Marine Biology

, Volume 144, Issue 4, pp 641–651

Coral recruitment: a spatio-temporal analysis along the coastline of Eilat, northern Red Sea

Authors

  • D. Glassom
    • Interuniversity Institute for Marine Science
    • Faculty of Life SciencesBar Ilan University
    • Oceanographic Research Institute
  • D. Zakai
    • Interuniversity Institute for Marine Science
    • Faculty of Life SciencesBar Ilan University
    • Israel Nature and National Parks Protection Authority
    • Interuniversity Institute for Marine Science
    • Faculty of Life SciencesBar Ilan University
Research Article

DOI: 10.1007/s00227-003-1243-0

Cite this article as:
Glassom, D., Zakai, D. & Chadwick-Furman, N.E. Marine Biology (2004) 144: 641. doi:10.1007/s00227-003-1243-0
  • 337 Views

Abstract

Recruitment rates of stony corals to artificial substrates were monitored for 2 years at 20 sites along the coast of Eilat, northern Red Sea, to compare with those recorded at other coral reef locations and to assess variation in recruitment at several spatial scales. Coral recruitment was low compared to that observed on the Great Barrier Reef in Australia, but was similar to levels reported from other high-latitude reef locations. Pocilloporids were the most abundant coral recruits in all seasons. Recruitment was twofold higher during the first year than during the second year of study. There was considerable spatial variability, with the largest proportion of variance, apart from the error term, attributable to differences between sites, at a scale of 102 m. Spearman’s ranked correlation showed consistency in spatial patterns of recruitment of pocilloporid corals between years, but not of acroporid corals. During spring, when only the brooding pocilloporid coral Stylophora pistillata reproduces at this locality, most coral recruitment occurred at central and southern sites adjacent to well-developed coral reefs. During summer, recruitment patterns varied significantly between years, with wide variation in the recruitment of broadcasting acroporid corals at northern sites located distant from coral reefs. Settlement was low at all sites during autumn and winter. This work is the first detailed analysis of coral recruitment patterns in the Red Sea, and contributes to the understanding of the spatial and temporal scales of variation in this important reef process.

Introduction

Patterns of stony coral recruitment are important to the recovery and replenishment of coral reefs after disturbances and to reef health in general (Gittings et al. 1988; Sammarco et al. 1991; Johnson and Preece 1992). Knowledge of recruitment processes facilitates implementation of conservation and management decisions on coral reefs (Dunstan and Johnson 1998) and has long been identified as a research priority (Harrison and Wallace 1990; Wells 1995). Studies of coral recruitment have been carried out for several decades, but they have increased recently in number and scope (e.g. Fisk and Harriott 1990; Maida et al. 1994; Dunstan and Johnson 1998; Tioho et al. 2001). The descriptive and methodological focus of early studies (Loya 1976a; Wallace and Bull 1981; Harriott and Fisk 1988) has been complemented by work on larval dispersal and on the question of whether reefs are self-seeding (Jokiel 1984; Williams et al. 1984; Richmond 1987; Sammarco and Andrews 1989; Black et al. 1990; Fisk and Harriott 1990; Adjeroud and Tsuchiya 1999; Nishikawa et al. 2003). Recent work has paid explicit attention to the elucidation of spatial and temporal variation in coral recruitment at different scales, the mechanisms underlying such variation, and the role of recruitment in determining benthic community structure (Babcock 1988; Baird and Hughes 1997; Connell et al. 1997; Dunstan and Johnson 1998; Hughes et al. 1999, 2000; Harii et al. 2002).

Some important issues in coral recruitment remain unresolved, such as the potential sources of larvae, their patterns of dispersal, and the effects on spatial variation in recruitment. While some population genetic studies (e.g. Ayre and Hughes 2000) indicate that most coral larvae do not travel far, others indicate genetically undifferentiated populations over distances of up to 50 km (Ayre et al. 1997; Adjeroud and Tsuchiya 1999; Ridgway et al. 2001; Nishikawa et al. 2003).

The only published study of coral recruitment at Eilat, located at the northern tip of the Red Sea, was conducted by Loya (1976a) on a single artificial substrate at one reef site. To date, there has been no detailed analysis of spatial or temporal variation in coral recruitment in the Red Sea. The present study describes variation in coral recruitment at a range of spatial and temporal scales along the coastline at Eilat. The reefs near Eilat are experiencing severe anthropogenic disturbance, with loss of coral cover and evidence of eutrophication in recent years (Fishelson 1995; Zakai and Chadwick-Furman 2002). Since coral recruitment may be affected by poor water quality (Richmond and Hunter 1990; Richmond 1993), this process may serve as an important indicator of the condition of local reefs.

