Butterflyfishes of the Southern Red Sea : Ecology and Population Dynamics

Transcription

1 RIJKSUNIVERSITEIT GRONINGEN Butterflyfishes of the Southern Red Sea : Ecology and Population Dynamics PROEFSCHRIFT ter verkrijging van het doctoraat in de Wiskunde en Natuurwetenschappen aan de Rijksuniversiteit Groningen op gezag van de Rector Magnificus, dr. F. Zwarts, in het openbaar te verdedigen op maandag 8 september 2003 om 16:00 uur door Zekeria Abdulkerim Zekeria geboren op 17 juni 1963 te Asmara, Eritrea

2 Promotor Prof. dr. J.J. Videler Beoordelingscommissie Prof. dr. R.P.M. Bak Prof. dr. R.F.G. Ormond Prof. dr. W.J. Wolff

3

4 Cover Printed by picture of Chaetodon larvatus Richard and Mary Field Grafisch Centrum Rijksuniversiteit Groningen Blauwborgje AC Groningen The Netherlands ISBN X The field work for the research was carried out at the University of Asmara, Massawa Marine Sciences Field Station, Massawa, Eritrea. The thesis was prepared at the University of Groningen, Department of Marine Biology, P.O.Box 14, 9750 AA, Haren, The Netherlands. The study was supported by the Netherlands Organization for International Cooperation in Higher Education (NUFFIC) as part of the Institutional and Human Resource Development in Marine Sciences and Fisheries project at the University of Asmara (MHO-UoA-MBF ERI/618 and ERI/619).

5 Table of Contents Acknowledgements Chapter 1 General Introduction and Outline of the Thesis 1 Chapter 2 The Distribution Patterns of Red Sea Chaetodontid 3 Assemblages Chapter 3 Correlation between Abundance of Butterflyfishes 13 and Coral Communities in the Southern Red Sea Chapter 4 Resource Partitioning among Four Butterflyfish species 21 Chapter 5 Territorial and Feeding Behaviour of the Brownface 43 Butterflyfish (Chaetodon larvatus) Chapter 6 Spawning Seasonality in Brownface Butterflyfish 53 Chapter 7 Temporal and Spatial Recruitment Patterns in 65 Chaetodontids and Pomacanthids in the Southern Red Sea Chapter8 Growth of the Brownface Butterflyfish 81 (Chaetodon larvatus) Chapter 9 Summary and Conclusions 97 References 113

6

7 Acknowledgements This thesis would not have come into existence with out the valuable help from various people and organizations. It is impossible to mention them all. I would like to thank all the of them who have helped me during the study and during the production of the thesis. Those of you who are not mentioned here please understand the problem. I want to thank my promotor, Prof. dr. John Videler, who has been supportive throughout the study. He was always on my side whenever I needed help. I spent many hours in his office and in his home discussing my study and he was always supportive. His critical thinking and wide experience has helped me in look at my results from different angles. John was more than a promotor for me. The hospitality of John and his wife, Hanneke, made my life in Groningen enjoyable. They made it a habit to invite me over dinners which was helpful in relieving me from the stress of postgraduate studies. The people of the marine zoology lab, especially Jos de Wiljes and Eize Stamhuis, were helpful in a number of practical aspects. They helped me find my way in the lab and facilitated retrieval of laboratory equipment. Mr. Dick Visser prepared the figure of the study site. The photograph of the brownface butterflyfish on the cover page was posted with the kind permission of Mr. Richard and Mrs. Mary Field. I want to thank the Netherlands Organization for International Cooperation in Higher Education (NUFFIC) for sponsoring my study. The Office of International Relations of the University of Groningen arranged my travel between Asmara and Groningen and made official arrangements for my stay in the Netherlands. I am indebted to the staff of the Office of International Relations for their help. The field work was conducted in the Eritrean Red Sea coast. During the field work I was assisted by number of individuals and organizations. The University of Asmara allowed me to make full use of the facilities of the Marine Sciences Field Station in Massawa. I also enjoyed free accommodation in the guesthouse of the university during my stay there. Mebrahtu Ateweberhan shared the field station and guesthouse facilities with me. He was the only person for me to turn to when ever I was faced with a practical and theoretical problem during my stay in Massawa. The discussions I had with him while sipping cappuccino in Mobile Mart helped me clear my thoughts. At the initial stage of my project I benefited from practical advise of Dr. Henrich Bruggemann, who worked very hard to make the field station a functional research unit.

8 I am indebted to Ato Hassan Mohammed Ali who provided professional service as a skipper of the field station boat. He was my buddy during diving and he helped me in collecting fishes. Many students from the Department of Marine Biology and Fisheries helped me in collecting data from the field and in the laboratory. I would like to thank Dawit Yemane, Ghebremedhin Sibahtu, Naser Mohammed, Dawit Berhe, Issam Yasin, Biniam Samuel, and Tesfamichael Kaleab for their assistance in the field. Sven Weertmen, a graduate student at the University of Groningen, also helped me in polishing otoliths in the lab. I would like to thank the Eritrean Diving Center and the Research Division of the Ministry of Fisheries for the technical support they provided me with at the beginning of the study. The efforts of all these people and organazations were invaluable to the study. Thus, I want to thank them all for their assistance. I want to thank the following professors for their help. Prof. Rupert Ormond, the Director of the University of Marine Biological Station Millport, and Prof. Ernst Reese, University of Hawaii, sent me a number of their publications and my correspondences with them had improved the quality of my work. Prof. J.H. Choat, James Cook University, provided me with practical advice on the techniques of polishing otoliths. Last but not least, I would like to thank my family for their moral support and encouragement. My father, my late uncle Abdulkader Zekeria, my brothers, and my sisters were of great help in this regard. My special thanks goes to my wife, Huda Saleh, who took full responsibility of our newly established family during my absence. She raised our first child, Aymen, while I was away from Eritrea for the study.

9 Chapter 1 Butterflyfishes of the Southern Red Sea 3 General Introduction and Outline of the Thesis

10

11 Butterflyfishes of the Southern Red Sea 5 INTRODUCTION Butterflyfishes (Chaetodontidae) are small, deep-bodied and colourful fishes. They are mostly diurnal feeders on coral polyps and on other invertebrates (Findley & Findley 2001). Chaetodontids are easily recognizable, site attached, and inhabitants of shallow tropical oceans. Their feeding and social behaviour can be investigated by following marked individuals. As a result, chaetodontids are among the most studied families of coral reef fishes. Distribution patterns, feeding habits, resource partitioning and social behaviour are the main aspects studied. Despite extensive studies on the ecology of butterflyfishes, little is known about their population dynamics. Data on the behaviour and timing of spawning are available for a few species only (Ralston 1981, Lobel 1989a, 1989b, Tricas & Hiramoto 1989, Yabuta 1997, Yabuta & Kawashima 1997); information on growth is limited to juveniles of a few species (Fowler 1989); and reports on recruitment are scarce (Walsh 1987). Part A of this chapter provides a summary of the current state of knowledge of the ecology and population dynamics of butterflyfishes. Global distribution, the effect of coral cover on local abundance, feeding habits, territorial behaviour and social systems are discussed. The life history characteristics of butterflyfishes are summarized. Part B summarizes the physical and biological features of the Red Sea. Part C presents the objectives of the study and an outline of the thesis. A. Butterflyfish Biology 1. Distribution and abundance Globally 116 chaetodontid species are recognized. The highest diversity of butterflyfishes is recorded from the Indo-Pacific but the diversity decreases as one moves away from this region. Global patterns of distribution may be determined by the evolutionary history of tropical reefs but local distributions may depend on the coral cover. Areas with high coral cover are inhabited by more butterflyfishes than regions with lower cover (Bell & Galzin 1984, Bouchon-Navaro et al 1986). However, the abundance of some chaetodontid species is probably related to the distribution of specific coral genera and not connected with total coral cover. For example, in the East Indo-Pacific the abundance of Chaetodon trifascialis was related with the cover of Acropora coral. This fish feeds specifically on Acropora and it is hypothesized that the fish and the coral species co-evolved for many years (Reese 1977). Based on this observation Reese (1977) proposed butterflyfishes as

12 6 Chapter 1. Introduction and Outline of the Thesis indicators of reef health. However, some workers questioned this hypothesis (Roberts et al 1988). II. Feeding, territorial and social habits Chaetodontids live close to the substrate and feed diurnally. Roberts and Ormond (1992) identified five feeding categories among butterflyfishes. These categories are: Obligate hard coral feeders; Feeders on sessile and sedentary invertebrates (including some amount of coral polyp); Feeders on motile benthic invertebrates; Generalist omnivores (opportunistic feeders on broad range of food including algae); and Planktivores (feeding primarily on zooplankton). The majority of the butterflyfishes belong to the first two dietary categories (Reese 1977, Anderson et al 1981, Harmelin-Vivien & Bouchon-Navaro 1982, Harmelin-Vivien & Bouchon-Navaro 1983). Only few species are known to feed on motile benthic invertebrates or on zooplankton (Hobson 1978, Harmelin-Vivien & Bouchon- Navaro 1982). The feeding habits of butterflyfishes vary among geographical regions. In the Great Barrier Reef about 80% of the butterflyfish species are corallivores. In the western Indian Ocean, the proportion is 72% while in Hawaii it is less than 60%. In contrast, the highest percentage of zooplankton feeders was reported from Hawaii (Sano 1989). Hourigan (1989) attributed the unusually large proportion of planktivores in Hawaii to the absence of many other planktivores from the area. Butterflyfishes occur as solitary individuals, as pairs, or in small groups (Reese 1977, Burgess 1978, Allen et al 1998). Food resources appear to be the major determinants of different social and mating systems among butterflyfishes. Corallivorous fishes are predominantly pair forming fishes while planktivores usually live in schools (Hourigan 1989). Plankton occurs in patches of varying density around which schools of plaktivorous fish aggregate. Corals, on the other hand, are fixed and predictable in their distribution and can be exhausted if preyed upon intensively. III. Life history characteristics Butterflyfishes are broadcast spawners releasing eggs into the water column. The pelagic eggs are externally fertilized and are dispersed by currents. Females typically spawn thousands to hundreds of thousands of eggs at a time. Spawning may take place as often as every two days or once or twice per month. However, usually spawning is seasonal and the spawning season extends for about four months (Thresher 1984).

13 Butterflyfishes of the Southern Red Sea 7 Embryos hatch about 30 hrs after spawning (Suzuki et al 1980) and the larvae spend an average of 40 days in the plankton before metamorphosing and settling on reefs (Hourigan and Reese 1987). A life history characterized by pelagic eggs and larvae implies absence of parental care for young and eggs. The advantage of this type of life history is the wide dispersal of the eggs and larvae. However, pelagic spawning results in high larval and egg mortality. The loss of eggs and larvae of pelagic spawners is usually compensated by high female fecundity. Juvenile chaetodontids grow very fast but attain maturity at relatively large size. Compared with pelagic fishes most coral reef species have longer life spans. Many species have extended spawning seasons where large clutches are spawned at intervals over extended breeding periods. The eggs hatch after about 30 hours and develop into pelagic larvae. The larvae spend about 40 days in the water column and end the planktonic life stage by settling on a coral reef as recruits. Butterflyfishes reach 70-75% of their maximum size and attain maturity at the age of one-year (Ralston 1976, Tricas 1986). Young fish are subject to heavy predation but the predation pressure on adults is relatively low (Reese 1977). The compressed bodies of adult fishes and the presence of sharp spines on the dorsal and anal fins discourage predators from attacking adult butterflyfishes. There is little information on the longevity of butterflyfishes. The highest longevity recorded was from aquarium fishes where an individual was known to have lived for 25 years (Allen et al 1998). The longevity recorded from natural habitat is much lower. For example, marked pairs of C. paucifasciatus lived for at least six years in the Northern Red Sea (Fricke 1986). Reese (1981) observed pairs that lived together for ten years. Compared with longevity estimates for other coral reef fishes, butterflyfish life is much shorter. For example, surgeonfish and parrotfish are known to live for 35 and 50 years respectively (Choat et al 1996). B. Physical and biological features of the Red Sea The Red Sea is a long, narrow body of water situated between Northeast Africa and the Arabian Peninsula. The sea extends for about 2000 km in Northeast Southwest direction and is connected to the Indian Ocean by a narrow and shallow sill at Bab-el-Mendab. The Suez Canal, which was built in 1869, connects the Red Sea with the Mediterranean Sea in the North. The Bab-el-Mendab and the Suez Canal allow only limited exchange of surface water between the Red Sea and the neighbouring water bodies. Geologically the Red Sea lies between the African and Arabian plates and is essentially a product of their divergence. Crustal sagging is believed to have started

