The role of swimming in reef fish ecology

Transcription

1 The role of swimming in reef fish ecology Christopher J. Fulton School of Botany and Zoology, The Australian National University, Canberra, ACT 0200, Australia. I. Introduction Swimming plays a crucial role in the ecology and evolution of fishes. Swimming is the primary means by which fish interact and move through their environment. As such, swimming performance determines how well a fish can overcome the daily challenges of acquiring food, avoiding predators, gaining access to different habitats and conducting reproductive activities (Webb, 1994; Batty and Domenici, 2000; Plaut, 2001; Blake, 2004). Since swimming can also be one of the largest costs in the daily energy budget of fishes (Feldmeth and Jenkins, 1973; Kitchell, 1983; Boisclair and Tang, 1993), even small variations in swimming activity can have profound implications for patterns of energy allocation among growth, maintenance and reproduction (Koch and Wieser, 1983; Boisclair and Sirois, 1993). In this way, swimming can influence the evolutionary fitness of fishes and be a primary target for natural selection (Huey and Stevenson, 1979; Hertz et al., 1988). Fish are faced with numerous challenges for locomotion in shallow demersal systems such as coral and rocky reefs. These challenges can range from extracting food from complex benthic structures (Wainwright and Bellwood, 2002) and escaping high-speed attacks from hidden predators (Hixon and Beets, 1993), to negotiating wave-swept habitats characterised by extreme flow speeds and turbulence (Webb, 2004; Fulton and Bellwood, 2005). The magnitude of these challenges can vary substantially over space and time and interact with the swimming abilities of species to strongly influence their patterns of distribution and abundance (e.g. Bellwood and Wainwright, 2001; Fulton and Bellwood, 2004). Such environmental challenges may also represent an opportunity, with those species able to adapt their swimming abilities gaining access to new habitats and sources of food. While these direct links between swimming performance and fish ecology appear intuitive, until recently there has been a distinct lack of empirical evidence to support such direct biophysical 1

2 interactions in demersal habitats such as reefs (Arnold, 1983; Wainwright and Reilly, 1994; Webb, 1994; Blake, 2004). A major impediment to exploring links between swimming and fish ecology has been the doubt surrounding the ecological relevance of different performance metrics. While sustained speed (i.e. activity maintained for 200+ minutes; Blake, 2004) is generally thought to be the most relevant measure for questions concerning patterns of habitat-use in fish (Hammer, 1995; Plaut, 2001; Blake, 2004), there is less consensus on how best to measure this level of performance. One performance measure in particular, critical swimming speed (U crit ), has received considerable attention in this regard (Hammer, 1995; Plaut, 2001; Blake, 2004). First developed by Brett (1964), U crit is obtained from fishes subjected to incremental changes in speed over time in an experimental flume. While Brett (1964) originally intended U crit to provide a time-effective means of estimating maximum sustained speed, some studies have found U crit to overestimate sustained speed as a result of anaerobically powered performance in the final stages of the critical speed test (Kolok, 1991; Hammer, 1995; Plaut, 2001), unnatural swimming behaviours (Peake, 2004), restrictive experimental flumes (Peake, 2004), and/or variations in the specific stepwise changes in speed and time used in each trial (Farlinger and Beamish, 1977; Kolok, 1999; Plaut, 2001). Despite this, many other studies have found U crit to be a robust and reliable metric for comparing the relative swimming speed capabilities of fishes (Williams and Brett, 1987; Hartwell and Otto, 1991; Peake and McKinley, 1998; Hogan et al., 2007). Moreover, U crit has been found to relate directly to other aspects of swimming performance, with strong correlations to sustained (Johansen et al., 2007a) and routine swimming speeds in a wide range of fish taxa (Fisher et al., 2005; Fulton et al., 2005; Fulton, 2007). As such, U crit is now generally viewed as a robust measure of swimming speed performance, as measured over prolonged time periods (20 seconds minutes; Hammer, 1995; Plaut, 2001; Fisher et al., 2005; Fulton, 2007). In adult reef fishes, we do find that U crit overestimates sustained and routine field speeds by %, depending on the swimming mode employed (Fulton, 2007; Johansen et al., 2007a). Nonetheless, once allometric effects are taken into account, U crit also provides a strong predictor of routine field speeds measured in individuals swimming undisturbed on the reef (Fig. 1a; Fulton et al., 2005; Fulton, 2007). Moreover, an almost isometric relationship between lab- and field-based measurements of swimming speed may be achieved by ensuring lab-based estimates 2