Materials and methods

Eilat lies at the extreme northwestern end of the Gulf of Aqaba, a semi-enclosed basin forming one of two northern extensions of the Red Sea. The Jordanian city of Aqaba borders the gulf to the northeast. The northern end of the gulf is ~4.5 km wide and 800 m deep. A short, 1-km-long fringing reef occurs along part of Eilat’s coastline (at the Nature Reserve, Fig. 1), and patch reefs of varying size and coral cover extend to the north and south of the fringing reef. The northern coastline contains only a few patch reefs, and is comprised mostly of soft substrate. Due to the steep bathymetry of the area, coral reef communities are limited to a narrow band along most of the coast. Net water movement is northward along the coast from October to February, and southward during the rest of the year (Genin and Paldor 1998). At the extreme northern end of the gulf, between the cities of Eilat and Aqaba, water movement changes from easterly to westerly at intervals of several weeks (Brenner et al. 2001).
Fig. 1

Map of Eilat in the northern Gulf of Aqaba, Red Sea, showing 20 study sites examined for coral recruitment. The border with Egypt is shown at lower left, and that with Jordan at upper right (IUI Interuniversity Institute for Marine Science)

Coral recruitment was monitored at 20 sites along the coast, ranging from the Egyptian to the Jordanian borders (Fig. 1). The sites varied in density of coral cover and in geographic position. Sites were assigned to three areas. The southern area contained nine sites between the Egyptian border at Taba and the Interuniversity Institute for Marine Science (IUI) (sites 1–9, Fig. 1). The central area encompassed four sites, two within the Nature Reserve and two immediately north and south (sites 10–13, Fig. 1). The northern area contained seven sites, ranging from the oil terminal to the city of Eilat (sites 14–20, Fig. 1).

In July 1998, three galvanized metal racks were deployed 1–3 m apart at 6 m depth at each site. All racks were constructed so that tiles were suspended at a 45° angle, since coral recruitment has been positively correlated with the angle of recruitment surface (after Carleton and Sammarco 1987; English et al. 1997). Four unglazed ceramic plates (10×10 cm each) were attached to each rack. Thus, spatial variation in coral recruitment was examined at four scales: between plates (N=4 per rack), racks (N=3 per site), sites (N=4–9 per area), and areas (N=3).

The recruitment plates were collected and replaced with new plates at intervals of 3 months for 2 years. The relatively short intervals allowed observation of seasonal variation in coral recruitment in an area where spawning is asynchronous between coral species (Shlesinger and Loya 1985; Shlesinger et al. 1998). Plates were examined under a binocular microscope, and all corals counted and identified to family where possible (after English et al. 1997). Due to seasonal differences in the release of propagules between genera of pocilloporid corals at Eilat (Shlesinger and Loya 1985), some genera of coral spat were inferred from the time of year at which they settled. Each coral was measured, and the number of polyps counted.

At 18 of the sites (excluding sites 11 and 12), the percent cover of other sessile organisms was determined using a point–intercept method on a grid with 122 points. At the same 18 sites, two additional plates were placed on each rack (i.e. six plates per site) during March 1999 and left undisturbed for 15 months, to examine the longer-term survival of settled corals and temporal variation in the recruitment of corals belonging to different families. For plates that had been in the water for 15 months, the surface area of each coral was calculated. For large corals, the length and mid-point diameter of each branch was measured, and branch area was calculated. Areas of all branches were added to yield the surface area of each entire colony.

Statistical analyses were conducted using SAS version 8. The experimental design was unbalanced, due to variation in the number of sites per area examined, so model-2, nested ANOVAs were conducted using the GLM procedure (SAS 1999). Figures in the text are presented as means (±SE).

Results

Short-term coral recruitment

We observed 935 stony coral recruits over 2 years. Most recruits (72.7%) belonged to the coral family Pocilloporidae, 11.0% belonged to the family Acroporidae, and 16.1% belonged to other coral families. In addition, a single colony of the hydrocoral genus Millepora (0.1% of all recruits) was observed on the plates. No coral recruits from the family Poritidae were observed. Recruitment to the plates varied both spatially and temporally (Fig. 2). Only colonies of Stylophora pistillata, a member of the family Pocilloporidae, were observed to release propagules between December and March (Shlesinger et al. 1998; D. Glassom, personal observations). All coral recruitment in March consisted of pocilloporid corals, most likely individuals of S. pistillata. During both years in March, recruitment at site 16, at the northern end of the commercial port (Fig. 1), was twice that at any other site (Fig. 2). In contrast, during June of both years, the highest recruitment was at site 8 (IUI), and site 12 (Nature Reserve, Fig. 1). Adult colonies of the stony coral S. pistillata occur at high densities on reefs adjacent to both these sites (D. Glassom, personal observations).
Fig. 2

Spatial and temporal variation in stony coral recruitment along the coast of Eilat, northern Red Sea. Shown are 20 sites, ranging from south (site 1) to north (site 20), examined seasonally over 2 years

During late summer (September), recruitment was overall lower than in June, but more evenly spread among sites (compare y-axis scales in Fig. 2). Pocilloporid corals again were numerically dominant, but at this time of year likely consisted of individuals of both Pocillopora and Stylophora, as inferred from their spawning seasons (Shlesinger et al. 1998). Data from December were excluded from all analyses, since recruitment was very low (a total of six and four coral recruits were found at all sites combined, in December 1999 and 2000, respectively).