14 8 Chapter 1. Introduction and Outline of the Thesis about 180 million years ago but the sea was established as a linear trough about 38 million years ago. The Red Sea is the place where the earth s largest geological feature, the mid-ocean rift system, strikes the continental platform, and splits it. The Red Sea rift has been separating Arabia from Africa for about 70 million years. Widening of the fault paused in the first half of the Tertiary and recommenced between 2 to 5 million years ago (Edwards & Head 1987). The climate of the Red Sea is largely controlled by the distribution of atmospheric pressure and its changes over a vast area. The pressure centres involved are generally distant from the Red Sea and vary during the course of the year (Edwards 1987). The pressure distributions undergo widespread and sometimes drastic seasonal changes over extensive areas. However, the effects in the Red Sea are small. The weather characteristics over the whole Red Sea basin show a remarkable uniformity throughout the year, with some variation due to the quite large range of latitude. As pointed out by Edwards (1987) this could be attributed to two main factors. Firstly, almost all the air that enters the Red Sea is dry although it may come from different directions. The Red Sea lies within the belt of the Northeast Trade Winds, which forms the basis of much of the airflow to the sea. Further, the surrounding desert and semi-desert areas contribute to the dryness. Secondly, mountain ranges along the side of the Red Sea ensure that the main wind systems blow predominantly along the length of the sea, with only localized air movement at right angles to the shoreline. As a result, the prevailing wind directions are remarkably constant and there is virtually no exchange of air masses with different properties, which might give rise to changeable conditions or spatial variability in the weather. In the south, measurements near Massawa show the lowest sea surface temperature in February, with values of 25 C. After February, the surface temperature increases gradually to reach a maximum value of 32 C in September. From October to January, temperature declines by about 1 C per month. In the central Red Sea, at about 18 N, a sea surface temperature of 20 C was recorded in February while temperature values above 30 C occurred during the summer months. There is strong evidence that the production potential of the Red Sea is low. Over most of the basin, thermoclines and haloclines prevent the cycling of nutrients from deeper water to the euphotic zone. There is little nutrient input to the pelagic system from land surface runoff to compensate for the steady loss by sinking of nutrients out of the productive zone. On this basis, productivity can be expected to be low over most of the central Red Sea. Production increases somewhat to the north and south where mixing processes are known to occur. Among these, the interoceanic water exchange via the straits of Bab-el-Mendab is the main process. This

15 Butterflyfishes of the Southern Red Sea 9 exchange is most intense in winter when the plankton-rich Gulf of Aden water flows into the Red Sea at the surface, counterbalanced by an outflow of Red Sea deep water over the Hanish sill. Seasonal changes of the primary production of the Red Sea are not well studied. The limited studies conducted in the area suggest a relatively higher production in the summer months (Ponomareva 1968). In the southern Red Sea a secondary peak in primary production and phytoplankton standing stock was observed in winter. The effect of summer eutrophication on the plankton biota may, however, be confined to the coastal and the northern and southern ends of the Red Sea. The vast oceanic region between 27 N and 18 N is less affected by seasonal changes. Indeed the most substantial import of phosphate into the Red Sea occurs by subsurface inflow of Gulf of Aden water from July to September. The mass development of blue-green algae provides further evidence of depletion of plant nutrients in oceanic waters during the summer season. The distribution of corals in the Red Sea is not well known. Some information is available for the Gulf of Aqaba and for some regions along the northern and central coasts of the Red Sea (Bouchon-Navaro 1980, Loya 1972, Roberts et al 1992). The available data suggest that the most common reef type in the area is fringing reef. The reefs in the north and central coasts are up to 40 meters deep and seven depth zones can be distinguished (Roberts et al 1992). In contrast, in the southern Red Sea, the depth of the reef is limited to about 10 m and the reefs are probably less developed. Only four depth zones can be recognized (Roberts et al 1992). Information regarding the fish communities of the Red Sea is also limited. About 1200 species of fishes are known to occur in the Red Sea (Goren 1984, Ormond & Edwards 1987). The majority of these inhabit coral reefs where they constitute a dominant component of the fish fauna. There are marked differences among the different regions of the Red Sea in fish species richness, assemblage compositions and species abundance (Sheppard et al 1992). C. Objectives of the study and outline of the thesis I. Objectives of the study The Chaetodontidae are represented by 14 species in the Red Sea and the Gulf of Aden of which seven are endemic (Randall 1983). The Gulf of Aqaba is the most studied part of the Red Sea as far as the ecology of butterflyfishes is concerned. A team of French scientists investigated the feeding (Harmelin-Vivien & Bouchon- Navaro 1982), distribution patterns (Bouchon-Navaro 1980), and resource

16 10 Chapter 1. Introduction and Outline of the Thesis partitioning (Bouchon-Navaro 1989). Fricke (1986) studied the social habits and territorial behaviour. Gharaibeh and Hulings (1990) investigated reproductive seasonality of three species. The butterflyfishes of the northern Red Sea, especially those off the Egyptian and Saudi coasts, have been the subjects of studies by British scientists. The distribution patterns (Roberts & Ormond 1987, Roberts et al 1988, Roberts et al 1992, Righton et al 1996), feeding habits (Ormond 1972) and territorial behaviour (Roberts & Ormond 1992, Wrathall et al 1992, Righton et al 1998) were investigated. Roberts et al (1992) surveyed 367 sites and investigated the large-scale variation in assemblage structure of chaetodontids and pomacanthids. The study sites were spread throughout most of the Red Sea. Between 6 and 47 sites were visited at every degree latitude from 29 N down to 16 N. Moreover, 20 sites were surveyed in the Gulf of Aden at 12 N. The southern Red, between 16 N and 12 N, was not surveyed during this investigation. The Sudanese, Eritrean and Yemeni coasts of the Red Sea were not included in the study. To date there is no information on the ecology of butterflyfishes from these areas. Roberts et al (1992) suggest the occurrence of marked differences in the assemblage patterns of chaetodontids between the southern Red Sea and the northern and central parts. Most of the information comes from surveys in the Northern Red Sea and the Gulf of Aqaba. Investigations on feeding, territoriality and social habits were conducted on C. austriacus and C. paucifasciatus, the dominant chaetodontids in the northern Red Sea. Both species are not reported from the south. On the other hand, nothing is known about the behaviour of C. larvatus, the most dominant chaetodontid in the southern Red Sea. The objective of this thesis is to assess the ecology and population dynamics of chaetodontids in the southern Red Sea. Due to logistic reasons, the fieldwork was carried out mainly on the reefs that are found near Massawa on the Eritrean coast. Distribution patterns were investigated on the reefs around Massawa and in the Dahlak archipelago. Results form this study showed that Chaetodon larvatus is the dominant species in the area. Hence, the species was selected for detailed ecological and life history studies. Distribution patterns, feeding habit, territorial behaviour, social systems and population dynamics of C. larvatus are investigated. II. Outline of the thesis Chapter 2 presents results of surveys of chaetodontid distribution from three regions along the Eritrean Red Sea coast. The data collected are compared with the

17 Butterflyfishes of the Southern Red Sea 11 regional distribution of butterflyfishes in the Red Sea obtained from the literature. The combined information is assessed and the biogeography of butterflyfishes in the Red Sea and the Gulf of Aden is discussed. In Chapter 3, the local distribution of chaetodontids is assessed for the central Eritrean coast. Abundance and coral cover data are presented for seventeen sites located near Massawa and in the Dahlak Archipelago. The correlation between fish abundance and coral cover is analysed and the possibility to use the abundance of butterflyfishes as indicators of reef health is discussed. Feeding habits, social systems and spacing of four common chaetodontid species are compared in Chapter 4. These species are C. larvatus, C. semilarvatus, C. mesoleucos and Heniochus intermedius. Diets are analysed by comparing stomach contents. Feeding rates, territorial behaviour, and social grouping of the four species are investigated and compared. Detailed investigation of social and feeding behaviour of C. larvatus is presented in chapter 5. Relationships between feeding rates, territory size and live coral cover are assessed. Food preference is determined by comparing the proportion of corals with the selection consumed by the fish. A comparison is made between the behaviour of C. larvatus and that of C. austriacus, a territorial corallivore in the northern Red Sea (Wrathall 1992, Righton et al 1996). In chapter 6 three approaches are used to determine the spawning seasonality of C. larvatus. In the first approach, changes in gonad histology is monitored. Egg development provides an indication of the time of spawning. In the second approach, changes in relative gonad mass were monitored by comparing monthly gonadosomatic values over a two years period. In the third approach, field observations were used to determine the spawning period. Recruitment patterns of C. larvatus, C. semilarvatus and Pomacanthus spp. are the subject of chapter 7. Field data were collected from the reefs around Massawa. Seasonal and inter-annual variations in recruitment are compared by analysing data collected over a four-year period. The field data set is also used to compare spatial aspects of the recruitment patterns. Growth patterns of C. larvatus are investigated in chapter 8. Two independent methods are used. The first method employs readings of growth rings of fish otoliths. Length changes of recruits in the field are monitored as the second method to study the growth of populations of young fish. Finally, in chapter 9 results from the whole study are summarized and conclusions drawn.

18

19 Chapter 2 Butterflyfishes of the Southern Red Sea 13 The Distribution Patterns of Red Sea Chaetodontid Assemblages Zekeria Z.A., Yohannes A.T. & Videler, J.J. Submitted to Aquatic Conservation

20 14 Chapter 2. Regional Distribution Abstract The occurrence and abundance of butterflyfishes was investigated in northern, central and southern areas of the Eritrean Red Sea coast. Visual census was used to estimate the presence and abundance of the species along 100-meter long transects. The assemblages of butterflyfishes of the three areas differ markedly. Two species are restricted to the north, while three others occur only in the south. The central area contains six species occurring in both the south and the north. The results show that the northern area is more related to the central and northern regions of the Red Sea while the southern area has many species in common with the Gulf of Aden. The areas sampled are considered to represent different biogeographic regions within the Red Sea. A barrier has been suggested to exist in the Red Sea, which prevents dispersal between the northern and southern regions. Our results indicate that this barrier is situated in the Eritrean central area. The nature of this barrier remains unclear.

21 Butterflyfishes of the Southern Red Sea 15 INTRODUCTION In their family-wide systematic revision, Allen et al (1998) recognized 116 chaetodontid species arranged in 10 genera. Nearly all of these inhabit coral reefs in tropical seas around the world. Butterflyfish species richness is greatest in the central Indo-West Pacific region where at least 40 species are recorded. Species richness generally declines with increase in distance from the centre of this region. Eastwards across the equatorial Pacific numbers drop gradually to 28 in the Society Islands and 14 in Marquesas. In the tropical east Pacific only 4 species have been reported and in the Caribbean 6 species are known. Proceeding westwards from the Indo-West Pacific the species richness declines reaching 27 in the western Indian Ocean, 22 in the Seychelles and 14 in the Red Sea (Findley & Findley 2001). Seven of these are endemic to the Red Sea and the Gulf of Aden (Randall 1983). The difference between the western Indian Ocean and the Red Sea chaetodontid assemblages is attributed to the existence of biogeographic barriers in the southern Red Sea and the northwestern Indian Ocean preventing the dispersal of endemics (Ormond & Edwards 1987, Blum 1989). In the Red Sea, Bouchon-Navaro (1986), Roberts & Ormond (1987), Roberts et al (1992), Righton et al (1996), and Righton et al (1998) observed regional differences in the distribution of butterflyfishes. Roberts et al (1992) conducted a detailed investigation on the large-scale distribution of chaetodontids. They grouped the species into four categories: Endemic to the Red Sea: found in the northern and central Red Sea only (Chaetodon austriacus and C. paucifasciatus) Endemic to the Red Sea region also occurring in the Gulf of Aden and sometimes in the Arabian Gulf (C. larvatus, C. fasciatus, C. semilarvatus, C. mesoleucos, and Heniochus intermedius) Present in the Gulf of Aden but not in the Red Sea (C. melapterus and C. vagabundus). Occurring throughout the Indian Ocean and the whole (or major parts) of the Red Sea (C. auriga, C. lineolatus, C. melannotus, C. trifascialis, and H. diphreutes). Roberts et al (1992) surveyed 345 locations extending from 29ºN in the Gulf of Aqaba to 16ºN on the Saudi Arabian coast. The Eritrean and Yemeni coasts were not included in the survey. We are not aware of any report on the distribution of butterflyfishes in the Red Sea south of 16ºN. This part of the Red Sea is about 2000

22 16 Chapter 2. Regional Distribution km long and extends from 18ºN to 11ºN. We will refer to the Red Sea north of Eritrea (>18ºN) as the northern and central Red Sea. The objective of the present work is to assess the distribution patterns of butterflyfishes in the southern Red Sea, along the Eritrean coast. The occurrence and abundance of butterflyfish species was investigated in three selected areas. The data collected are compared with published distribution patterns of the species concerned. The biogeography of butterflyfishes of the southern Red Sea is discussed. MATERIALS AND METHODS Study site The study was conducted in the southern Red Sea along the Eritrean coast (Fig. 2.1). Three areas along the coast were surveyed. The northern area is located at about 18ºN and is found near Hasmat. The central area is around Massawa and in the Dahlak Archipelago and is located at about 16º30'N. The southern area is located at about 13ºN and is near the port city of Asseb. At least ten sites were selected in each area to estimate the presence and abundance of butterflyfish species. Figure 2.1. Study site.

23 Butterflyfishes of the Southern Red Sea 17 Presence and abundance data The presence of chaetodontid species was determined by surveying each site for at least 15 minutes while snorkelling or diving close to the bottom. Each butterflyfish species encountered was recorded on a PVC slate. Randall (1983) and Lieske & Myers (1994) where used for fish identification. A belt transect method was used to estimate the abundance of the butterflyfishes (Crosby and Reese 1996). At each site three 100m long transects were laid parallel to the shoreline. Visual census was conducted by slowly swimming or diving along the transects and counting all the butterflyfishes observed within 2.5m distance on either side of the line. Three abundance categories were used: Abundant (>10 individuals m -2 ); Common (10-5 individuals m -2 ); Rare (<5 individuals m -2 ). RESULTS AND DISCUSSION Distribution along the Eritrean Coast Eleven butterflyfish species are recorded from the Eritrean coast (Table 2.1). Species richness varies among the three areas studied. There are eight species in the northern area of which one, C. fasciatus is abundant. The densities of the other seven species are very low. The central area contains only six species. Four of these, C. larvatus, C. semilarvatus C. mesoleucos and H. intermedius, are abundant. C. fasciatus is common and C. lineolatus is rare. Nine species occur in the south. Three of these, C. larvatus, and H. intermedius are abundant, five species are rare and H. acuminatus is common. The chaetodontid assemblage varies among the three areas. C. auriga and C. trifascialis are found in the northern area only. Three species, C. vagabundus, C. melapterus and H. acuminatus, occur only in the south. The remaining six species are found in all three areas but the abundance varies. Roberts et al (1992) regard C. melapterus and C. vagabundus as Indian Ocean species present in the Gulf of Aden but not in the Red Sea. Randall (1983) did also not consider these as Red Sea species. However, in the more recent field guides of Debelius (1993) and Lieske & Myers (1994) both species are reported from the southern Red Sea. H. acuminatus, has never been reported from the Red Sea before. According to Steene (1978), the distribution of H. acuminatus extends from East Africa and the Persian Gulf in the west to the Society Islands in the east and from southern Japan in the north to the Lord Howe Island in the South.