3 are taken from individuals using the same gait as they would during their daily swimming activities. This can be accomplished by setting an alternative cut-off point in experimental speed tests, the gait transition speed (U trans ), which is the point where the fish shifts from either a pectoral-only to pectoral-caudal gait, or from a steadycaudal to burst-coast caudal gait (Fig. 1b; Drucker and Jensen, 1996; Walker and Westneat, 2002; Fulton et al., 2005; Cannas et al., 2006). While this concept may have limited utility for some groups such as larval fishes (Fisher and Leis, Chapter 12), the method appears robust in adult reef fishes from the 10 families examined to date (Fulton et al., 2005; Fulton, 2007). In discussing the ecological implications of swimming performance, I present both experimental (lab-based) and field-based measurements of speed performance, with U trans being the metric of choice when available. While manoeuvrability is undoubtedly a crucial aspect of swimming performance for fishes interacting with demersal habitats (Gerstner, 1999; Webb, 2002, 2004; Webb et al., Chapter 2), my main focus here is on the implications of speed performance for reef fish ecology. In this chapter I explore the role of swimming in shaping patterns of habitatuse in adult reef fishes. Firstly, I examine the diversity of swimming modes employed by reef fishes and how these may relate to the environmental challenges of living a reef-associated lifestyle. I then draw upon a decade of research to review our current understanding of labriform swimming and how this has provided new insights into the forces structuring coral and rocky reef fish assemblages around the globe. In particular, I explore the importance of fin morphology for patterns of performance, and how the acquisition of a unique fin shape may have been a key determinant in the success of labriform-swimming fishes on reefs. Finally, I discuss the importance of behaviour as a modulating factor in the link between fish swimming morphology, performance and patterns of resource use. Reef fishes are an ideal group to examine these concepts, given their wide diversity of body forms, modes of locomotion and evolutionary lineages, matched only by the diversity of available habitats and environmental conditions found in reef ecosystems. In several areas I am forced to make speculative assertions on the underlying mechanisms due to a lack of empirical evidence, particularly in the area of swimming energetics; many of these I expect to be adapted or rejected in light of future research. 3

4 II. Swimming modes and the reef-associated lifestyle Reefs are occupied by an abundance of fishes that appear to take full advantage of the diversity of food resources, home sites and predator refuges provided by these complex ecosystems. While occupying reefs may seem an ideal lifestyle for fishes, both coral and rocky reefs can be demanding environments that pose substantial challenges for fish locomotion. Wave-induced hydrodynamic forces can often exceed the equivalent of cyclone-force winds on land (Denny, 1988), and even under relatively calm conditions (<1 metre significant wave height), flow speeds in shallow reef habitats (<4 metres depth) can average cm s -1 and exceed 400 cm s -1 on a regular basis (Young, 1989; Symonds et al., 1995; Kench, 1998; Fulton and Bellwood, 2005; Gourlay and Colleter, 2005). Many reef fishes are simply unable to cope with the magnitude of these flows (Fulton, 2007); other species may be capable of such speeds, but excessive energy expenditure incurred in maintaining this level of performance may prove too costly for the long-term occupation of high-flow habitats (Clarke, 1992; Korsmeyer et al., 2002; Aseada et al., 2005; Fulton and Bellwood, 2005). Moreover, the unsteady and multidirectional nature of wave-induced flows can prove challenging for fishes to maintain stable swimming trajectories and complete fine-scale tasks such as extracting small food items from benthic microhabitats (Gerstner, 1999; Webb, 2002, 2004; Liao, 2007). Why would reef fishes want to gain access to shallow wave-swept habitats? Some of the richest food resources can be found in the most wave-exposed reef habitats. For example, loadings of detritus, a nutritious dietary resource targeted by many reef fish taxa (Wilson et al., 2003), is often greatest in the wave-swept reef crest and flat habitat zones (Crossman et al., 2001; Purcell and Bellwood, 2001; Wilson et al., 2003). Similarly, the biomass and productivity of the epilithic algal matrix (EAM), a resource actively targeted by most herbivorous reef fishes, is highest in habitats of high water movement (Carpenter et al., 1991; Steneck and Dethier, 1994; Purcell and Bellwood, 2001). Moreover, planktivorous fishes may find their suspended food items delivered at greater rates in high-flow, wave-swept environments (Holzman and Genin, 2003; Clarke et al., 2005). Besides an apparent abundance of food resources, fishes may also find less interspecific competition, with the total number of fish species within wave-swept reef flat habitats often being the lowest of all habitat zones in a given locality (Bellwood and Wainwright, 2001; 4