The abundance of coral recruits ranged from a low of 0.16±0.49 recruits per 0.01 m2 area in March 2000 to a high of 1.70±0.63 in June 1999 (Fig. 2). The abundance of coral spat in June 2000 (0.81±1.28) was approximately half that observed in the previous year. The total number of coral recruits was significantly higher (ANOVA, F=15.82, df=2, P=0.0001) in 1999 than in 2000, but spatial patterns of recruitment between sites were similar during spring and summer of both years. Spearman’s correlation analyses showed that sites were significantly correlated for March and June between years (Table 1). Only in late summer (September), did the spatial pattern of recruitment differ significantly between years (Fig. 2; Table 1). Differences in the abundance of acroporid coral recruits at the northernmost sites between years largely accounted for this variation. In 1999, most acroporids recruited to northern sites, while pocilloporids were more abundant in the south (Fig. 3). In 2000, few acroporid recruits were found at any sites, resulting in a concomitant drop in total recruitment toward the north.
Table 1

Spearman’s rank correlations comparing spatial patterns of coral recruitment between 2 years (1999 and 2000) during each of three seasons (ending in March, June, and September) at Eilat, northern Red Sea. A significant correlation (P<0.05) indicates similar spatial patterns of recruitment during 1999 and 2000. The pattern of recruitment of pocilloporid corals was separated from that of other coral families only during late summer (September), because this was the only season during which both non-pocilloporid and pocilloporid recruits appeared on the plates. During March and June, all coral recruits were pocilloporids

 

All corals

Pocilloporidae

N

Rs

P

N

Rs

P

March

60

0.508

0.001

June

60

0.690

0.001

September

60

0.019

0.885

60

0.260

0.057

Fig. 3

Variation in recruitment between families of stony corals during July to September 1999, at Eilat, northern Red Sea. No data were available for sites 11 or 12 (see “Results”)

Variation in recruitment at each spatial scale was revealed by the variance component of each factor of the ANOVA model containing spatial data (Table 2). Differences between plates within racks (the error term in the model) accounted for the highest proportion of variance; this represents variation at a scale of 100 m. This was followed by differences between sites (102 m) in two of the three seasons. Neither area (103 m), rack (101 m), nor year contributed substantially to overall variance, even where these factors were statistically significant. In the analysis for September, variance attributable to the interaction between year and site overshadowed that attributable to site alone (Table 2).
Table 2

Spatio-temporal analysis of coral recruitment during three seasons at Eilat, northern Red Sea. Analyses are model-2 ANOVAs, with all spatial factors nested in a hierarchical fashion. Statistics are combined for all families of corals. The error term represents individual plates on settlement racks. Each level of the spatial data is nested within the higher levels, i.e. there are four plates per rack, three racks per site, and several sites in each area examined. Note that recruitment varies significantly between sites during all three seasons, but not between racks within site [n.s. not significant (P>0.05), *P<0.01, **P<0.001]

Variance source

df

F

P

Percent of variance

March

  Year

1

1.70

n.s.

0

  Area

2

11.17

**

0.7329

  Site

17

8.45

**

21.05983

  Rack

40

0.76

n.s.

0

  Year×Area

2

2.87

n.s.

1.2459

  Year×Site

17

1.26

n.s.

2.7262

  Error

434

74.2356

June

  Year

1

37.64

**

6.6603

  Area

2

38.32

**

5.9724

  Site

17

18.34

**

28.9324

  Rack

40

1.01

n.s.

1.1729

  Year×Area

2

5.87

*

1.8427

  Year×Site

17

2.81

**

7.2377

  Error

430

48.1816

September

  Year

1

12.38

**

2.5147

  Area

2

0.87

n.s.

0

  Site

17

3.17

**

0.0136

  Rack

40

1.31

n.s.