24 18 Chapter 2. Regional Distribution Distribution in the Red Sea and the Western Gulf of Aden The distribution figures from the present work are compared with distribution data obtained from the literature in Table 2.2 The table shows that six species, C. lineolatus, C. fasciatus, H. intermedius, C. larvatus, C. semilarvatus, and C. mesoleucos, are widely distributed in the Red Sea and in the Western Gulf of Aden. The last five species are endemic to the Red Sea and the Gulf of Aden but C. lineolatus occurs in the Red Sea and in the whole Western Indian Ocean (Randall 1983). An unusual distribution is noted for C. trifascialis, which is recorded from the northern Red Sea region (including the northern Eritrean coast) and the Western Gulf of Aden but not from the central and southern Red Sea areas. Our results show that C. auriga occurs in the northern Eritrean area but not in the central and southern areas. This species is also found in the northern and central Red Sea region. H. diphreutes occurs in the northern and central Red Sea but not along the Eritrean coast and in the western Gulf of Aden. C. melannotus is present in the northern and central Red Sea and in the western Gulf of Aden. H. diphreutes and C. melannotus are widely distributed in the Indian ocean (Roberts et al 1992). Table 2.1. Distribution data of butterflyfishes along the Eritrean Red Sea coast. A, Abundant (>10 individuals 1000 m -2 ); C, Common (10-5 individuals 1000 m -2 ); R, Rare (<5 individuals 1000 m -2 ). Species Northern Central area Southern area Chaetodon auriga R C. trifascialis R C. larvatus R A A C. semilarvatus R A A C. mesoleucos R A R Heniochus intermedius R A A C. fasciatus A C R C. lineolatus R R R C. melapterus R C.vagabundus H. acuminatus C The northern and central Red Sea region contains two endemic species, C. paucifasciatus and C. austriacus. The southern area of the Eritrean coast and the Gulf of Aden share two species, C. vagabundus and C. melapterus, which are not recorded from more northern parts of the Red Sea. Both species occur in the western Indian Ocean. There are no exclusive species in the central area of the Eritrean coast. R

25 Butterflyfishes of the Southern Red Sea 19 These results suggest that the chaetodontid assemblage of the northern Eritrean coast is more related to that of the northern and central Red Sea region. Hence, the northern Eritrean coast could be regarded as a continuation of the northern and central Red Sea. The chaetodontid assemblage of the southern area is unique because it has three species, which are not recorded, from the central and northern areas. This area probably represents a separate sub-region within the Red Sea. Its assemblage is similar to that of the western Gulf of Aden and deviates substantially from the more northern assemblages. The central Eritrean area forms a transitional region between the southern and the northern Eritrean coastal areas. Klausewitz (1972), Briggs (1974), and Ormond and Edwards (1987) suggested the presence of an ecological barrier or a region of vicariance between the southern and central areas of the Eritrean coast. Our data provide a more precise indication of its position at about 16ºN. Table 2.2. Distribution of chaetodontid species in the northern and central Red Sea (NCRS), along the Eritrean coast, and in the Western Gulf of Aden (WGA). Northern Species NCRS 1 area 2 (18 N) Eritrean coast Central area 2 (16 N) Southern area 2 (13 N) WGA 3 Chaetodon paucifasciatus C. austriacus X X C. melannotus X X Heniochus diphreutes C. auriga X X X C. trifascialis X X X C. lineolatus X X X X X C. fasciatus X X X X X H. intermedius X X X X X C. larvatus X X X X X C. semilarvatus X X X X X C. mesoleucos X X X X X C. melapterus X X C. vegabundus X X H. acuminatus X 1 Roberts et al (1992) 2 Present study 3 Kemp (1998)

26 20 Chapter 2. Regional Distribution Roberts et al (1992) attributed the ecological barrier to the sparsely distributed, poorly developed and turbid nature of the southern Red Sea reefs that appear to restrict the distribution of the northern and central endemics, C. austriacus and C. paucifasciatus. The same barrier could also restrict the distribution of C. vagabundus, C. melapterus and H. acuminatus from spreading to the north. The barrier does not seem to affect the distribution of the more widely distributed species. These species might have a prolonged pelagic larval life facilitating longrange dispersal. Further investigation is needed to discover the nature of the barrier and its differential effect on fish dispersal.

27 Chapter 3 Butterflyfishes of the Southern Red Sea 21 Correlation between the Abundance of Butterflyfishes and Coral Communities in the Southern Red Sea Zekeria Z.A. & Videler, J.J. Published in the Proceedings of the 9 th International Coral Reef Symposium, Bali, Oct , 2000, Volume I, PP

28 22 Chapter 3. Local Distribution Abstract Relationships between some substrate parameters and the abundance of butterflyfishes were investigated across seventeen reefs around Massawa and in the Dahlak Archipelago in the southern Red Sea. Visual censuses of butterflyfishes were conducted along 100 m belt transects and the nature of the substrate was investigated using a quadrat method. Five chaetodontid species and fifteen scleractinian coral genera were found in the study sites. The surveyed reefs had different proportions of live coral cover. The abundance of Chaetodon larvatus, C. semilarvatus and C. mesoleucos showed significant correlation with live coral cover. However, the relationships between coral cover and the abundance C. fasciatus and Heniochus intermedius were not significant. C. larvatus, C. semilarvatus and C. fasciatus are corallivores while C. mesoleucos and H. intermedius feed mainly on non-coralline invertebrates. Substrate rugosity did not show correlation with the abundance of the fish species. The results suggest the existence of strong links between corallivorous chaetodontids and the cover of scleractinian corals.

29 Butterflyfishes of the Southern Red Sea 23 INTRODUCTION Fourteen chaetodontid species are recorded from the Red Sea and the Gulf of Aden of which seven are endemic to the region (Randall 1983). Roberts et al (1992), Righton et al (1996) and Kemp (1998) investigated the large-scale distribution patterns of these species. They showed that there is marked variation among the northern, central and southern parts of the Red Sea (See Chapter 2). Roberts & Ormond (1987), Bouchon-Navaro & Bouchon (1989), and Roberts et al (1988) reported within-habitat-variation in the distribution of chaetodontids. There is fairly good knowledge of the regional distribution of the butterflyfishes in most parts of the Red Sea but little is known about the local distribution of the species in the southern part. The structure of fish communities can be influenced by the physical complexity of the substrate (Luckhurst & Luckhurst 1978, Roberts & Ormond 1987) and by the amount of coral cover (Bouchon-Navaro et al 1986, and Ohman et al 1998). Increased substrate complexity provides a greater diversity of shelter and feeding sites, thus enhancing species richness (Bell & Galzin 1984). Although many fishes do not depend directly on live coral for food, a number of chaetodontids are either obligate or facultative coral feeders. Reese (1977) first proposed that obligate corallivores, such as butterflyfishes, could serve as indicator organisms. This butterflyfish bio-indicator hypothesis led a number of reef-fish ecologists to assess the relationship between the abundance of chaetodontids and the cover of live corals. Bell & Galzin (1984), Bouchon-Navaro et al (1986), Findley & Findley (1985), Sano et al (1984), Bouchon-Navaro & Bouchon (1989), and Ohman et al (1998) found significant correlation between the scleractinian coral cover and chaetodontid abundance and diversity. On the other hand, Luckhurst & Luckhurst (1978), Bell et al (1986), and Roberts & Ormond (1987) found weak or no correlation between the two parameters. Most of the above studies showed significant correlation between the abundance of chaetodontids and the cover of live corals. However, whether these fishes can serve as indicators of reef health has become a subject of debate. For example, Roberts et al (1988) argued that butterflyfish fish abundance could not be used to compare the health of corals from different sites. Their argument was based on the observation that pristine reefs may naturally have low live coral cover. Reese & Crosby (1998) noted that the indicator species concept is more useful, sensitive and accurate when within site comparisons are made over time.

30 24 Chapter 3. Local Distribution The objective of this study is to investigate the distribution patterns of chaetodontids and to assess the influence of live coral cover and substrate rugosity on their abundance in the southern Red Sea. The results will also be used as arguments in the discussion about the reef health indicator hypothesis. Materials and Methods Substrate and butterflyfish distributions were surveyed at seventeen sites in the southern Red Sea in March, April and August The sampling sites, their locations, and the depths of the surveys are listed in Table 3.1. Ten sites were located around Massawa and seven in the Dahlak Archipelago. Most sites are covered by fringing reefs that extend to maximum depths of 8-10 meters. In thirteen sites the surveys were conducted at two meter depth; in four sites deeper zones were selected. A belt transect method was used to estimate the abundance of the butterflyfishes (Crosby & Reese 1996). At each site, three 100 m long lines were laid parallel to the shoreline. Visual census of fishes was conducted by slowly swimming or diving along a line and counting all the butterflyfishes observed within 2.5m distance on either side of the line. Crosby & Reese (1996) used 10 m wide transects but poor visibility and the narrowness of many surveyed reefs forced us to reduce the transect width to 5 m. The fishes were identified to species level and their abundance recorded on a PVC slate. Data collection took place between 09:00 and 13:00 hrs. At each site one of the line transects was selected for the determination of surface rugosity and estimation of percentage bottom cover. A stratified sampling strategy was used to study the nature of the substrate. The 100 m transect was divided into ten, 10 m line segments and from each segment two points were randomly selected for the estimation of bottom cover. A one-meter square quadrat was laid on each random point and substrate type and cover within the quadrat was then visually estimated (English et al 1994). Substrate was classified into seven types: scleractinian corals, soft corals, other invertebrates, macro-algae, dead corals, coral rubble, and sand. Scleractinian corals were further identified to genus level. Corals were grouped in one of five-growth forms following English et al (1994): massive, sub-massive, encrusting, branching and solitary. Surface rugosity was determined using the chain and tape method described by McCormick (1994). A 50 m long leaded rope was laid directly under the line transect and was made to confirm to all contours as closely as possible. A measure of surface rugosity was the ratio of linear distance covered by the rope to its actual length.

31 Butterflyfishes of the Southern Red Sea 25 Table 3.1. List of the study sites: Their geographical positions and depths. Site Name Latitude ( o N) Longitude ( o E) Depth (m) I Gurgussum II Agip depot III Sheik Said Island east (Shallow) IV Schuma Island (shallow) V Durgella Island VI Kutmia VII Dahlak Hotel (Massawa) VIII Dissie Island (shallow) IX Twalot reef X Durgaam Island XI Schuma Island (deep) XII Dissie Island (deep) XIII Sheik Said Island east (deep) XIV Sheik Said Island west XV Madot Island XVI Resimedri reef (shallow) XVII Resimedri reef (deep) Data analysis The Shannon & Weaver index and the Pielou evenness index (Pielou 1966) were computed to characterize the diversity of the fish and coral communities. The Spearman correlation coefficient was used to investigate the relationships between the abundance of fish species and coral community parameters. The following coral community parameters were tested: The number of fish species (fish species richness); the number of coral genera (coral generic richness); the density of all butterflyfishes (total fish density) and the density of corallivorous species (corallivore density). RESULTS Fish distribution and abundance Five butterflyfish species were counted in 17 sites. The abundances differed significantly (table 3.2). The most abundant species, Chaetodon larvatus, was present at all sites. It had high densities in sites XIII, XIV, XVI and XVII, with values exceeding 38 individuals per 500 m 2. It was less abundant in sites I, II, and III with less than five individuals per 500 m 2. Heniochus intermedius was also observed

32 26 Chapter 3. Local Distribution in all sites but its densities were much lower than those of C. larvatus. The highest abundances of H. intermedius were in sites X, XI and XII with densities of 11.7, 9.3, and 9.0 individuals per 500 m 2 respectively. C. mesoleucos was recorded from 11 sites and its abundance was very low except in two sites (XII and XIV with 8 and 7 individuals 500 m -2 respectively). The sites with relatively high C. semilarvatus abundance were X (5.3 per 500 m 2 ) and XV (6 per 500 m 2 ). The least abundant species, C. fasciatus, was recorded from only five sites, with a highest density of 3.3 per 500 m 2 at site X. The fish community parameters varied among the sites. The highest species richness values of five abundant species are found in sites IV, VIII, X, XI and XV. The lowest values of only two were found in sites I and III. Total butterflyfish abundance was high in sites XIV, XV and XVII where the densities were 52.3, 50.0 and 53.0 individuals per 500 m 2 respectively. The lowest density was recorded from sites I, II, and III, with mean abundance values of 2.0, 5.7 and 6.7 individuals per belt transect respectively. The value of the diversity index also varied from 1.4 at sites VIII and X to a lowest value of 0.3 at site VII. Substrate type and distribution Fifteen scleractinian coral genera were found in the study sites. Porites, Echinopora, Stylophora, Platygyra and Montipora were the dominant and widespread genera. Coral distribution varied remarkably among the sites (Table 3.2). The highest number of coral genera in one site was 11 in site XII. The number of genera recorded from sites XVII, X and XIV were 10, 8 and 8 respectively. Site II had only one genus. The percentage of coral cover also differed among sites. The scleractinian coral cover of sites XVI and XVII was high with 56.6 % and 67.3 % respectively. Sites I, II, and III had less than 10 % coral cover. The cover on the remaining sites varied from 13.3 % (IV) to 47.5 % (XV). Massive and sub-massive corals dominated most sites. Branching and solitary growth forms were present in fewer sites and their cover was much lower. Encrusting corals densely covered sites XVI and XVII.

