication). Even if the approach of professional divers is very approximate and qualitative, Marschal et al. (2004) found a basal diameter growth range of 0.14–0.75 mm, thus confirming the possibility of a wide range of growth rates depending on environmental conditions. Consequently, a diameter growth rate of 0.5 mm per year (a value between Bramanti et al. 2005 and Marschal et al. 2004) was used here to convert basal diameter into age.

Using this conversion, the mean age of our colonies is 9.6±4.2 years, and the most frequent age class is 8.2 years old. Thus, compared to earlier data from the same site, colonies now show a much younger average age (García-Rodríguez and Massó 1986a). This change is due partially to a bias toward older colonies as the initial study subsampled colonies from professional harvests (Table 6). However, given the legal minimum harvest size of 7 mm, an average diameter of 7.8 mm in a professional harvest appears extremely low and indicates nearly depleted stocks. The present branch morphology of the Costa Brava population also leads to the conclusion that the populations are relatively young, as the vast majority of the observed colonies lack the arborescent growth form that is characteristic of older colonies (see Garrabou and Harmelin 2002). It appears that no example of such a situation has been recorded for other octocorals, leaving overharvesting as the only explanation.

Comparison with other species

A paucity of unharvested red coral populations makes it difficult to distinguish human-induced mortality from natural mortality. Thus a comparison with other species serves to obtain a more general view of the biology of octocorals when interpreting size distributions (Table 5).

Colony abundance of the octocoral Leptogorgia sarmentosa in the Ligurian Sea, Italy (6.8±3 colonies m−2, Mistri 1995), is low compared to red coral. This difference can be explained by the fact that L. sarmentosa occupies a different ecological niche of soft bottom plains characterized by gravel and small boulders (Mistri 1995; Rossi et al. 2004). Furthermore, it is a fast growing species compared to red coral, with a soft gorgonine skeleton unlike the calcareous one in C. rubrum (S. Rossi, unpublished data). Thus, as abrasion due to wave movement may increase mortality (Grigg 1976), dense populations are disadvantageous for L. sarmentosa.

The age distribution in L. sarmentosa differs markedly from that in red coral populations, as extremely old individuals have been found in the L. sarmentosa population. The most frequent age recorded was 9 years (the first age class is usually missed by most studies due to the difficulty of seeing small colonies), with a considerable part of the population reaching 30 years (Mistri 1995). This age distribution shows that the studied L. sarmentosa population consists of a higher quantity of old individuals than any studied red coral population, presumably due to the lack of human-induced mortality. Low mortality is expected for an unharvested species such as L. sarmentosa. However, a black coral (Antipathes aperta) population studied in a New Zealand fjord showed that even an unharvested octocoral population may consist of mainly young colonies: 90% were smaller than 50 cm, while the largest colonies reach 4 m (Grange and Singleton 1988). The authors explained this size structure by a high mortality, possibly due to landslides common at the site.

Another octocoral living in the Mediterranean coralligenous community is the aposymbiotic gorgonian Paramuricea clavata. Its colonies may reach 1 m in height and have a similar depth distribution to that of red coral (True 1970; Weinberg 1978). The abundance of this non-exploited cnidarian varies between 22 and 54 colonies m−2, which is less than in red coral colonies (Weinberg 1978; Coma et al. 1994). However, it is unlikely that a natural red coral population could maintain high colony abundance if it contained many colonies of maximum size (50 cm in height), because the dominance of a few very large colonies may hinder the growth of newly settled ones, mainly due to trophic shadowing (Zabala and Ballesteros 1989).

In shallow populations (<9.5 m depth) of Pocillophora meandrina on the islands of Oahu (Grigg 1976) and Hawaii (Grigg and Maragos 1974), large colonies are more abundant on the leeward than on the windward side. This indicates that the size frequency, distribution, and abundance of P. meandrina depend less on fishery mortality than on natural mortality (Grigg 1976). Thus, in shallow water gorgonians natural mortality from wave action plays a significant role, unlike in red coral which is a sciaphilic gorgonian with exceptionally low mortality (García and Massó 1986c; Garrabou and Harmelin 2002).

Conclusions

This study demonstrates that the three-dimensional structure created by red coral populations at the Costa Brava has degraded to a “grass-plain”-like structure, from the original “forest”-like structure, which was still observable 20 years ago (García-Rodríguez and Massó 1986a). The transformation of the population structure of this exploited benthic cnidarian may be even more dramatic if we consider the qualitative data gathered in Tescione (1973), and it seems that all over the western Mediterranean the situation is similar (Santangelo et al. 1993b; Garrabou and Harmelin 2002; Santangelo et al. 2004). The major part of C. rubrum populations off the Costa Brava consists of young, small colonies. A long history of exploitation has changed its population structure in all unprotected habitats. Size differences between harvested areas and the Medas Islands MPA are in part due to deeper, less accessible sites inside the MPA. The relatively short time of protection (14 years) has allowed this area to recover sufficiently to result in a significantly different population structure compared to exploited areas. Still, 14 years has not been long enough to allow the population to recover to a completely natural state: virtually no colonies larger than 30 cm height were found, which is considerably less than the maximum size recorded for this species. Completely protected marine reserves will undoubtedly prove useful for future research on red coral and are well worth the effort. Future work should employ longtime monitoring, as it is the most reliable way to contribute to our understanding of growth, mortality, and recruitment in this species.

Size structures like the ones described in this study are a useful descriptor for the state of a population, as shifts indicate changes in recruitment and/or mortality. In combination with knowledge about the biology of a species (especially growth rates, reproductive cycle, and feeding ecology), size structure is key to detecting species responses to environmental and anthropogenic factors, and forms a basis for management decisions.

Ackowledgements

We are grateful to C. Orejas, T. Brey, G. Santangelo, S. Thatje, and to three anonymous reviewers whose comments greatly improved the manuscript. N. Fernández and L. Bramanti provided invaluable help during SCUBA diving fieldwork. We thank M. Aranguren and L. Vera for their help in the processing of photographic samples. Many thanks to the family Mörker for logistics at Roses and to JM Llenas for support at the Medas Islands. Thanks also to the professional divers harvesting red coral in the Costa Brava for advice. G.T. was supported by a PhD scholarship from the University of Bremen, Germany. This study was funded through European Union Funds by the Department of Fisheries and Agriculture of the Government of Catalonia (Spain), PCC:30103.

Copyright information

© Springer-Verlag 2006