2.6545

  Year×Area

2

7.52

**

1.7829

  Year×Site

17

2.61

**

10.5018

  Error

434

82.5325

The largest coral found on the short-term plates was a pocilloporid colony of 6.1 mm diameter that consisted of 35 polyps, corresponding to a maximum possible age of approximately 3 months (as inferred from the length of time the plates were submerged). Pocilloporid corals on the tiles were on average 2.31±0.04 mm diameter and had 6.77±0.21 polyps (N=469). The largest acroporid recruit had a diameter of 2 mm and consisted of six polyps. Acroporid corals were on average 1.04±0.03 mm diameter and had 1.34±0.11 polyps (N=61). There were no clear differences in recruit size between sites, seasons or years.

Long-term coral recruitment

Recruitment to plates that had been in the water for 15 months mirrored patterns found for the early summer period (June) during both years of the study (compare Figs. 2 and 4a), reflecting the dominance of the pocilloporid coral S. pistillata in determining overall recruitment patterns. Of the 268 corals observed on these plates (2.53±3.03 corals per plate, N=106 plates), there were 237 pocilloporids, 5 acroporids, 3 poritids and 23 others. Recruitment of coral spat after 15 months showed similar patterns of variation to those after 3 months, and almost mirrored between-site variation for June of each year (Table 3), despite the fact that these data integrate two brooding seasons and an intervening spawning season (Shlesinger et al. 1998). Plates within racks again accounted for the largest portion of the variation, followed by site within area. Area comprised a relatively small proportion of the variation, despite being statistically significant, and, notably, “racks within sites” was once again insignificant, both statistically and in terms of contribution to the total variance (Table 3).
Fig. 4

a Spatial variation in the recruitment of stony corals to long-term plates at Eilat, northern Red Sea. b Size distribution of pocilloporid coral spat on the long-term plates. Note that the x-axis is a log scale. The largest coral was approximately 43.4 mm in diameter

Table 3

Spatial analysis (model-2 ANOVA) of coral recruits of all families combined on long-term plates immersed for 15 months at Eilat, northern Red Sea. Levels of spatial data are nested as in Table 2 [df 1 for “area” due to lack of data for sites 11 and 12 in the nature reserve (area 2). Note that recruitment does not vary significantly between racks within each site; n.s. not significant (P>0.05), *P<0.001]

Variance source

df

F

P

Percent of variance

Total

107

2.44

*

100

Area

1

14.04

*

9.6079

Site

16

4.66

*

32.0038

Rack

36

1.14

n.s.

3.8024

Error

54

54.5858

A comparison of the number of corals on plates left in the water for 15 months with those changed every 3 months showed no significant difference in terms of either the total number of corals or the number of pocilloporids (Table 4). However, significantly fewer acroporids than expected were found on plates that had been in the water for 15 months (Table 4), indicating that acroporids may have suffered higher mortality on the plates than did pocilloporids. This impression is supported by the sizes of corals from these two families that were found on the 15-month plates. Despite the fact that the plates were taken out of the water some 9 months after the spawning season for most Acropora corals (Shlesinger et al. 1998), the largest acroporid found had only 25 polyps and a surface area of 26 mm2. The largest pocilloporid, in contrast, had several branches and a surface area of 1887 mm2. The number of polyps on this pocilloporid coral could not be counted directly, but was approximately 1085, as estimated from a regression of polyp number on coral colony surface area (linear regression test, R2=0.93, P<0.01, N=43 corals examined).
Table 4

Comparison of the abundance of coral recruits (number of individuals per 0.01 m2 area, mean+SE), on plates immersed continuously for 15 months, versus aggregate numbers, on five consecutive sets of plates that were each immersed for 3 months (5×3 months), at Eilat, northern Red Sea [all corals those from both families, plus unidentified corals; n.s. not significant (P>0.05), *P<0.01]

5×3 months

15 months

N

t

P

Acroporidae

0.269 (0.057)

0.057 (0.024)

18

09.34

*

Pocilloporidaes

1.866 (0.558)

2.236 (0.519)

18

0.3

n.s.

All corals

2.667 (0.699)

2.519 (0.552)

18

0.18

n.s.

The size-frequency distribution of pocilloporid corals on the long-term (15-month) plates is shown in Fig. 4b. After log-normal conversion, this distribution approximates normality, showing a similar size-frequency distribution to natural coral populations (e.g. Bak and Meesters 1998).