33 Butterflyfishes of the Southern Red Sea 27 Table 3.2. Fish density (individuals 500 m -2 ), fish community parameters (species richness, total density, corallivore density, Shannon and evenness indexes), coral community parameters (live coral cover, generic richness, Shannon and evenness indexes), and substrate rugosity at 17 sites in the southern Red Sea. Fish species Site I II III IV V VI VII VIII IX X XI XII XIII XIV XV XVI XVII C. larvatus C. mesoleucus C. semilarvatus C. fasciatus H. intermedius Fish community Species richness Total fish density Corallivore density Diversity index Evenness Index Coral growth forms Massive Sub-massive Encrusting Branching Solitary Coral community Live coral cover Generic richness Diversity index Evenness Index Substrate rugosity

34 28 Chapter 3. Local Distribution Table 3.3. The correlation coefficient between fish species and coral community parameters. Significance level given at P<0.01 * and <0.001 ** Fish species Coral community parameters Coral growth forms Substrate rugosity Generic index % coral cover Diversity index Massive Sub-massive Encrusting Branching C. larvatus ** * C. mesoleucus 0.74* 0.67* C. semilarvatus * * 0.27 C. fasciatus H. intermedius Fish community Species richness Total fish density ** * 0.30 Corallivore density ** ** 0.29 Diversity index Evenness index

35 Butterflyfishes of the Southern Red Sea 29 There was no obvious variation in rugosity among the sites. The rugosity values varied from 0.6 for site VI to 0.9 for sites I, II, III and XVII. Correlation between the abundance of butterflyfishes and coral growth forms The values of the study of the correlation between the abundance of butterflyfish species and coral community parameters are summarized in Table 3.3. Significant correlation was obtained between scleractinian coral cover and the abundance of C. larvatus, C. mesoleucos, and C. semilarvatus. The correlation was very strong (P<0.001) for C. larvatus and weaker for the two other species (P<0.01). The abundance of C. mesoleucos was also correlated with the generic richness of the coral. The abundance of C. fasciatus and H. intermedius showed no correlation with any of the coral community parameters. The total chaetodontid density strongly correlated with the scleractinian coral cover (P<0.001) and with the cover of branching corals (P<0.01). Corallivore density correlated at the P<0.001 level with the percentages of both coral cover and the cover of branching corals. Abundance of C. larvatus was correlated with the cover of encrusting corals and the abundance of C. semilarvatus was correlated with branching-coral-cover (P<0.01). Fish density did not show any correlation with other coral community parameters. Abundance of C. larvatus was correlated only with the cover of Montipora out of fifteen coral genera. There was no correlation between the abundance of the other fish species and any of the individual coral genera. Butterflyfish fish species richness and diversity did not show any correlation with coral diversity and generic richness. The coral growth forms were not correlated with the fish community parameters. Neither the individual fish species nor the fish community parameters showed correlation with the substrate rugosity. DISCUSSION The five chaetodontids in this study are endemic to the Red Sea and the Gulf of Aden (Randall 1983). Roberts et al (1992) recorded all five species in the northern Red Sea. Bouchon-Navaro & Bouchon (1989) found only C. fasciatus and H. intermedius in the Gulf of Aqaba. The most abundant species near Massawa and in the Dahlak area, C. larvatus, has a relatively low density in the northern Red Sea. On the other hand, species dominant in the north, such as C. austriacus and C. paucifasciatus, are absent from the south. C. fasciatus, abundant in the northern Red Sea, has a lower abundance in the southern part. Our studies on the feeding habits show that C. larvatus and C. semilarvatus feed mainly on coral polyps while C. mesoleucos and H. intermedius forage a diverse group of mobile and sessile invertebrates including corals in the southern Red Sea (see chapter 4).

36 30 Chapter 3. Local Distribution The results in this chapter reveal that the abundance of the corallivores, C. larvatus and C. semilarvatus is correlated with live-coral cover. Among the non-corallivore fishes, the abundance of C. mesoleucos correlated also significantly with coral cover. Positive correlation between the abundance of butterflyfishes and live-coral cover was also found in the Gulf of Aqaba (Bouchon-Navaro & Bouchon 1989), in the Indian Ocean (Ohman 1998) and in French Polynesia (Bell & Galzin 1984). However, Luckhurst & Luckhurst (1978) found no correlation between the parameters in the Caribbean and Roberts & Ormond (1987) found only a weak correlation in the northern Red Sea. C. fasciatus feeds mainly on corals (Harmelin-Vivien & Bouchon-Navaro 1982) but its abundance did not correlate with the coral cover. This could be due to the low density of the fish species in the study area. On the other hand, positive correlation was observed between coral cover and the abundance of the non-corallivore C. mesoleucos. In both cases, food alone may not explain the observed correlation. Results of the present study agree with the suggestions of Crosby & Reese (1996) that C. larvatus could serve as a bio-indicator in the Red Sea. As mentioned above, C. larvatus is rare in the Gulf of Aqaba and its abundance is very low in the northern Red Sea (Roberts et al 1992). Therefore, its use as a bio-indicator should be restricted to the southern Red Sea. C. semilarvatus could also be considered as indicator species in the southern Red Sea. However, it should only be used as an indicator of reef health when comparing reefs over time (Roberts et al 1988). The correlation between the abundance of C. larvatus and C. semilarvatus with the cover of specific coral growth forms may seem to indicate predator-prey relationships between the two biota. However, observations on the feeding behaviour of C. larvatus revealed no preference for a specific growth form (see chapter 5). Both encrusting and branching corals cover relatively small areas in the studied sites. Therefore, their cover may not explain the observed variation in fish abundance. The significant correlation between butterflyfish abundance and substrate structural complexity found by Luckhurst & Luckhurst (1978) and Bell et al (1986) was not supported by the findings of Roberts & Ormond (1987) and Ohman (1998). In the present study, that correlation was not significant, possibly due to the small range (0.3) of the substrate rugosity values.

37 Butterflyfishes of the Southern Red Sea 31 Chapter 4 Resource Partitioning among Four Butterflyfish Species Z. A. Zekeria, Y. Dawit, S. Ghebremedhin, M. Naser and J. J. Videler Published in Marine and Freshwater Research. 2002,vol. 53, pp.1-6.

38 32 Chapter 4. Resource Partitioning Abstract Feeding habits and territorial behaviour of four sympatric Red Sea butterflyfishes were investigated in the Eritrean coastal waters. Feeding habits were studied by focal animal sampling. Individual bite rates and types of food consumed were recorded. Stomach contents of 125 specimens were analysed in the laboratory. The food items in the stomach were sorted and their volume estimated. The four species showed marked variation in their food preferences and feeding habits. The most abundant butterflyfish Chaetodon larvatus, an obligate corallivore, forms monogamous pairs. Each pair defends a relatively small territory against conspecifics and C. semilarvatus. The latter species also feeds on scleractinian corals but lives solitary or in small aggregations. The third species, Heniochus intermedius, feeds on non-coralline benthic invertebrates (mainly polychaetes); it usually lives in pairs or in aggregations of up to 24 individuals. Both C. semilarvatus and H. intermedius occupy undefended and overlapping home ranges. The least abundant species, C. mesoleucos, forms monogamous pairs, defends a territory and feeds mainly on non-coralline benthic invertebrates (mainly nematodes and polychaetes). The study reveals that the four species co-exist in the same habitat where they partition the food resources. Both C. larvatus and C. semilarvatus feed on scleractinian corals, but partition this food source by feeding at different times. While C. larvatus was observed to feed only during daytime C. semilarvatus feeds by day and night.

39 Butterflyfishes of the Southern Red Sea 33 INTRODUCTION Fourteen species of butterflyfish are recorded from the Red Sea (Randall 1983). The distribution shows marked variations from north to south (Roberts et al 1992), as well as locally (Bouchon-Navaro & Bouchon, 1989 and Roberts & Ormond, 1987). Only twelve chaetodontid species are found in the southern Red Sea (Kemp 1998), of which four are common on the reefs around Massawa and the islands of the Dahlak archipelago (See chapter 3). The butterflyfish assemblage in the southern Red Sea differs from that in the north (Righton et al 1996). Chaetodon austriacus and C. paucifasciatus dominate the northern reefs and are absent in the south. C. larvatus, the most dominant chaetodontid in the south, has a very low density in the north (Roberts et al 1992). Butterflyfishes are one of the best-studied fish families on coral reefs (Motta 1989). Their feeding habits have been investigated in the Pacific (Reese 1975, 1981), Japan (Sano 1989), the Red Sea (Ormond 1972, Harmelin-Vivien & Bouchon-Navaro 1982) and in French Polynesia (Harmelin-Vivien & Bouchon-Navaro 1983). Distribution patterns of chaetodontids are documented for the Great Barrier Reef (Fowler 1990a), French Polynesia (Bell et al 1986 Bouchon-Navaro 1986), the West Indies (Alevizon et al 1985), and the Red Sea (Bouchon-Navaro & Bouchon 1989, Roberts et al 1992). These studies have shown that a number of chaetodontids co-exist in the same habitat. Many butterflyfish species are known to feed on corals. However, there is little information on how these closely related species co-exist in the same habitat while most of them use similar food resources. Bouchon-Navaro (1986) and Pitts (1991) examined resource partitioning among butterflyfishes in the northern Red Sea and western Atlantic respectively. Results from those studies show partitioning of either food or space resources among co-existing species. In the present work, trophic and spatial partitioning among four chaetodontid fish species in the southern Red Sea was investigated. MATERIALS AND METHODS Study area The study was conducted in the southern Red Sea near Massawa (Fig. 4.1). The fishes used for analysis of stomach contents were collected from the reef east of Sheikh Said Island. Observations on the feeding and ranging behaviour of the fish species were made on Resimedri reef near Massawa, and surveys of the distribution and abundance of the butterflyfishes took place on seventeen reefs around Massawa and in the Dahlak archipelago. Most of the reefs in the study area are of the fringing type and reach a depth of 10 m. The dominant coral genera found in the study area are Porites, Echinopora, Montipora and Stylophora. The coral cover of eastern Sheikh Said and Resimedri reefs is

40 34 Chapter 4. Resource Partitioning 45% and 57% respectively. These values are relatively high compared with the mean coral cover of 23% found for the seventeen reefs studied (See chapter 3). Distribution and abundance of fishes A belt-transect method was used to estimate the abundance of the butterflyfishes (Crosby & Reese 1996) at seventeen sites in the study area (see chapter 3). At each site, three lines, each 100-m long, were laid parallel to the shoreline. Visual census of fishes was conducted by slowly swimming or diving along the line and counting all the butterflyfishes observed within 2.5 m distance on either side of the line. Crosby & Reese (1996) recommended a transect width of 10 m. However, owing to poor visibility and because of the narrowness of many surveyed reefs we reduced the width to 5 m. The fishes were identified to species level and their abundance was recorded on a PVC slate. Data collection took place between 0900 and 1300 hours in March, April and August Figure 4.1. Study site. Asterisks indicate locations of surveyed reefs. RM = Resimedri; SS = Sheik Said Island; TW = Twalot Island. The two arrows indicate the observation and the fish collection sites.