5 Fulton et al., 2001; Fulton and Bellwood, 2004). While gaining and maintaining access to high flow reef habitats can be a considerable biophysical challenge for fishes, the ecological advantages may also be substantial. Recognizing both the apparent desirability and difficulty of occupying waveswept habitats, we may ask which reef fishes can access such habitats and why? A good starting point in answering this question is to examine the different swimming modes utilized by reef fishes. Since the concept was first formalized by Breder (1926), swimming modes (otherwise known as gaits) have provided a primary framework for examining fish swimming behaviour and performance. Each mode involves the use of different fins or body regions during swimming, such as the employment of either the body and caudal fin (BCF) or the median-paired fins (MPF) for propulsion (Webb, 1984, 1994; Blake, 2004). In reef fishes we find a wide range of swimming modes being used during routine activities, ranging from steady BCF undulation (e.g. subcarangiform), to pectoral-augmented BCF (e.g. chaetodontiform) and rigid-body MPF propulsion using just the pectoral (e.g. labriform) or dorsoventral (e.g. balistiform) fins (Table 1; Webb, 1984, 1994; Blake, 2004; Fulton, 2007). Despite this diversity of behaviours, most reef fish taxa tend to use just one of three modes during routine swimming: labriform, subcarangiform, or chaetodontiform (Table 1; Fulton, 2007). Indeed, recent surveys of an entire reef fish assemblage (156 species, 7 families) revealed that more than 60% of reef-associated species use labriform propulsion during daily swimming activities (Fulton and Bellwood, 2005; Fulton, 2007). Furthermore, three of the most abundant reef fish families are composed primarily of labriform swimmers (the Labridae, Pomacentridae and Acanthuridae; Fulton, 2007). Consequently, reef-associated fish communities are characterized by labriform-swimming individuals, both in terms of numerical abundance and taxonomic diversity. Each swimming mode has inherently different performance traits and energetic costs, largely as a consequence of using different body parts and/or muscle groups to produce thrust (Webb, 1984, 1994; Weihs, 2002; Blake, 2004). Within the broad division of BCF (subcarangiform, chaetodontiform) and MPF (labriform) propulsion, we generally find greater slow-speed manoeuvrability in fishes using an MPF mode (Webb, 1984, 1994; Gerstner, 1999; Webb and Fairchild, 2001; Blake, 2004). Such high manoeuvrability is undoubtedly a useful trait for close interaction with complex reef habitats, and may help explain the prevalence of the labriform 5

6 (MPF) mode in reef fish assemblages. However, as discussed above, swimming speed performance may also be crucial for gaining access to wave-swept reef habitats. Traditionally, BCF-swimming has been viewed as the superior mode for attaining high swimming speeds, since MPF taxa seem to have a body shape and musculature optimized towards slow-speed manoeuvres (Alexander, 1967; Webb, 1984; Blake, 2004). However, this paradigm appears to have developed from comparisons of fishes of disparate size and lifestyle typically large pelagic or anadromous fishes (predominantly BCF swimmers) matched against small, bottom-associated taxa (a mix of MPF and BCF swimmers; Wardle, 1975; Lindsay, 1978; Walker and Westneat, 2002). When we compare the results of swimming performance trials in solely marine demersal taxa of equivalent size, we find that MPF-swimmers attain either similar or higher speeds than their BCF-swimming counterparts (Fig. 2). Moreover, these demersal MPF-swimmers maintain these high speeds while going about their daily swimming activities, with routine field speeds of 85% U crit in labriform-swimming species, versus field speeds of below 50% U crit in BCFswimming taxa (Fig. 2; Fulton, 2007; Johansen et al., 2007a). In addition to their taxonomic and numerical dominance of reef fish assemblages, labriform-swimming fishes also appear to be excellent swimming speed performers. Fundamental differences in propulsion and swimming performance, such as those described above, can have profound ecological implications for reef fishes. Evidence of this can be found in the distribution of fishes according to swimming mode across a typical coral reef. Here we find labriform-swimming fishes widely distributed across the full range of reef habitats, while BCF-swimmers are restricted mostly to sheltered, low-flow areas (Fig. 3; Fulton and Bellwood, 2005). A primary reason for this trend appears to be a direct interaction between wave-induced flows and the relative swimming speed performances of fishes. Indeed, the average swimming performance of fishes in a given habitat (regardless of species or mode) tends to be closely matched to the relative flow conditions in that habitat, such that habitats with higher flows are occupied by fishes with higher swimming performance (Fig. 4). Moreover, this interaction between flow velocity and swimming speed performance is found at multiple spatial scales. At the finer within-habitat scale, we find patterns of water-column use closely linked to variations in fish swimming mode and performance in a pattern that mimics the profile of flow velocities in the water column. Specifically, we find fishes occupying the upper layers of the water column 6