Recruitment of other sessile organisms

Bryozoans, bivalves, and serpulid worms were the major occupiers of space on the recruitment plates. The relative abundance of each group, on plates changed every 3 months, varied over the 2 years of the study. The percent cover of bivalves declined at the southern sites during the second year, while those of serpulids and bryozoans increased (Fig. 5). By September 2000, bivalves were almost entirely absent from the recruitment plates, comprising only 0.27% of total cover, while serpulid worms occupied 47.5% of space on the plates.
Fig. 5

Spatial and temporal variation in the percent cover of major sessile organisms on recruitment plates changed every 3 months at Eilat, northern Red Sea, during 1999–2000. No data were available for sites 11 or 12 (see “Results”) (filled bars 1999; open bars 2000)

For all dates combined, the number of stony corals that recruited to the plates was negatively correlated with the percent cover of serpulids and bryozoans (Table 5). However, this pattern was not consistent between years or seasons. In June 2000, the number of corals was positively correlated with the percent cover of bryozoans and bivalves, although the correlation coefficients were low (Table 5).
Table 5

Pearson’s correlations between the most abundant groups of sessile organisms on recruitment plates at Eilat, northern Red Sea. Only significant correlations are shown here. A further 28 tests, in which no significant correlation was detected, were excluded for the sake of clarity. Loss of plates in some seasons resulted in differences in N (*P<0.05, **P<0.01, ***P<0.001)

Organisms correlated

Date

N

R

P

Corals and serpulids

All dates combined

1051

−0.10

**

Corals and bryozoans

All dates combined

1267

−0.08

**

Corals and serpulids

Jun 1999

212

−0.277

***

Corals and bivalves

Sep 1999

211

−0.215

**

Corals and bivalves

Mar 2000

211

−0.202

**

Corals and bivalves

Jun 2000

212

0.186

**

Corals and bryozoans

Jun 2000

212

0.259

***

Serpulids and bryozoans

Sep 1999

211

−0.143

*

Serpulids and bryozoans

Mar 2000

211

−0.233

**

Serpulids and bryozoans

Jun 2000

212

−0.202

**

Serpulids and bryozoans

Sep 2000

205

−0.431

***

Serpulids and bivalves

Jun 1999

212

−0.295

***

Serpulids and bivalves

Sep 1999

211

−0.504

***

Serpulids and bivalves

Mar 2000

211

−0.357

***

Sessile organisms appeared to interact on the plates. In September 1999, serpulids were negatively correlated with bryozoans. Serpulids and bivalves also were negatively correlated in June and September 1999, and in February 2000 (Table 5). While the correlation coefficients of these interactions were generally higher than those of the interactions with corals, the majority of tests (28 of 42) showed no significant correlation between the abundance or percent cover of groups of sessile organisms on the plates.

Discussion

Coral recruitment

Scleractinian corals in the Red Sea do not mass-spawn synchronously, but colonies of most species spawn sometime during the summer (Shlesinger et al. 1998). The only stony corals that release propagules at other times of the year are colonies of the pocilloporid coral Seriatopora hystrix and of the poritid coral Alveopora spp., which release brooded planulae during autumn and winter, and those of the pocilloporid Stylophora pistillata, which release brooded planulae from late winter to early summer. Hence, during spring to mid-summer (March and June), coral recruits were almost exclusively pocilloporids, and most likely those of S. pistillata. Pocilloporid recruits also were numerically dominant during late summer, but probably included individuals of other genera such as Pocillopora. The asynchronous spawning of corals in the Red Sea, in contrast to highly synchronous spawning in Australia, may occur due to differences in environmental conditions such as annual temperature range (Richmond and Hunter 1990) and variation in the intensity of competition for settlement space (Shlesinger and Loya 1985; Loya 1990). About half of the 24 coral species examined in the Red Sea are known to spawn during periods similar to those of allopatric populations in the Great Barrier Reef of Australia, despite the seasons being reversed (Shlesinger and Loya 1985; Richmond and Hunter 1990; Shlesinger et al. 1998).

The wide variation between years in the recruitment rate of acroporid corals observed here in the Red Sea is similar to patterns known for reefs in Australia (Harriott and Banks 1995; Dunstan and Johnson 1998). Recruitment of acroporid coral spat at Eilat was relatively low and occurred in 1999 predominantly at northern sites (Fig. 3).

In a previous study of coral recruitment at Eilat, Loya (1976b) found up to 60 recruits m−2 on a single artificial substrate. This recruit density is similar to that observed in the present study at the site closest to that of Loya’s (site 9 in Fig. 1) during June 2000, but is <25% of the coral recruit density found in June 1999. During the in situ study of Loya (1976b), only recruits >1 month old were detectable; thus, coral recruitment may have been underestimated in his study.