41 Butterflyfishes of the Southern Red Sea 35 Collection of fishes 125 fishes belonging to four chaetodontid species were caught in a barrier net while diving and snorkelling. Immediately after capture, fishes were preserved in ice and transferred to a laboratory where lengths and weights were measured. The sampled fish were stored in a deep freezer until dissection. Collection took place at different times of the day from November 1999 to April On 14 April 2001, forty-five additional specimens of C. larvatus, C. semilarvatus and H. intermedius were collected to investigate the fullness index of their stomach. Five individuals from each species were captured in the morning ( hours), after noon ( hours) and in the evening ( hours). Dissection and stomach content analysis Total length, standard length and body depth were measured to the nearest mm; body mass was measured to the nearest g. After dissection mass of the stomach and length of the intestine were measured. Stomachs were opened under a dissecting microscope, and their contents were spread out in a Petri dish and examined under a microscope. The bottom of the Petri dish was divided into 0.25 mm 2 squares. Prey items were classified as turf algae, scleractinian corals (coral polyp, zooxanthellae, nematocyst and coral mucus), nematodes, sedentary polychaetes, errant polychaetes, hydrozoans, crustaceans, ascidiaceans and larvaceans. The volumetric percentage of a given food item was estimated by determining the number of grid squares covered by each food type as a fraction of total number of squares covered by the stomach content (Mol 1995). For the fish collected on 14 April 2001, the gutted mass and the mass of the stomach content were determined. Field observations Field observations were carried out while snorkelling or diving on the reef slope at depths varying from 2 to 5 m. Each fish was followed for 15 min, and the number of bites and the type of coral consumed were recorded on a PVC slate. Feeding observations were made for at least three days for each species from January to April During each day at least three replicate feeding observations were made three times: morning ( hours), noon ( hours) and evening ( hours). Ranging habits were investigated by following the movement of fishes for 3 h per day and for five days for each fish species. Five individuals from each species were marked by applying subcutaneous injections of Alcian Blue (De Jonge & Videler 1989). The movements of the marked fishes were monitored for two months. Territories were marked with floats. Sizes of the territories were determined by plotting the territories to

42 36 Chapter 4. Resource Partitioning scale on graph paper. The ranging habit and feeding behaviour for two C. larvatus pairs were monitored during one year. Data analysis Several coefficients were calculated to determine the relative importance of prey items in the diet: the mean volumetric percentage (MVP) of a prey is the sum of individual volumetric percentages for the food item divided by the number of specimens examined; the percentage frequency of occurrence (PFO) is the number of stomachs containing a particular prey item as a percentage of the total number of stomachs containing food; and the ranking index (RI) is calculated by multiplying mean volumetric percentage and percentage frequency of occurrence. H ' = h Habitat width (H h) was calculated by use of the Shannon index (H ) as follows: ( P i ln P i ) where P i represents the relative abundance of a fish species in habitat i. The evenness of fish distribution (Eh) was determined by use of the Pielou evenness index (E) which was calculated as E h = H' h /lnn Where N is the total number of habitats in which the fish species was recorded (Pielou 1966). The same equations were also used to calculate the diet breadth (H d) and diet distribution (Ed) for the four fish species where P i represents the proportion by volume of a prey item i, and N is the total number of fishes with stomachs containing food. Overlap between two species in resource use was calculated on the basis of prey types and habitat use (Sala & Ballesteros 1997). Habitat overlap (T) was determined as: T = 0. 5 P x P y 1 hi hi where P xhi and P yhi are the proportions of abundance in the habitat hi for all fish species pairs x,y. α = Diet overlap (α H ) was determined as P P H x fi y fi where P xfi and P yfi are the proportions by volume in the stomachs of the prey item f i for all fish species pairs x, y.

43 Butterflyfishes of the Southern Red Sea 37 The overlap index varies from 0, when the two species use totally different resources, to 1, when they use the resources in the same proportion. An overlap value (α H ) equal to or above 0.60 was considered significant, following Keast (1978). Stomach fullness index (FI) was the mass of the stomach content as a percentage of the gutted mass of the fish. RESULTS Distribution patterns The four species were widely distributed, with H h being 2.52 for C. larvatus, 2.41 for C. semilarvatus, 2.26 for C. mesoleucos and 2.47 for H. intermedius. They showed little variation in their distribution on the seventeen reefs, with E h being 0.09 for C. larvatus, 0.91 for C. semilarvatus, 0.91 for C. mesoleucos and 0.87 for H. intermedius. There was high overlap of habitat between pairs of species among all species (Table 4.1), particularly between C. larvatus and C. semilarvatus (T = 0.71). Table 4.1. Habitat overlap (T) among four butterflyfish species in the Red Sea Fish species C. larvatus C. semilarvatus C. mesoleucus H. intermedius C. larvatus C. semilarvatus C. mesoleucus H. intermedius 1.00 Diet C. larvatus (H d 0.24, E d 0.03) and C. semilarvatus (H d 0.41, E d 0.07) mainly ate polyps of scleractinian corals. The mean volumes occupied by this food item were 96.1% and 90.1% of the total food volume respectively (Table 4.2). As a result, the dietary overlap between the two species was very high, α H = 0.94 (Table 4.3). C. semilarvatus Table 4.3. Diet overlap (α H ) among four butterflyfish species in the Red Sea Fish species C. larvatus C. semilarvatus C. mesoleucus H. intermedius C. larvatus C. semilarvatus C. mesoleucus H. intermedius 1.00

44 38 Chapter 4. Resource Partitioning supplemented its diet with errant polychaetes and a small quantity of nematodes. C. mesoleucos (H d 1.50, E d 0.19) and H. intermedius (H d 1.88, E d 0.17) consumed different types of food with little overlap (α H = 0.48): The diet of C. mesoleucos was composed mainly of unidentified matter, polychaetes, nematodes and scleractinian corals, whereas that of H. intermedius was mainly polychaetes and scleractinian corals, supplemented with larvaceans, crabs and amphipods (Table 4.2). Table 4.2. Mean volumetric percentage (MPV), percentage frequency of occurrence (PFO) and ranking index (RI) of food items in the diets of four butterflyfish species collected from Red Sea C. larvatus C. semilarvatus C. mesoleucus H. intermedius No of fish Size range (cm) Food item MVP PFO RI MVP PFO RI MVP PFO RI MVP PFO RI Coral Polychaet (sed) Polychaet (err) Nematod Copepod Amphipod Shrimp Crab Hydrozoa Ascidacia Larvacea Turf Algae Others Unidentified Feeding habits The average feeding rate of C. larvatus was 10.5 bites min 1, whereas C. mesoleucos and C. semilarvatus fed at 6.4 and 6.6 bites min 1 respectively. H. intermedius spent most of the daytime hiding below coral and was only occasionally observed feeding at a rate of 1.1 bites min 1. With the exception of C. mesoleucos the species showed significant diurnal variation in feeding rate (P <0.05). The feeding rate of C. larvatus was low in the morning, increased around noon and decreased during the afternoon (Fig. 4.2). C. larvatus and C. mesoleucos defend

45 Butterflyfishes of the Southern Red Sea 39 territories during the day (see below) and sleep by night in coral crevices within their territories. We have no indication of feeding activity during the night in these two species The feeding rate of C. semilarvatus increased gradually over the day (Fig. 4.2). The fishes were observed wandering over the reef after sunset. Night conditions made it difficult to observe feeding activities. C. semilarvatus collected at different times of the day showed significant variation in the stomach fullness index (P <0.05): individuals captured early in the morning had more food in their stomachs (mean FI = 1.76) than those investigated later in the day (mean FI = 1.27). This offers circumstantial evidence for nocturnal feeding Feeding rate (bites mn -1 ) :30 12:00 18:30 Time (hr) Figure 4.2. Mean (± s.d.) number of bites min -1 in four chaetodontids in the southern Red Sea., g, C. larvatus;, C. semilarvatus;, C. mesoleucos;, H. intermedius. The feeding rate of H. intermedius was very low throughout the day but increased sharply in the evening just before dusk (Fig. 4.2). H. intermedius was active after sunset. Fishes captured early in the day had more food in their stomach (FI = 0.84) than those collected either at noon or in the evening (FI = 0.48 and 0.50 respectively). These findings, as in C. semilarvatus, offer circumstantial evidence for nocturnal feeding by H. intermedius. The guts of the corallivorous species, C. larvatus and C. semilarvatus, were longer than those of the omnivores (Table 4.4). The relative length of intestine also shows the relation with the feeding habit of these species. C. larvatus had a relatively long gut and the highest feeding rate, whereas H. intermedius had the shortest gut and the lowest feeding rate. Although the observed feeding rate of C. semilarvatus was much lower than that of C. mesoleucos, if our argument of night feeding by C. semilarvatus is correct the

46 40 Chapter 4. Resource Partitioning total number of bites taken by C. semilarvatus throughout the day and night could be higher than the bites consumed by C. mesoleucos by day only. Ranging and social behaviour The spacing behaviour of the chaetodontids was investigated by following movements of marked individuals. C. larvatus lives in heterosexual pairs occupying territories ranging from about 24 m 2 to 66 m 2. Pairs are monomorphic and swim close to one another; heterosexuality was confirmed by examining the gonads of twenty pairs. The abundance of C. larvatus in most sites was so high that the whole reef was divided into continuous territories. Both pair members participate in defending their territory against conspecific neighbours and occasionally against C. semilarvatus. They mainly use advertisement or display behaviour to chase intruders. More time was spent feeding than patrolling the territory. Table 4.4. Relative intestine length in four chaetodontids in the Red Sea Species Name N Fish length (FL) Intestine length (IL) Relative length (RL) Mean s.d. Mean s.d. RL = IL / FL C. larvatus C. semilarvatus C. mesoleucus H. intermedius C. mesoleucos also lives in heterosexual pairs (confirmed by examining the gonads of twenty pairs) but defends relatively larger territories, ranging from 947 m 2 to 966 m 2. Defence was directed only against conspecifics. More time was spent patrolling the territory than feeding. C. semilarvatus and H. intermedius are not territorial. Tagged individuals were not seen again on the reef after their release. C. semilarvatus is solitary or lives in groups of up to 20 individuals and actively wanders over the reef during the day. H. intermedius usually lives in pairs or in groups of up to 24 individuals. Pairs spend most of the daytime hiding below large corals. Both C. semilarvatus and H. intermedius were observed wandering over the reef after sunset. DISCUSSION Sano (1989) classified butterflyfish species into three categories based on food diversity. He grouped fishes with H d values <0.3 as specialists and fishes with values >1.0 as generalists. The third category, low diversity feeders, has intermediate H d values.

47 Butterflyfishes of the Southern Red Sea 41 According to this classification, the present study suggests that, in the southern Red Sea, C. larvatus belongs to the specialists, C. semilarvatus to the low diversity feeders and H. intermedius and C. mesoleucos to the generalists. The composition of the stomach contents classifies C. larvatus and C. semilarvatus as corallivores whereas C. mesoleucos and H. intermedius feed mainly on non-coralline benthic invertebrates. In the northern Red Sea, C. larvatus feeds mainly on scleractinian corals whereas the diet of C. semilarvatus consists almost completely of non-coralline invertebrates (Ormond 1972). According to Harmelin-Vivien & Bouchon-Navaro (1982), H. intermedius in the Gulf of Aqaba is planktivorous, feeding mainly on larvaceans. The observed differences in the feeding habits between the chaetodontid populations of the southern and northern Red Sea could be due to the availability of food. Although coral polyps are abundant in both areas, the presence of competing corallivores may make this food source difficult to exploit for less competent species. Depending on food availability, diets may vary among individuals within sites, among sites and temporally (Roberts & Ormond 1992). Regional variation in feeding behaviour has also been noted for Chaetodon auriga, C. lunula, C. unimaculatus and Forcipiger flavissimus (Harmelin- Vivien & Bouchon-Navaro 1983), for C. kleinii (Sano 1989), for C. miliaris (Ralston 1981) and for C. trifascialis (Irons 1989). In the Gulf of Aqaba, potential competitors occupy different regions on the reef: C. austriacus and C. trifascialis on the reef flat and C. fasciatus at 5 m depth (Bouchon- Navaro 1986). In the southern Red Sea, however, the two corallivorous species share the same habitat. Lack of space partitioning among the fishes may be due to the nature of the reef in the study area. The reefs in the Gulf of Aqaba extend to depths greater than 30 m while the reefs of the southern Red Sea rarely exceed 10 m (Roberts et al 1992). C. mesoleucos and H. intermedius seem to partition diverse non-coralline invertebrates while C. larvatus and C. semilarvatus consume scleractinian corals. The observed agonistic behaviour of C. larvatus against C. semilarvatus indicates competition for food resources between these rival species. The niche compensatory hypothesis (Ebeling & Hixon 1991) asserts that cooccurring species showing a high degree of overlap in one niche dimension (e.g. habitat) separate along another dimension (e.g. diet). Taking both distribution and feeding habits into consideration, partitioning of resource use among the four chaetodontid species is evident. C. larvatus and C. semilarvatus show significant feeding overlap but partition the resources temporally. As indicated by the observational feeding studies and by the diurnal variation in the stomach fullness index, C. larvatus feeds mainly during daytime whereas C. semilarvatus seems to concentrate its feeding effort late in the evening and during the night. According to Bemert and Ormond (1981) C. semilarvatus is a nocturnal species.

48 42 Chapter 4. Resource Partitioning Schoener (1974) pointed out that habitat separation is far more effective than food separation in preventing species overlap. In the present study, at least at a local scale, food partitioning appears to be more important than habitat partitioning in structuring fish assemblages. However, when the whole Red Sea is taken into consideration, it is likely that habitat partitioning has played an important role in the regional distribution of the butterflyfishes. The butterflyfish assemblage of the northern Red Sea is different from that of the south. For example, C. larvatus dominates the reefs in the south whereas C. austriacus, which is absent in the south, is dominant in the north. These two species exhibit a number of similar characteristics: they are obligate corallivores, live in heterosexual pairs, and defend small territories (Bouchon-Navaro & Bouchon 1989, Righton et al 1998, Roberts et al 1992, Wrathall et al 1992). Since both are generalist obligate coral feeders, C. larvatus and C. austriacus occupy similar niches on the reefs. The similarities in ecological role could be the reason for their dominance in different parts of the Red Sea. The four chaetodontid species studied here show significant variation in their feeding rates. C. larvatus has a higher feeding rate than the non-coralline invertebrate feeders. The observed feeding rate for C. semilarvatus is low in the morning and increases during the day. Results from feeding observation and fullness index suggest that C. semilarvatus remains active and feeds also by night. Bemert and Ormond (1981) found C. semilarvatus to feed entirely by night. The total number of bites taken during day and night by C. semilarvatus could be much higher than those taken by C. mesoleucos during the day only. Since coral tissue is poor in nutrients, the fishes take less energy per feeding bite. Coral has high water content and its energy content is relatively low (Tricas 1989). Moreover, corallivorous butterflyfishes have very low absorption efficiencies (Hourigan 1989). The feeding habits of the four fish species are also reflected in the length of their guts. The longest guts occur in microphagous and herbivorous species, and carnivorous fishes have the shortest guts (Kapoor & Smit 1975). Hence, corallivorous species require higher feeding rates (Reese 1991).