7 employing a labriform mode and exhibiting higher levels of speed performance for their size (Fig. 5). While this evidence suggests that swimming speed performance is a crucial determinant of habitat-use in reef fishes, such trends may not simply be due to differences in raw speed performance amongst modes. In fact, speed estimates taken in experimental trials indicate that fishes using both MPF and BCF swimming can attain the raw speeds needed to access wave-swept habitats (Fig. 2; Blake, 2004; Fulton, 2007). However, it seems that the critical performance aspect is their ability to maintain high routine speeds over daily timescales, with routine field speeds in labriform taxa regularly exceeding 40 cm s -1 versus less than 40 cm s -1 in BCFswimming fishes (Fig. 2; Fulton, 2007). Incidentally, this apparent speed threshold around 40 cm s -1 appears to coincide with an environmental threshold in the levels of ambient flow found on reefs, this being the typical lower boundary of flow velocities recorded in shallow, wave-swept habitats (Fig. 5b; Young, 1989; Symonds et al., 1995; Kench, 1998; Fulton and Bellwood, 2005; Gourlay and Colleter, 2005). As such, fishes incapable of maintaining routine speeds above this level could be simply excluded from wave-swept locations. Collectively, this evidence suggests that swimming speed performance may set overriding limits on the habitat-use patterns of reef fishes. In particular, the ability to maintain high speeds over daily timescales appears to be a primary determinant of whether a reef fish can occupy habitats exposed to high levels of wave-induced water motion. In this, labriform fishes appear to be ideally matched to the challenges of a high-flow environment and have taken their place as the dominant occupants of wave-swept habitats and locations on reefs. Notably, these links between swimming speed performance and habitat-use seem to apply regardless of any differences in fish trophic status. Indeed, fish from the same trophic group display markedly different swimming abilities that seem to match the flow conditions in their respective habitats. For instance, the most abundant planktivores on the relatively calm reef slopes, Pomacentrus brachialis and Cirrhilabrus punctatus, display average field speeds of 19.6 cm s -1 and 20.6 cm s -1 respectively, whereas Chromis atripectoralis and Thalassoma amblycephalum dominate wave-swept crest habitats with average speeds of 36.1 cm s -1 and 43.8 cm s - 1 respectively (Fulton, 2007). Likewise, in the roving herbivores and detritivores we find Ctenocheatus binotatus occurring primarily on the reef slope and displaying a mean field speed of 29.3 cm s -1, while C. striatus (43.1 cm s -1 ) and Acanthurus 7