Comparison of coral recruitment between reef sites worldwide is hindered by differences in the methods employed. Substrate material (Harriott and Fisk 1988), method of attachment (Mumby 1999), and duration and depth of plate deployment all vary between recruitment studies, and may affect results. In addition, environmental conditions that vary naturally between sites, such as water temperature (Coles 1985; Wilson and Harrison 1997), current patterns (Willis and Oliver 1988), and the effects of herbivores (Gleason 1996) and allelopathic organisms (Koh and Sweatman 2000; Maida et al. 2001) all may influence coral recruitment rates. The use of recruitment plates of different sizes also may affect apparent abundances of coral recruits, if abundances are standardized and reported as number of individuals per unit area (Birkeland et al. 1981; Glassom 2002). Nonetheless, an overview of recent studies using artificial substrates (Table 6), in combination with a survey of earlier work (Smith 1992), allows patterns of coral recruitment in the Red Sea to be considered in a global context.
Table 6

Comparison of recruitment patterns of stony corals at coral reef locations around the world, illustrating the range of techniques used and results obtained. For studies prior to 1992, see the summary by Smith (1992). Abundances of coral recruits were averaged across sites and years within each location, except where experimental conditions made this inappropriate, in which case only numbers on control plates were used. Where sampling was seasonal, means for different seasons were added to give numbers per square meter per year. Where means were not given, they were calculated or estimated from points on graphs (GBR Great Barrier Reef in Australia; CE ceramic; CO coral skeleton; PL plastic; LI limestone; S single plates; T stack of plates; P pairs of plates with gap between them; D directly on reef substrate; R on racks; H horizontal; V vertical; A angled; BRD brooding corals; SPA broadcast-spawning corals)

Reef location

Material

Lone/Pair

Method attached

Orientation

Depth (m)

Area examined (cm2)

Recruitment (no. m−2 year−1)

Most common mode

Time immersed (months)

Source

Pacific

French Polynesia

CE

S

R

H/V

225

131

BRD

4

Gleason (1996)

Guam

PL

S

D

H/V

6–36

75/225

57

1.5–6

Birkeland et al. (1981)

North GBR

CE

P

R

H

2

122

1840

BRD

2

Baird and Hughes (1997)

CE

P

R

H

2

122

526

BRD

2

Baird and Hughes (1997)

CE

T

R

4–5

225

1135

Varied

4 & 9

Maida et al. (1994)

Mid GBR

CE

S

D

H

1

122

4222

SPA

2

Hughes et al. (2000)

CE

S

D

H

1

122

4590

SPA

2

Hughes et al. (1999)

CO

S

R

A

3,14–18

3–400

2092

6 & 12

Sammarco et al. (1991)

CE

P

R

H

7–10

144

633

BRD

1.5

Babcock (1988)

CE/CO/PL

P

R

4

2044

SPA

4.5

Harriott and Fisk (1988)

South GBR

CE

S/P

R/D

H/V/A

9

122

150

BRD

Mumby (1999)

CE

S

R

H

9–12

400

307

BRD

5 & 12

Dunstan and Johnson (1998)

SE Australia

CE

P

R

9–19

225

173

Seasonal

0.5–1

Banks and Harriott (1996)

CE

P

R

6–9

225

132

BRD

5

Harriott and Banks (1995)

Red Sea

Eilat

CE

S

R

A

6

100

190

BRD

3–4

Present study

Caribbean

Bahamas

LI

S

R

H/V

10–100

225

106

4–8

Avery and Liddell (1997)

Barbados

CE

P

R

H/V

225

79

17

Hunte and Wittenberg (1992)

Bermuda

CE/LI

S

R

5–7

37

BRD

Smith (1992)

Numbers of coral spat per plate at Eilat (Fig. 2; Table 6) were low relative to those observed at most locations along the Great Barrier Reef (GBR) of Australia, particularly in the central GBR, but were similar to those found in the Caribbean Sea, and in other parts of the tropical Pacific Ocean (Table 6). Coral recruitment at Eilat also was similar to levels observed at high-latitude reefs located south of the GBR in Australia (Harriott and Banks 1995; Banks and Harriott 1996). The predominance of spat of brooding corals observed here also is similar to the pattern found in other reef areas, except for at the GBR (Table 6).

Several sites in the present study were located >5 km distant from most reef patches, at the north beach in Eilat (Fig. 1). This may have affected the arrival of larvae of brooding corals, which may have limited dispersal distances (Ayre and Hughes 2000; Tioho et al. 2001; Harii et al. 2002; Nishikawa et al. 2003; but see Sammarco and Andrews 1988, 1989). High spatial variation has been observed in the recruitment of brooding corals in Australia (Babcock 1988) and Japan (Tioho et al. 2001; Harii et al. 2002), and is similar to patterns observed here in the Red Sea for February and June (Fig. 2). In contrast, during September, when a large proportion of recruits was likely from broadcast-spawning corals (Shlesinger et al. 1998), there is no significant difference in the spatial distribution of recruits between sites in the northern versus southern areas (Table 2).