49 Butterflyfishes of the Southern Red Sea 43 Chapter 5 Territorial and Feeding Behaviour of the Brownface Butterflyfish (Chaetodon larvatus) Z. A. Zekeria and J. J. Videler

50 44 Chapter 5. Territorial and Feeding Behaviour Abstract Feeding behaviour of Chaetodon larvatus was investigated in the southern Red Sea. Fifteen C. larvatus pairs were followed and their ranging behaviour, territory size, feeding habits and activity patterns were investigated. Results show that pairs occupy feeding territories, which they defend against conspecific and congeneric intruders. The size of the territories, which ranged from 24 to 66 m 2, showed no correlation with live coral cover. C. larvatus are corallivores and feed on different coral polyps. Comparison between the cover of corals on the substrate and the coral types in the diet shows that C. larvatus feeds on corals in proportion to their abundance in the substrate. The feeding rate, which averages 13.8 bites min -1, was related to neither the coral cover nor the size of the territory. C. larvatus is diurnal and spends most of the daytime feeding. The remaining time is spent in other activities such as swimming, hovering and territorial defence.

51 Butterflyfishes of the Southern Red Sea 45 INTRODUCTION Butterflyfishes (Chaetodontidae) are among the most studied families of fishes in the tropical oceans (Motta 1989). The fishes live among corals in shallow waters where individual fishes can easily be approached and their behaviour observed at close range. As a result, the social and territorial behaviour of many chaetodontid species has been investigated. Studies on the feeding and territorial behaviour show that most butterflyfish species feed on corals by taking bites of polyps and mucous and that the fishes usually occur as pairs and appear to be spread out over the reef (Reese 1975, Fricke 1986, Hourigan 1989, Roberts & Ormond 1992). A number of authors have investigated the territorial and feeding behaviour of butterflyfishes in the northern Red Sea, targeting mainly Chaetodon austriacus and C. paucifasciatus. The distribution of these species is restricted to the northern and central part of the sea (Fricke 1986, Roberts & Ormond 1992, Wrathall et al 1992, Righton et al 1998). However, little is known about the behaviour of C. larvatus, the dominant butterflyfish in the southern part of the Red Sea. C. larvatus, the brownface butterflyfish, is endemic to the Red Sea and the Gulf of Aden and is widely distributed throughout the region (Roberts & Ormond 1992, Righton et al 1996). It is the dominant chaetodontid species in the southern Red Sea where its abundance correlates with the distribution of live coral cover (see chapter 3). Preliminary investigations on the feeding and territorial behaviour of the species indicate that C. larvatus lives among corals and feeds on coral polyps (see chapter 4). The objective of the present work is to investigate the territorial behaviour, food preference and activity patterns of C. larvatus in the southern Red Sea. The behaviour of C. larvatus was observed at four sites near Massawa. A food preference was investigated by relating the cover of coral available in the fish territories with the proportion of the corals in the diet. The territorial behaviour, activity patterns and time budget were investigated by following marked individuals. MATERIALS AND METHODS Study site Data were collected from Resimedri reef, near Massawa harbour, in the southern Red Sea (Figure 5.1). The fringing reef is approximately 2 km long with an average width of about 100 m. The depth of the reef drops gently from 0.5 m near the reef flat to a maximum of 10 m. The reef is rich in coral diversity but the corals are unevenly distributed and form distinct reef zones and monogeneric coral stands. The dominant coral genera in the area include: Echinopora, Montipora, Pavona and Porites. The live

52 46 Chapter 5. Territorial and Feeding Behaviour coral cover of the reef is 57%; a relatively high cover compared to the mean value of 23% for other reefs in the area (see chapter 3). Figure 5.1. Study site. MP = Massawa proper; RR = Resimedri reef; SSI = Shiek Said Island; TW= Twalot. Field observations Field observations were carried out between 9:00h and 13:00h while snorkelling or diving on the reef slope at depths varying from 2 to 5m. Fifteen pairs of fishes were selected in four sites and their territorial behaviour, feeding behaviour, and activity patterns investigated. The size of territories and territorial behaviour The size of territories was determined by following each fish pair for 30 min. During the first 15 min, the position of the fishes was marked every 30 sec using floats weighed with galvanised nails. A preliminary boundary of the territory was then determined based on the position of the markers. The edges of the territories were mapped in a more refined way during the second 15 min when fishes were followed and their position marked whenever they moved outside the marked area. A pilot project showed that increasing the observation time beyond 30 min did not change the sizes of the territories significantly. The sizes of the territories were estimated using a modified method of Crosby & Reese (1996). The length of the territory along the longest dimension was measured using a measuring tape. The width of the territories was then measured on three equidistant positions along the long axis (at 25%, 50%, and 75% of its length). All width

53 Butterflyfishes of the Southern Red Sea 47 measurements were taken perpendicular to the long axis. The territories were plotted to scale on graph paper and their sizes estimated by counting the number of square millimeters enclosed within the territories. Persistence of fishes in their territories was investigated by monitoring the territorial range of three tagged fish pairs. These fish were tagged using a subcutaneous injection of Alcian Blue (De Jonge & Videler 1989) and the size of their territories was monitored for six months. Feeding behaviour Feeding data were collected by closely following individual fishes for 10 min. Three replicate observations were made in each territory. Bite rates were tallied and the consumed type of coral recorded. Since C. larvatus are sexually monomorphic it is not possible to identify the sex of individuals by their external appearance. Preliminary investigation showed no differences in the feeding rates between individuals of a pair in a C. larvatus territory. Hence, one of the pair was randomly selected and followed for the entire observation period. Time budget Data on activity patterns were collected while slowly snorkelling or diving. Fifteen pairs of C. larvatus were followed for 15 min each and their activities recorded at 15 sec intervals. A total of 60 point observations were made from each territory and the data were used to estimate the time budget. The activity performed by a fish was identified into one of four categories: feeding, swimming, hanging, and agonistic interactions (Righton et al 1998). During feeding a fish was either inspecting a coral at close range or taking a bite. In the former case the fish swims close to the substrate at low speed. Swimming refers to the act where a fish swims at a faster speed a few cm above the substrate. During agonistic interaction a fish displays aggressive behavior in defending its territory or in response to aggression by others. On a few occasions the fishes were hanging in the water column which was sometimes followed by 'cleaning' by cleaner wrasses. Coral cover Live coral cover was estimated by dividing each territory into as many one-squaremeter quadrats as possible. The percentage cover within each quadrat was then estimated. The coral was identified to the genus level and the cover of each genus estimated separately. The average cover of all quadrats within a territory was used as a cover value for the entire territory.

54 48 Chapter 5. Territorial and Feeding Behaviour Data analysis Relationships among size of territory, coral cover, bite rate, and feeding time were determined using Pearson's correlation coefficient. The Student t-test was employed to compare the gastrosomatic indices of male and female C. larvatus. The types of corals consumed by the fish were compared with their availability in the substrate using Vanderploeg and Scavia's relativized selectivity index, E* (Lechowicz 1982) and the significance of variation of the food items in the diet and their proportion in the substrate was compared using chi square (χ 2 ) test. RESULTS Coral Cover Table 1 shows the size and coral cover of 15 territories in 4 sites. The mean size of the territories ranged from 33 to 45 m 2 but the observed variation in the size was not significant (P=0.61). The territories in the four sites differ in coral cover. The dominant coral genera in sites I, II and II were Porites, Echinopora and Montipora respectively. Site IV was covered mainly by Echinopora and Pavona. The highest coral cover was 77% (site II) and the lowest value was 53% (site I) but the observed variation in coral cover among the sites is insignificant (P=0.08). No correlation was observed between the size of the territories and the coral cover (r=0.27; P=0.33). Table 5.1. Mean (±s.d.) territory size of C. larvatus; coral cover (%); proportion of corals in the field; and proportion of corals in the diet. Sites I II III IV No of territories Size of territories (m 2 ) 38.6 (12.5) 45.1 (7.8) 32.8 (6.3) 41.6 (21.8) Coral cover (%) 53 (17.0) 76.5 (11.2) 64.5 (5.3) 56.4 (13.1) Proportion on the substrate Porites 0.27 (0.09) 0.03 (0.02) (0.01) Echinopora 0.11 (0.10) 0.52 (0.15) (0.21) Montipora 0.04 (0.01) 0.01 (0.01) 0.62 (0.06) 0.02 (0.01) Pavona 0.02 (0.02) 0.16 (0.11) (0.24) Stylophora 0.08 (0.05) 0.03 (0.02) 0.02 (0.01) 0.03 (0.01) Other 0.01 (0.01) 0.02 (0.03) (0.04) Proportion in the diet Porites 0.55 (0.31) 0.01 (0.01) (0.02) Echinopora 0.2 (0.18) 0.54 (0.19) 0.02 (0.04) 0.39 (0.24) Montipora 0.03 (0.05) (0.15) 0.00 Pavona 0.11 (0.07) 0.43 (0.18) 0.08 (0.14) 0.49 (0.26) Stylophora 0.11 (0.07) 0.01 (0.01) 0.11 (0.05) 0.01 (0.01) Other 0.01 (0.01) 0.01 (0.02) 0.04 (0.04) 0.09 (0.01)

55 Butterflyfishes of the Southern Red Sea 49 Ranging behaviour Paired C. larvatus occupy territories which they defend against intruders. Dissection of twenty territorial pairs collected from a nearby reef showed that all pairs were made up of a male and a female partner of similar size (Figure 5.2). The correlation between the size of male and female partners was significant (r=0.838, paired t-test P=0.636). Both sexes were observed defending their territory against neighboring conspecifics and against intruding congeners (mainly the corallivorous C. semilarvatus). The defending fishes mainly display advertisement behaviour to chase intruders out of the territories. Occasionally, however, the defendants directly attack the intruders until they leave the invaded territory. C. larvatus are active during the day and remain hidden in coral crevices by night. During the entire observation period the fishes were never seen outside their territories. However, in a separate study, where the spawning behaviour of C. larvatus was investigated, the fishes were observed leaving their territories to form aggregations with their neighbors. These aggregations, which lasted for only a few minutes at a time, took place in the evening before the fishes retreated to their resting places for the night (see chapter 6). Long term monitoring of the ranging behaviour of three tagged C. larvatus pairs revealed that the fish occupied the same territories for six months. We have no information showing if the fishes stayed in the same territories after the end of the monitoring period Length of males (cm) Figure 5.2. Correlation between the sizes of male and female partners of C. larvatus. Feeding habits Length of females (cm) Ninety nine percent of all feeding bites were taken from live coral. The feeding rate in the four sites varied from 16.1 to 11.6 bites min -1 with an average of 13.8 bites min -1.

56 50 Chapter 5. Territorial and Feeding Behaviour The rate of feeding was not correlated with the size of the territory (P>0.05). C. larvatus fed on all types of corals which were available in their territories (Figure 5.3). Vanderploeg and Scavia's relativised selectivity index, E*, showed that Porites was a slightly preferred prey item (E*=0.18), Montipora was slightly avoided (E*=-0.29), while the other coral species were neither preferred nor avoided (0.1>E*>-0.1). However, since all the above E* values lay between -0.5 and 0.5, C. larvatus can not be regarded as selective in its feeding habits. Furthermore, chi square tests showed no variation between the cover of coral genera and their consumption rate (P>0.05) proportion Pori Echi Mont Styl Goni Pavo other coral genera Figure 3. Relationship between the proportion of coral cover ( the ratio of the cover of a genus to the total live coral cover) (sold bars) and the ratio of the number of bites taken from a genus to the total number of bites(open bars). Lables of coral genera: Pori = Porites, Echi = Echinopora, Mont = Montipora, Styl = Stylophora, Goni = Goniopora, Pavo = Pavona, other = other coral genera. Time budget Feeding accounted for about 80% of the time budget in C. larvatus (Table 5.2). A feeding bout started with a close inspection of the substrate followed by nipping off polyps from scleractinian corals. Up to 30 bites are taken during one bout and the fishes change feeding positions after each bout. Individuals of a pair feed close to each other and occasionally take bites from the same colonies.