8 nigrofuscus (45.6 cm s -1 ) occupy the crest in abundance, and A. triostegus (68.8 cm s - 1 ) dominate the most wave-swept reef flat habitat (Fulton, 2007). Spanning reef fish taxa from several perciform lineages (e.g. the Percoidei, Labroidei and Acanthuroidei; Nelson, 2006), it appears that the link between swimming mode, performance and patterns of habitat-use in reef fishes is a general phenomenon that transcends any differences in trophic status or taxonomic affiliation. Similar relationships between swimming performance and habitat-use have been found in freshwater fish communities, where freshwater fishes of greater swimming performance tend to occupy habitats of higher flow (e.g. Sagnes et al., 1997, 2000; Rosenfeld et al., 2000; Kodric-Brown and Nicoletto, 2005). While this selection of either low- or high-flow habitats has often been attributed to differential predation risk in freshwater systems (Power, 1984; Gilliam and Fraser, 1987; Lonzarich and Quinn, 1995), marked differences in fish growth and condition prompted several workers to examine the energetic implications of habitat choice. Certainly, the energetic cost of swimming increases exponentially with speed (Brett, 1964; Webb, 1975; Sepulveda et al., 2003; Alexander, 2005), and swimming can consume substantial amounts of energy on a daily basis (Feldmeth and Jenkins, 1973; Kitchell, 1983; Boisclair and Tang, 1993), so that even small changes in the cost of locomotion could play a large role in shaping fish habitat selection. Accordingly, the application of bioenergetic cost-benefit analyses, which contrast the cost of swimming against energy intake from invertebrate drift, revealed that avoidance of high-flow riffle habitats by freshwater fishes could in fact be driven primarily by locomotor energetics (Hughes and Dill, 1990; Hill and Grossman, 1993; Nislow et al., 2000; Rosenfeld and Boss, 2001). Moreover, similar bioenergetic analyses have indicated that many freshwater fishes may choose flow conditions that allow them to swim at an energetically optimal speed during upstream migrations and station-holding (Hinch and Rand, 1998; Haro et al., 2004; Castro-Santos, 2005). Could the energetic cost of locomotion be a determining factor in the habitatuse of reef fishes? While there is a paucity of information on the comparative energetics of reef fish swimming, relationships between cost-of-transport and speed in some representative MPF and BCF taxa provide some insight. Fish using a labriform (MPF) mode have been found to display minimum costs of transport across a broad range of swimming speeds ( BL s -1 ), while fish using a subcarangiform (BCF) mode display a narrow minima at relatively low speeds ( BL s -1 ) and steep 8

9 increases in cost-of-transport with increasing speed (Korsmeyer et al., 2002; Tolley and Torres, 2002; Lee et al., 2003; Cannas et al., 2006; Jones et al., 2007). Based on rates of oxygen consumption, these cost-of-transport relationships provide a relative measure of energy consumption during continuous swimming for a set distance or time period. As such, these relationships indicate that labriform swimming provides the most efficient means of swimming continuously over a broad range of speeds, while fishes of similar size using a subcarangiform (BCF) mode would incur much greater energetic costs. Certainly, multi-species comparisons of speed performance have found that labriform taxa display a wide range of swimming speed performances, experience lower drag as a result of rigid-body locomotion, and show a propensity for higher field speeds than demersal BCF taxa of the same size (Fig. 2; Walker and Westneat, 2000; Wainwright et al., 2002; Blake, 2004; Fulton, 2007). This ability to maintain high speeds over daily time periods with minimal energetic cost may be the crucial advantage that allows labriform fishes to successfully occupy high-flow habitats on reefs, while fishes using BCF modes of propulsion can not (Boisclair and Tang, 1993; Pettersson and Hedenström, 2000; Aseada et al., 2005). Although more empirical data on reef fish locomotor energetics is needed, it appears that energetic efficiency may be a critical factor in the link between fish swimming performance and patterns of habitat-use on reefs. Bio-energetic analyses such as those conducted in freshwater systems may go a long way towards explaining the high prevalence of labriform-swimming fishes in wave-swept reef habitats. Indeed, costbenefit analyses in planktivorous reef fishes has already provided some unique insights in patterns of habitat selection according to flow and food delivery (Clarke, 1992; Holzman and Genin, 2003; Clarke et al., 2005). Acknowledging that much work remains to be done on the comparative energetics of reef fish swimming before we can take these assertions any further, I now move on to explore the apparent success of labriform-swimming reef fishes in greater detail. III. Labriform swimming: the ecological implications of form and function Fishes that employ labriform locomotion during their daily swimming activities encompass an astounding level of diversity. Over two-thirds of the noncryptic fish species living on reefs adopt labriform locomotion during their routine swimming (Fulton and Bellwood, 2005; Fulton et al., 2005; Fulton, 2007). Moreover, 9