There is considerable debate regarding the likelihood and ecological significance of the dispersal of coral propagules between distant reefs (Williams et al. 1984; Sammarco and Andrews 1988, 1989; Richards et al. 1995; Roberts 1997; Mumby 1999; Tioho et al. 2001; Nishikawa et al. 2003). At Eilat, a high proportion of acroporid coral recruits occurred at northern sites, where adult coral densities are low. Circulation models for the Gulf of Aqaba, which show circular gyres with net southward water movement along the Eilat coast during the summer when most coral species spawn (Genin and Paldor 1998; Berman et al. 2000), raise the possibility that recruits along the north beach were transported from reefs in Jordan on the eastern side of the gulf. However, these models are incomplete, and eddies may result in larval retention at small scales (A. Genin, personal communications). Water movement at the extreme northern end of the gulf alternates between eastward and westward flow (Brenner et al. 2001), with flow direction changing irregularly at intervals of up to 1 month. Whether coral larvae reach reefs in Eilat from those in nearby Jordan may thus depend partly on the direction of water flow at the time of spawning, possibly contributing to the observed variability of coral recruitment (Fig. 2). Although the present study was not designed to assess patterns of coral larval dispersal, recruitment patterns observed along the linear coastline of Eilat, in combination with predictable current patterns, indicate that corals potentially could recruit to reefs in Eilat from those in Aqaba. Genetic comparisons of corals from opposite sides of the gulf are needed to test this hypothesis.

Levels of coral recruitment at Eilat were significantly lower in 2000 than in 1999. However, significant differences between only 2 years of data must be interpreted with caution, and this is especially true of corals that are known to have wide temporal variation in their levels of reproduction (Hughes et al. 2000). Eilat’s coral reefs have deteriorated recently and suffer from intense anthropogenic disturbance (Loya 1990; Fishelson 1995; Zakai and Chadwick-Furman 2002). A study of coral recruitment in the Nature Reserve (sites 11 and 12 in the present study) indicates a pattern of declining recruitment over the past 4 years (D. Zakai, unpublished data). Thus, our observed decline in recruitment during the second year of the present study may not be due entirely to natural fluctuation.

Analysis of the components of variance allows interpretation of variation in recruitment at different spatial scales (Underwood 1997; Hughes et al. 1999). In all analyses here, plates within racks, equating to a spatial scale of 100 m, comprised the largest part of observed variance, reaching a maximum of >80% of the total variance in September, partly due to the low overall numbers of corals per plate (Table 2). A high proportion of variance in coral recruitment has been found previously at this scale (Dunstan and Johnson 1998), although the overall number of recruits was much higher. Moreover, coral spat may aggregate at scales as small as 10 cm2 (Wallace and Bull 1981) and tend to settle in proximity to other coral spat (Lewis 1974). At Eilat, differences between sites accounted for the next highest portion of the variance, in three of four cases (Tables 2, 3). In September, the variance attributable to the interaction between sites and years overshadowed that of the difference between sites, largely due to variation in the recruitment of acroporids at the northern sites.

Racks within sites (100–101 m scale) contributed little to the variance in any of the tests, implying that patchiness in coral recruitment occurs at a larger scale than this. Differences between areas contributed little to the overall variation in coral recruitment, although numbers of settled corals differed significantly at the 0.05 level for three of the four analyses conducted (Tables 2, 3). The scale at which most patchiness occurred (100) was similar for plates that had been in the water for 3 months and for 15 months, even though long-term plates were exposed to two brooding seasons and one intervening spawning season (Shlesinger et al. 1998). Thus, most patchiness in coral recruitment, excluding the error term (plates on racks) occurred at the level of sites (102 m), but was evened out at larger (area) and smaller (racks in sites) levels.

Understanding the scales at which ecological processes occur is important for predicting the impact of disturbances and for management of coral reefs (Johnson and Preece 1992; Connell et al. 1997; Hughes et al. 1999; Sale 1999). Analyses of components of variance have been used to determine scales of variation in several studies of coral recruitment. Similar to the present study, a large proportion of variation in coral recruitment at other reef locations has been attributed to plates within racks (or to replicate plates within sites, if attached directly to the reef: Baird and Hughes 1997; Dunstan and Johnson 1998; Hughes et al. 1999). Racks within sites accounted for very little variation at Eilat (present study) or at Heron Reef in Australia (Dunstan and Johnson 1998), but higher proportions of variation were attributable to this scale in the central GBR (Babcock 1988) and on Lizard Island (Baird and Hughes 1997), especially for corals of the family Pocilloporidae. Differences between sites were responsible for less variation at other locations than at Eilat, and were observed to be particularly low at Heron Island by Dunstan and Johnson (1998). However, sites in other studies were further apart on average than were those in the present work, and in some cases distances between sites were equivalent to distances between areas chosen at Eilat (e.g. Hughes et al. 1999). In this light, it is noteworthy that area contributed only a small percentage to the variation observed here (Table 2).