57 Butterflyfishes of the Southern Red Sea 51 Swimming is the major non-feeding activity and it took 17% of the daily time budget. Swimming was sometimes followed by a display of advertisement behaviour where the fishes spread their dorsal and anal fins. This behaviour was obviously a warning signal for potential invaders. Occasionally the warning display was followed by a direct attack by the defending pair towards intruders. Usually the fight ended abruptly when the intruders left the invaded area. Hovering was a rarely observed behaviour which accounted for only 2% of the time budget. The fishes were sometimes approached by cleaner wrasses (Labroides dimidiatus), who cleaned the gills and the skin of the butterflyfish. C. larvatus spend little time (1.5%) defending their territories or invading other territories (Table 5.2). Table 5.2. Time budget of C. larvatus in four sites in the southern Red Sea. Values represent mean (± s.d.) percentage of time spent. Site I II III IV Feeding 80.5 (2.4) 86.7 (3.5) 77.8 (17.1) 67.8 (12.3) Swimming 16.9 (2.6) 10.7 (3.2) 17.8 (12.9) 28.3 (10.4) Hovering 1.7 (1.4) 1.3 (1.4)) 3.3 (3.3) 2.8 (3.5) Social interaction 0.8 (1.0) 1.3 (1.4) 1.1 (1.0) 1.1 (1.9) DISCUSSION Corals represent an evenly dispersed, predictable and long-term renewable food resource (Tricas 1986a) which can be defended by territorial fishes (Reese 1991). Territoriality is a widespread strategy among obligate corallivorous butterflyfishes. Our result shows that C. larvatus is obligate coral feeder. It lives in heterosexual pairs, which defend territories against conspecific and congeneric intruders. However, Roberts & Ormond (1992) categorized C. larvatus as solitary. Their work was conducted in the northern Red Sea where the density of C. larvatus is relatively low (Roberts et al 1992, Righton et al 1996). According to Reese (1981), the frequency of pairing in fishes decreases in stressed fish populations. The solitary behaviour of C. larvatus and its lower densities in the northern Red Sea may be caused by stress due to competition with a congeneric species, C. austriacus. Corallivorous butterflyfishes spend most of their time feeding and are classified as energy maximizers following Hixon (1980). According to Hixon's model the size of the territory of energy maximisers should decrease with increase in coral cover. Tricas

58 52 Chapter 5. Territorial and Feeding Behaviour (1989b) and Righton et al (1998) found an inverse relationship between the cover of live corals and the size of territories. However, our results show no correlation between these two parameters. The absence of correlation could be due to the relatively rich coral cover in the study area and to variation in coral cover among the territories. The coral cover in the study area ranged from 53% to 77%. While the coral cover recorded by Righton et al (1998) was about 24%. The coral cover in the Pacific where Tricas (1989b) investigated the territorial behaviour of C. multicinctus ranged from 8% to 90%. A number of butterflyfishes live in pairs (Roberts & Ormond 1992, Allen et al 1998). In C. quadrimaculatus and C. multicinctus females in monogamous pairs feed more than unmated females, while paired males feed less than their mates but contributed most to the defense of the territory. This division of labour increased the time available to females for feeding while enabling the paired male to share in his mate's increased fecundity (Hourigan 1989). In addition to its contribution to reproductive success, pairing may help butterflyfishes in protecting the territory resources more effectively. Fricke (1986) suggested that pair swimming is a form of advertisement display, similar to duetting in birds, and that it is essential for territory maintenance because unpaired fishes are unable to defend territories economically. In the present study, preliminary investigations of stomach contents showed higher consumption of food by females. It was not possible to identify the sex of the fishes in the field. However, both males and females were observed defending territories. David Strang (personal communication) found no difference in feeding rate between the two sexes. Further study with tagged individuals is needed to verify absence of division of labor between the two sexes. C. larvatus feeds throughout the day at a high rate. Reese (1991) compared the feeding rate of corallivores butterflyfishes with the high grazing rate of cattle. Since corals, like vegetation, are energetically poor food resources (Tricas 1989a), the fishes need to consume large amounts. Corals have high water content, and a low caloric value. Less than 25% of coral energy content is absorbed in the gut of the fishes (Tricas 1986a & Hourigan 1989). As a result, C. larvatus spent a large proportion of their time feeding. The time spent defending territories is relatively low because butterflyfishes keep intruders out of their territories mainly by advertisement, such as bright coloration and aggressive display (Roberts & Ormond 1992, Righton et al 1998).

59 Chapter 6 Butterflyfishes of the Southern Red Sea 53 Spawning seasonality in the Brownface Butterflyfish (Chaetodon larvatus) Z. A. Zekeria and J. J. Videler

60 54 Chapter 6. Spawning Seasonality Abstract The spawning behaviour of C. larvatus was investigated in the southern Red Sea. Three methods were employed to study seasonal patterns of spawning: 1. A monthly change in gonadosomatic index was monitored for two years. 2. Changes in histological development of gonads were monitored by analysing monthly samples of gonads. 3. Field observations of spawning behaviour were made and spawning seasonality monitored. Results from the three studies indicate that May and June were the major spawning months. Gonad size was highest during these months; most of the gonads attained maturity when investigated histologically, and highest courting frequency was recorded during these months. The results agree with the recruitment period, which was found to be in June and July.

61 Butterflyfishes of the Southern Red Sea 55 Introduction Reproduction in coral reef fishes has been extensively studied in Pomacentrids (Sale 1980). Five methods have been employed to detect the spawning periods: field observations of spawning events, visual evaluation of maturity of gonads using macroscopic characteristics, the use of gonad indices, the examination of frequency distribution of oocyte diameters, and the observation and analysis of histological sections of gonads. These methods differ in their accuracy and each has its own advantages and disadvantages. Some of these methods have been employed to study spawning seasonality of butterflyfishes. For example, Lobel (1989a) and Yabuta (1997) monitored the spawning behaviour of butterflyfishes in the field. Seasonal change in the gonadosomatic index was investigated by Ralston (1981) and Lobel (1989b). Tricas and Hiramoto (1989) analysed the histology of gonad development in butterflyfishes. Histological investigation provides better information on the spawning seasonality of tropical reef species, which have a protracted spawning season (Tricas & Hiramoto, 1989). Butterflyfishes are among the best-studied families of coral reef fishes (Motta 1989). However, information on reproduction habits exists only for a few species (Ralston 1981, Lobel 1989a, 1989b, Tricas & Hiramoto 1989, Yabuta 1997, Yabuta & Kawashima 1997). Most of these studies were conducted in the tropical east Pacific or in Japan. There are only two reports on the spawning of butterflyfishes from the Red Sea. Spawning in Chaetodon paucifasciatus was reported by Fricke (1986) while Ghraibeh and Hulings (1990) investigated reproduction seasons of three chaetodontids. Both studies were conducted in the Northern Red Sea. There is no information in the literature on the reproduction of chaetodontids in the southern Red Sea. Chaetodon larvatus, the most abundant chaetodontid in the southern Red Sea, is endemic to the Red Sea and the Gulf of Aden (see chapters 2 & 3). The adults are corallivores and live in heterosexual pairs, defending relatively small territories (see chapter 5). Growth and recruitment behaviour of C. larvatus has been investigated (see chapter 7 & 8). The objective of this study is to investigate the reproduction pattern of C. larvatus in the southern Red Sea. Three approaches have been employed to determine the spawning season of this species. First, spawning behaviour was monitored by following focal fishes in the field. Second, fish samples were periodically collected and seasonal changes in the size of gonads monitored. Third, changes in oocyte development were investigated using histological techniques.

62 56 Materials and Methods Chapter 6. Spawning Seasonality Study site Field observations were conducted on Resimedri reef, near Massawa harbor, in the southern Red Sea (Figure 6.1). This fringing reef is reaches a maximum depth of about 10m and it is exposed to moderate waves. Fish specimens were collected from a reef located east of Sheik Said Island; about 1.5 km south of Resimedri reef. Figure 6.1. Study site. FCS = fish collection site; MP = Massawa proper; OS = observation site; RR = Resimedri reef; SSI = Sheik Said (Green) Island; TW= Twalot. Field Observation Spawning in C. larvatus was monitored by snorkeling on the study site in the evening hours. Activity of the fishes and their social behaviour was recorded. Field observations started two hours before sunset and continued until the fishes retired to coral crevices for the night rest, a few minutes after dusk. The observations took place twice a month from January to December Studies on recruitment suggested April and May as spawning period for C. larvatus. Hence, during these months the frequency of observation was increased to twice a week.

63 Butterflyfishes of the Southern Red Sea 57 Gonadosomatic Index Monthly collections of about thirty fishes were made from August 1998 to August Fish collection took place between 9:00 and 13:00 using a barrier net and handnets. Immediately after capture the fish samples were transported on ice to the laboratory where the total and standard lengths (to the nearest mm) and the body mass (to the nearest g) were measured. The fish were then preserved in a deep freezer until dissection. After a couple of days in the freezer, the fishes were dissected; the gonads removed; and the mass and colour of the gonads recorded. Five gonads from each month were selected for further histological analysis. The gonadosomatic index was calculated as the ratio of gonad mass to the body mass expressed as expressed in a percentage. Histological analysis of gonads The gonad samples were fixed in Bouin s solution for at least 24 hours. Gonad tissue was dehydrated and cleared in ethanol and xylene respectively. The tissues were then embedded in paraffin and 7 µm-thick sections cut on a microtome. The sections were mounted on microscope slides and stained using hematoxylin / eosin following the procedure given in Preece (1965). Oocytes were examined and their developmental stages determined using a light microscope. At least 100 oocytes were randomly selected from each oocyte stage. The area of each oocyte was measured using an image analyser and the diameter of the oocytes was calculated from the area assuming a spherical shape of the oocytes. Results Sex ratio and size of maturation A total of 619 fish specimens were collected of which 598 were sexed. The sex ratio for these specimens was 48 males to 52 females. The size of most of the specimens was more than 80mm (Table 6.1). The smallest mature male and the smallest mature female recorded from the sample were 58 mm and 68 mm respectively. Fifty-percent maturity was attained at 66 mm in females and 71 mm in males. The slightly higher abundance of females and their earlier maturity could be due to the fact that females can be distinguished at a smaller size than males. Unlike testis, ovaries are yellowish in colour and lack fatty tissue. Hence, the female gonads can easily be distinguished at a smaller size. Gonad Histology The ovaries in C. larvatus are heavy, rounded, bilobed and yellowish in colour while testes are elongated, slender and white. In gravid females, the ovaries occupy about half of the visceral cavity. Fatty tissues, which surround the tests and other internal organs

64 58 Chapter 6. Spawning Seasonality of ripe males, are not present in females. Since ovaries are larger and easier to stage compared to testes only ovaries were used for histological analysis. A. Oocyte development Oocytes of C. larvatus were classified into five stages following the description given by West (1990). These stages are chromatin nucleolar stage, perinucleolar stage, yolk-vesicle formation, vitellogenic stage, and ripe stage. Table 6.1. Sex ratio by size class expressed in mm total length (TL) for Chaetodon larvatus Size class n Percentage mm (TL) Females Males undifferentiated < Chromatin nucleolar (oogonia) stage At this stage, the ovary consists of numerous ovigerous folds extending from the ovarian wall towards the centre of the ovary. Nests of oogonia arise within the ovarian luminal epithelium during the primary growth. The chromatin nucleolar stage is characterised by small oogonia (7µm) and by the presence of only one nucleolus. Ovaries collected in January and February had oogonia, which were not well preserved during the histological preparations. 2. Perinucleolar stage The oocytes are characterised by their small size (±15 µm) and by the presence of a relatively large central nucleus that occupies more than half of the cell volume (Fig. 6.2A). Concomitant with oocyte growth, the nucleus (germinal vesicle) increases in size.

65 Butterflyfishes of the Southern Red Sea 59 At a later stage of development the nucleus contains many nucleoli around its periphery. Perinucleolar oocytes were present throughout the year in adult females but were dominant in samples collected in January, February and March. A B Y P C D T Z G Figure Stages of oocyte development in Chaetodon larvatus. A-perinucleolar oocytes (p); B-yolk vesicle formation stage (y); C-vitellogenesis stage; D-magnified vitellogenesis stage showing the three membranes (Z = zona radiata, T = Theca, G = Granulosa) 3. Yolk vesicle (cortical alveoli) formation This stage is characterised by the appearance of yolk vesicles in the cytoplasm. The vesicles first appear as vacuoles but later increase in size and number to form several peripheral rows and give rise to cortical alveoli, which surround the central nucleus (Fig 6.2B). A non-cellular membrane, the zona radiata, begins to form between the follicular layer (theca and granulosa) and the developing oocyte. The size of the yolk vesicle stage oocyte is still small (±81 µm). Cortical alviolar stage oocytes were observed in samples collected on February, March and August. 4. Vitellogenesis stage. The vitellogenic (yolk) stage oocyte is characterized by the appearance of yolk vesicles in the cytoplasm. During early vitellogenesis small yolk granules appear in the periphery of the oocyte. Later in the development the yolk, granules migrate towards the center and completely fill the cytoplasm (Fig. 6.2C). The mean size of vitellogenic oocytes is ±185 µm and the size ranges from 97 to 618 µm. As the cell enlarges, the

66 60 Chapter 6. Spawning Seasonality vitelline membrane thickens and develops radial striation to form the zona radiata (Fig. 6.2d). Vitellogenic stage oocytes were found throughout the year, except in samples from January and February. 5. Ripe (mature) stage This is the final stage of oogenesis where oocyte development leads to the release of mature oocytes into the ovary lumen. In many marine teleostes, there is concomitant rapid increase in size due to hydration of the oocytes. The final stage of oocyte maturation is difficult to follow because of shrinkage and distortion of the cells during histological processing. In addition, ovulated oocytes may be lost from the ovarian lumen during the initial tissue processing (West 1990). Tricas and Hiramoto (1989) found late phases of hydrated oocytes only in females collected a few hours before sunset. They argue that oocyte hydration starts a few hours before spawning. None of the ovaries examined during the present study were hydrated and no spent ovaries were observed. B. Gonadosomatic Index. Monthly changes in gonadosomatic index (GSI) for male and female C. larvatus are depicted in figure 6.3. The GSI data for the two sexes show significant variation among months (t-test, P<0.05). Relatively higher GSI values were recorded from March to July and lower values were recorded for January and February. The values remained relatively low during summer and autumn. For example, in 1999/2000 mean GSI for females was 0.8 in January and peaked in March to a mean value of 3.8 (Fig. 6.3B). This was followed by a steady decline from April to August. From September to December the GSI values remained low at 1.9. The trend is apparently similar in the two years studied (Figure 6.3A and 6.3B). The only notable difference between the two years is the very low GSI for the sample collected in April This sample was taken from fishes found aggregated in one spot. Fishes in this sample had small ovaries with an average mass of 0.46 g. The development of testis also showed a similar trend as that of the ovaries. Significant correlation was recorded between males and females in the monthly GSI patterns (r=0.78, p>0.05). However, the GSI curves for males are less clear (Figure 6.3C and 6.3D). For most of the samples, large variation was recorded in the size of testes collected in the same month. The large variation could be due to the error in weighing the testis. Testis is normally found embedded in fatty tissue, which is difficult to clear. As a result testis mass includes the mass of adhered fatty tissue which has a profound effect on the in calculation of the gonadosomatic index.