10 these species span several families (e.g. the Acanthuridae, Labridae, Pomacanthidae, Pomacentridae, Scaridae, Zanclidae; Table 1), an enormous range of body sizes (e.g cm total length; Wainwright et al., 2002), trophic groups (e.g. corallivores, detritivores, herbivores, planktivores, ectoparasite cleaners; Bellwood et al., 2006) and reproductive modes (e.g. haremic societies, pair-wise brooding, broadcast spawning; Randall et al., 1997). Indeed, it appears that the only common attribute shared by these fishes is that they use solely their pectoral fins to produce thrust during routine swimming activities (Webb, 1994; Blake, 2004; Fulton, 2007). Using this unifying trait and a remarkable link between pectoral fin shape and swimming performance, we have been able to transcend these vast differences in lifestyle and uncover some of the general mechanisms shaping reef fish communities at a global scale. Our understanding of labriform locomotion has increased dramatically over the past decade, largely due to the considerable attention devoted towards the consequences of pectoral fin shape for locomotor function and performance. Labriform-swimming fishes display a wide spectrum of pectoral fin shapes, ranging from deeply rounded to highly tapered fins (Fig. 6; Wainwright et al., 2002; Fulton et al., 2005). Kinematic analyses have revealed that these variations in fin shape are linked to two different forms of fin movement and thrust production: drag-based rowing using rounded fins, versus lift-based flapping using tapered fins (Blake, 1979, 1981; Drucker and Jensen, 1996; Walker and Westneat, 1997, 2002; Walker, 2004). As the names suggest, the former involves rowing the fin to exploit drag-based forces on a backward power stroke, while the latter involves flapping the fin in a figure-eight sweep to exploit lift-based thrust on both upward and downward strokes (Drucker and Lauder, 1999; Walker and Westneat, 2000, 2002). Since thrust is produced on only half of the fin strokes in drag-based locomotion (i.e. no thrust produced by forward recovery stroke), this form of labriform swimming is effective only over a limited range of slow speeds (Blake, 1981; Vogel 1994; Drucker and Lauder, 1999). Conversely, lift-based flapping produces thrust on all fin strokes and provides a highly efficient means of maintaining very high swimming speeds, albeit at the cost of less slow-speed power and acceleration (Vogel 1994; Walker and Westneat, 2000, 2002). As a result, we find that tapered-fin species tend to swim faster than rounded-fin species of the same size, both under experimental conditions and while swimming undisturbed on the reef (Walker and Westneat, 2000; Wainwright et al., 2002). 10

11 Representing two extremes at either end of a continuum, this trade-off in performance with fin shape produces a spectrum of swimming speeds among species, with even relatively small changes in fin shape having a marked impact on speed performance. Consequently, we find a consistent functional link between speed and pectoral fin shape in labriform fishes, regardless of their taxonomic affiliation (Fig. 6; Wainwright et al., 2002; Fulton et al., 2005). While highlighting the unifying importance of morphology for variations in swimming speed performance, this functional relationship also allows us to predict the relative performance of labriform fishes based solely on their pectoral fin shape. Applying this functional relationship to patterns of ecology, we find that fin shape also provides a strong predictor of habitat-use across gradients of water motion. Whether it is height in the water-column or the occupation of reef habitats of different wave energy, we find a distinct trend in the distribution of labriform reef fishes according to their fin shape. Specifically, we find tapered-fin species occurring in high abundance in high-flow habitats, while rounded-fin species are restricted to sheltered, low-flow areas (Fig. 7; Bellwood and Wainwright, 2001; Fulton et al., 2001). In a pattern that reflects the differential distribution of swimming modes described in section II above, this trend appears to be the result of differences in swimming performance according to pectoral fin type. Namely, tapered-fin species are capable of high levels of speed performance with high efficiency, whereas rounded-fin species are unable to maintain the high speeds needed to occupy highflow habitats (Vogel, 1995; Wainwright et al., 2002; Walker and Westneat, 2000, 2002). Notably, this dichotomous distribution is found in reef fish assemblages throughout the world, spanning high diversity fish assemblages on the Great Barrier Reef to intermediate and low diversity systems in French Polynesia and the Caribbean (Fig. 7; Bellwood et al., 2002). Despite distinct variations in species composition and a five-fold difference in species richness, labriform fishes in each of these locations are arranged in highly congruent patterns according to pectoral fin shape and the relative wave energy of habitats (Fig. 7). This phenomenon is driven primarily by the relative abundance of individuals, rather than a simple presence-absence of species (Bellwood et al., 2002). This is because relatively few tapered-fin species occupy wave-swept habitats, but do so in extremely huge abundances. Indeed, population sizes of tapered-fin species are typically an order of magnitude higher than roundedfin species in the same habitats. For example, the tapered-fin species Thalassoma 11