Year contributed little to the overall variance, partly as a consequence of the short duration of the study, despite differences in the numbers of coral spat recruiting to plates between years for two of the three seasons tested (Table 2). Longer term studies on coral recruitment at Eilat would improve the resolution of the temporal data. Nonetheless, the contribution of temporal variation to the total variance has been found previously to be low in a study conducted over a longer period of time (Dunstan and Johnson 1999).

Partitioning of variance among spatial scales over 2 years was consistent for both brooding and spawning corals in a study on the GBR (Hughes et al. 1999). Correlation between years in spatial patterns of recruitment also has been observed in other marine communities (Connell 1985). In the present study, variance partitioning was consistent for pocilloporid corals, most of which are brooders, but not for acroporids, which are spawners. At Eilat, high pocilloporid recruitment during March and June of each year occurred at sites immediately south of dense aggregations of adult Stylophora pistillata colonies (D. Glassom, personal observations). The planula larvae of some brooding corals can survive and remain competent to settle for up to 100 days (Richmond 1987). However, they are equally capable of settlement immediately after release, and brooding coral species generally are not considered to disperse over large distances (Babcock 1988; Sammarco and Andrews 1989; Veron 1995; Nishikawa et al. 2003). Philopatry may result in more spatial variation of recruitment in brooding coral species than in spawners, as recorded by Babcock (1988). This variation may be consistent over time, especially in patchy reef environments, as reflected here by the significant Spearman’s correlations for March and June (Table 1).

Consistent spatial patterns of recruitment between years for the brooding coral S. pistillata at Eilat likely result from: (1) low dispersal distances of planulae of this species (Babcock 1988; Nishikawa et al. 2003), (2) discrete coral populations caused by the patchy nature of local reefs, and (3) a linear coastline with southward currents during most of the year (Genin and Paldor 1998). Although most coral recruits at Eilat were pocilloporids, there was no significant positive correlation in spatial variation during September between years. Spawning data indicate that at this time of year, many pocilloporid coral recruits likely belong to the genus Pocillopora rather than to Stylophora (Shlesinger et al. 1998). The most common local species of Pocillopora, P. verrucosa, is a broadcast spawner in the Red Sea (Fadlallah 1983). Hence, a large proportion of broadcast-spawning corals during September may have concealed patterns that were apparent in other seasons (Fig. 2), due to the long-distance dispersal of these larvae.

The potential of artificial reefs to alleviate diving pressure on natural reefs at Eilat is currently a focus of investigation (Wilhelmsson et al. 1998; Zakai and Chadwick-Furman 2002). The consistency that we observed in coral recruitment patterns between sites during most seasons may contribute to management decisions on the location of artificial reefs at Eilat. Such consistency was attributable primarily to the recruits of a single abundant coral species, S. pistillata. This coral is an opportunist colonizer of new substrates (Loya 1976a, 1976b), and may serve as an important precursor to the settlement of other coral species on artificial marine substrata.

Recruitment of other sessile organisms

Some sessile organisms, such as bryozoans and bivalves, may affect coral recruitment negatively (Birkeland 1977; Dunstan and Johnson 1998). However, the calcareous tubes of serpulid worms may provide secondary substrates for coral settlement. On our plates, several coral recruits were observed to settle on serpulid tubes, although there also were some cases of corals being partially overgrown by serpulids. Additionally, serpulid genera such as Filograna and Filogranella formed dense mats loosely attached to the substrate, which may have prevented coral larval settlement, since no spat were observed on these mats. Overall, there was a negative correlation between the number of coral recruits per plate and the percent cover of bryozoans and serpulids. This pattern was not consistent between years or seasons, however, and there were even two cases in which abundance of coral spat was positively correlated with the cover of bryozoans (Table 5). The correlation coefficients were low for all these tests, and it is doubtful that they are of much biological significance. There also were numerous negative correlations between groups of sessile organisms. It is likely that the relatively low numbers of corals per plate, as well as interactions between the main groups of sessile organisms, obscured any clear effects of competition for space with coral recruits at Eilat.

Acknowledgements

We thank the staff of the Interuniversity Institute for Marine Science for technical assistance, and the Nature Reserve Authority of Israel for permits to work in restricted areas. K. Tarnaruder and numerous volunteers assisted with the fieldwork. Comments by J. Wielgus, A. Genin, and two anonymous reviewers greatly improved the manuscript. Funding was provided by a graduate fellowship from the Faculty of Life Sciences at Bar Ilan University, an Internal Grant from the Research Authority of Bar Ilan University, and a grant from USAID-MERC through the Red Sea Marine Peace Park Program.

Copyright information

© Springer-Verlag 2003