67 Butterflyfishes of the Southern Red Sea A (Gonad wt / somatic wt) X B C D Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Months Figure 6.2. Monthly gonadosomatic indexes (GSI). a) Females 1998/99 b) Females 1999/2000 c) Males 1998/99 d) Males 1999/2000

68 62 Chapter 6. Spawning Seasonality Histological analysis of gonads also shows seasonal patterns. In January and February, all ovaries were dominated by oocytes in early developmental stages (chromatin nucleolar and perinucleolar stage oocytes). No advanced oocytes (vitellogenic and mature stages) were observed during the two months. On the other hand, from April to December advanced stage ovaries dominated the gonads. During these months, a few gonads with cortical alveoli were recorded and most of the gonads had vitellogenic or mature oocytes. Perinucleolar oocytes where observed throughout the year. The results from the histological analysis show that the fishes do not spawn during the winter when the ovaries are dominated by oocytes in their early development stages. Fast changes in oocyte development occur in February and March. The changes in oocyte development may indicate the approach of the spawning season. C. Field observations. C. larvatus live in heterosexual pairs and defend small territories (see chapter 4). With the exception of occasional antagonistic interactions related to territorial defence, C. larvatus rarely show social interactions with neighbouring conspecifics. The only time neighbours showed positive interactions was when they formed aggregations during evening hours. These aggregations took place mainly in the evening hours of April and May. Each aggregation lasted for only a few minutes and was repeated many times per evening. During aggregation, groups of up to 16 individuals gather from neighbouring territories and swim together for about five minutes a few cm from the bottom. During this time, territorial borders are abolished and the neighbours freely mingle with one another. After swimming for a while the group separates into many pairs. The pairing lasts for about 30 seconds during which the fishes exhibit brief 'courting-like' behaviour. This behaviour takes place among pairs that may or may not be partners from the same territory. One individual of the courting pair (assumed to be a male) courts the other member of the pair (assumed to be a female) by swimming behind her placing his snout against her abdomen. After following the female for a few seconds the male dashes forward and abruptly brakes in front of the female, blocks her way, and waves his caudal fin. The courting lasts for about 30 sec and ends with the female chasing the male. At the end of the courting, the whole group breaks up and the fishes disperse pair-wise to their respective territories. Despite observations of courting behaviour on many occasions, no spawning in C. larvatus was observed. Discussion Spawning in butterflyfishes was observed in the western Atlantic (Colin 1989), in Hawaii (Lobel 1989a & b), in the Red Sea (Fricke 1986) and in Japan (Yabuta 1997,

69 Butterflyfishes of the Southern Red Sea 63 Yabuta & Kawashima 1997b). These studies point out that spawning takes place at dusk following a short period of courting. A male in a heterosexual pair courts the female by swimming closely behind her and by placing his snout close to her abdomen. After a brief courting display, the pair ascends from the bottom and they simultaneously release their gametes on the water column (Colin 1989, Lobel 1989, Yabuta 1997, Yabuta & Kawashima 1997). In some case, non-paired males intrude into a spawning pair and release sperm (Lobel 1989). In most of the observed cases courting and spawning took place between heterosexual pairs, which defend permanent territories. However, Yabuta and Kawashima (1997) reported spawning between one male and many female C. trifascialis. This species lives solitary and individuals defend territories against others of the same sex. Each male territory covers 2 3 female territories and mating occurs between a male and the females living in his territory. Courting and spawning usually takes place within the territories of paired butterflyfishes. In C. trifasciatus, however, pairs were observed to migrate to a temporary spawning territory located a few hundred meters from their feeding territories (Yabuta 1997). In the present study, despite substantial effort to investigate the spawning pattern of C. larvatus no actual spawning was observed. Several pairs displayed courting-like behaviour. During this time, territory borders were crossed but no agonistic interactions were observed. After brief courting, all fishes returned to their respective territories. It is not clear whether the observed aggregation and subsequent courting is a prologue to spawning or if it is a completely different ritual unrelated to reproduction. We are sure that spawning did not take place during our observations. The fishes were followed until after sunset when they retired to their night shelters. Further investigation is required to find out if spawning takes place during other times of the day. The gonadosomatic indices and histological analysis of gonads show that C. larvatus spawns seasonally from April to June. Studies of recruitment patterns of the fish species in the study area revealed that new recruits settle on the reef mainly during June and July (see chapter 7). The pelagic larvae spend about 20 days in the water column before settling on the reef as recruit (Zekeria unpublished data). The timing of spawning and the duration of the larval phase suggest May and June as spawning months, which coincides with the results obtained in the present study. Seasonal reproduction of butterflyfishes was observed in the Western Atlantic (Colin 1989), in the Pacific (Ralston 1981, Walsh 1987, Lobel 1989a & 1989b, Tricas & Hiramoto 1989), in eastern Pacific (Yabuta 1997, Yabuta & Kawashima 1997) and in the Red Sea (Fricke 1986, Gharaibeh & Hulings 1990). In the western Atlantic and the pacific spawning took place in winter and in spring whereas in the Red Sea spawning occurred during summer and autumn (June to December).

70 64 Chapter 6. Spawning Seasonality The seasonal pattern of spawning in coral reef fishes was found to follow annual variations in water temperature and day length (Walsh 1987). Colin (1986) correlated spawning of five Chaetodontids in the western Atlantic with the water temperature. He pointed out that winter, when water temperatures range between 25 C and 28 C, would be the most appropriate season for butterflyfish spawning and ruled out the possibility of spawning when water temperatures exceed 26 C. However, the present study showed that C. larvatus spawn from April to June with water temperatures ranging from 28 C to 33 C. Yabuta (1997) observed C. trifasciatus spawning between 25.6 C and 30.9 C. The observed spawning of C. larvatus at higher temperatures could be due to adaptation to the local conditions. The monthly mean water temperatures in the study area range from 27.7 C in January to 33.6 C in September. Juvenile C. larvatus grow at a fast rate for the first six months of their life (see chapter 8). The fast growth of the young fishes enables the juveniles to quickly pass the risk of a young age when they are vulnerable to heavy predation (Ralston 1981). Growth of young fishes is enhanced at higher temperatures (Sogard & Olla 2001). The delayed spawning of C. larvatus until late spring could be an adaptive strategy for juveniles to be recruited in early summer when the temperature is higher. Seasonality in ocean currents and food availability could also play important role in affecting the seasonality of spawning. However, little work has been done on the oceanographic and planktonic systems of the southern Red Sea. Hence, it is not possible to relate the spawning season with patterns in water movement and food availability.

71 Butterflyfishes of the Southern Red Sea 65 Chapter 7 Temporal and Spatial Recruitment Patterns in Chaetodontids and Pomacanthids in the Southern Red Sea Z. A. Zekeria and J. J. Videler

72 66 Chapter 8. Growth Patterns Abstract Temporal and spatial recruitment patterns were investigated by monitoring settlement of chaetodontids and pomacanthids on two reefs in the southern Red Sea. Recruitment data were collected by visual census on the reefs at monthly intervals. Seasonal patterns were studied by collecting recruitment data from two reefs for one year while long-term temporal patterns were investigated by monitoring settlement on one site for four years. Results show seasonal recruitment for Chaetodon larvatus, C. semilarvatus and Pomacanthus spp. Recruitment occurred mainly in June and July. The three species differed in their recruitment patterns. The chaetodontids were recorded mainly from sites rich in live coral where branching Montipora dominate the substrate while pomacanthids were abundant on sites where live coral cover was low.

73 Butterflyfishes of the Southern Red Sea 67 Introduction The life cycle of most marine organisms includes a pelagic larval stage preceding a demersal, less mobile adult stage (Sale 1980, Fowler et al 1992). For the species concerned, the input of young individuals to the local population comes from settlement of pelagic larvae. In coral reef population studies recruitment refers to the addition of individuals to the benthic environment via settlement from the pelagic phase (Sale 1980). Recruitment frequently shows pronounced spatial and temporal variation (Sale 1980, Doherty 1988, Fowler et al 1992), which can strongly affect the distribution and abundance of a species as well as its population dynamics Variations in densities of reef fish and invertebrate recruits have been shown to exist between patch reefs or sites separated by tens or hundreds of meters (Luckhurst & Luckhurst 1983, Fowler et al 1992), among reefs or sites separated by several kilometres (Williams & Sale 1981, Doherty 1983, Sale et al 1984, Victor 1984), and between regions separated by hundreds or thousands of kilometres (Eckert 1984, Sale et al 1984, Victor 1984, Doherty 1987, Doherty & Williams 1988, Fowler et al 1992). Temporal recruitment variation can be interannual (Hawikns & Hartnoll 1982, Williams 1983, Eckert 1984, Sale et al 1984, Walsh 1987), seasonal (Walsh 1987), or diel (Doherty 1983). Numerous factors, including microhabitat characteristics (Bell & Galzin 1984, Eckert 1984, Sale et al 1984), resident fishes (Jones 1987), and larval supply (Victor 1986, Milicich et al 1992), cause variations in recruitment. The reef substratum provides food and shelter from predation (Shulman & Ogden 1987), as well as shelter from aggression from adult fishes (Sale 1980). Substratum characteristics can also affect the outcome of interspecific interactions (Ebersole 1985). Variation in the distribution and abundance of recruits can be caused by habitat choice during settlement (Sale et al 1984, Wellington 1992) or survival after settlement (Wellington 1992). Furthermore, postsettlement processes such as predation or competition may alter the initial distributions of recruits. Investigations on the relative importance of pelagic and demersal processes with an impact on recruitment have, for the most part, been limited to pomacentrid and labrid species (Doherty & Williams 1988), and mainly been conducted on the Great Barrier Reef or in the Caribbean (Mapstone & Fowler 1988). It is thus unclear how generally applicable the results from these studies are. Specifically, it is doubtful that the results from such studies can be applied to non-pomacentrid or non-labrid species of coral reef fish living in other parts of the world.

74 68 Chapter 8. Growth Patterns The objective of the present work is to investigate the recruitment patterns of chaetodontids and pomacanthids in the southern Red Sea. Attempts will be made to find answers to the following questions. Is recruitment in chaetodontids and pomacanthids seasonal? Is there inter-annual variation in recruitment? What are the differences in recruitment between chaetodontids and pomacanthids? Does the nature of the substrate play a role in the recruitment of these fish populations? Material and Methods Study site Recruitment surveys were conducted on the Resimedri and Twalot reefs near Massawa (Fig 7.1). Resimedri is a fringing reef, which extends for about 2000 m along the southeastern coast of Massawa proper. The depth of the reef drops gently from about 0.5 m on the reef flat to a maximum depth of 10 m. Mounds of Porites dominate the shallower part of the reef while encrusting and branching corals are abundant in the deeper part of the reef. Monogeneric stands of Montipora, Echinopora, and Pavona are common on the central part of the reef. Twalot reef is situated on the southern end of Twalot Island. It is a narrow fringing reef, which runs parallel to the shore for about 1000 m and extends to maximum depth of 4 m. Dead coral dominates the reef and sub-massive Montipora and Porites are the most abundant corals in the area. Dead coral is full of crevices providing shelter for young fish. The bottom below the reef base is covered with fine sediment, which prevents extensive growth of corals. Survey methods Recruitment data were collected using visual census while diving or snorkelling close to the substrate. Butterflyfish and angelfish encountered within the quadrats were counted and their sizes estimated to the nearest cm. Length estimation was facilitated by a cm-scale glued to a PVC plate. Occasionally samples of young fish were collected from nearby sites (using the anaesthetic quinaldine) to validate the size estimates. Comparisons between estimated and actual lengths showed that the former are one third higher than the actual lengths. Hence, all estimated lengths were accordingly reduced by one third.

75 Butterflyfishes of the Southern Red Sea 69 Figure 7.1. Study site. CS = comparison site; MP = Massawa proper; MS = monitoring site; RR = Resimedri reef; TR = Twalot reef; TW= Twalot. Growth of the species studied appears to be sufficiently rapid for an experienced observer to be able to distinguish one month's recruit from those that arrived during the previous month (see chapter 8). Hence, juvenile fish less than or equal to 4 cm in size are assumed to have recruited within a month. This method of estimating recruitment may underestimate recruitment in two ways (Williams 1983). First, transient individuals that both recruited and disappeared between successive censuses will be missed. Second, rapid replacement of individuals on the same site by conspecifics will be difficult to detect. However, in the study area juvenile fishes were not observed leaving the recruitment site. The only source of new recruits to the area was through settlement of larvae from the water column. Moreover, the mortality rate of newly recruited C. larvatus is relatively low during the first few weeks (see chapter 8). Restricted distribution of juveniles and the low mortality of recruits minimise the underestimation mentioned above. Recruitment data were collected from two sites on Resimedri reef and one site on Twalot reef. The first site on Resimedri reef (referred henceforth as the monitoring site) was monitored for over three years to investigate seasonal and inter-annual variations in recruitment. Spatial variation in recruitment was investigated by comparing the data recorded from this site with recruitment data collected from Twalot reef. The second site on Resimedri reef (referred to as the comparison site ) contained two stations with different coral composition. Recruitment data from the two stations were compared and the effect of the nature of coral on recruitment investigated. Due to the difference in the