12 hardwicke (aspect-ratio (AR) = 1.90) and T. jansenii (AR = 1.96) can attain average densities of 16 individuals per 100 m 2 on wave-exposed reef crests of the Great Barrier Reef, while rounded-fin species such as Cheilinus chlorurus (AR = 0.97) and Halichoeres melanurus (AR = 1.11) display densities of 0.8 per 100 m 2 in the same crest habitats (Bellwood and Wainwright, 2001; Fulton et al., 2001). Even greater differences are found in open-ocean systems such as French Polynesia, where we find densities of up to 380 per 100 m 2 (T. quinquevitattum, AR = 1.97) versus 0.2 per 100 m 2 (C. trilobatus, AR = 0.90) in species occupying reef crests exposed to oceanic swells (Bellwood et al., 2002). Taking population size as a measure of success, it would seem that labriform taxa with highly tapered pectoral fins are the most successful group of fishes found on reefs. Reef fish assemblages in temperate latitudes display a notable variation on this general theme. At first glance, it appears that temperate fish assemblages exhibit a similar link between fin shape and wave energy to that found in coral reef locations (Fig. 7). However, increases in fin aspect-ratio with increasing wave energy appears to be less pronounced in the temperate reef fish assemblage occupying offshore reefs near Port Stephens, south-east Australia (Fig. 7). This is not due to a breakdown in the functional link between fin shape and speed performance in temperate species. Indeed, we find a similarly strong relationship between fin shape and speed performance to that seen in tropical reef fishes, despite the fact that the temperate assemblage contains several unique taxa (e.g. Achoerodus, Ophthalmolepis, Notalabrus; Fulton and Bellwood, 2004). Moreover, rounded-fin temperate species also display similar patterns of habitat-use to their tropical contemporaries, with low AR species largely restricted to low-flow, sheltered rocky habitats (Fulton et al., 2001; Fulton and Bellwood, 2004; Denny, 2005). Where we do find marked differences between the tropical-temperate assemblages is in wave-exposed habitats: temperate reefs are almost devoid of fishes in wave-swept areas, while similar habitats on coral reefs are replete with tapered-fin individuals (Fulton and Bellwood, 2004; Fulton et al., 2005). The reason for these differences appears to be that species with highly tapered pectoral fins are not present in temperate fish assemblages, with the maximum fin aspect-ratio in temperate species (AR = 1.43 in Coris picta; Fulton and Bellwood, 2004) being markedly lower then the maximum found in tropical species (e.g. AR = 2.1 in Stethojulis bandanensis; Wainwright et al., 2002). Unlike tropical taxa, a highly tapered fin shape has not evolved within temperate reef fish 12

13 assemblages. As a consequence, only temperate fishes which grow to a large size seem able to attain the high speeds needed to access wave-swept habitats (Fulton and Bellwood, 2004). Certainly, those few fishes found in exposed rocky reef habitats are much larger (26-30 cm total length) than individuals in more sheltered areas (14-18 cm total length; Fulton and Bellwood, 2004), which is in distinct contrast to the consistent body size of coral reef fishes distributed across wave energy environments (7-9 cm total length; Bellwood and Wainwright, 2001; Fulton and Bellwood, 2004). Despite this alternative mechanism, we see very few individuals in wave-swept habitats on rocky reefs, possibly because of the long timeframe and high degree of mortality incurred in reaching this large body size (Fulton and Bellwood, 2004; Denny, 2005). Ultimately, temperate systems provide an example of an assemblage devoid of tapered-fin fishes and the striking results a complete lack of abundant fish populations in wave-swept habitats. This raises some pertinent questions about the evolution of labriform locomotion in reef fishes. Could the acquisition of a tapered fin and the lift-based mechanism of propulsion facilitated the expansion of reef fishes into wave-swept habitats, and if so, why has this not occurred in temperate reef fish assemblages? While the evolution of labriform swimming remains to be fully resolved (Drucker and Lauder, 2002; Thorsen and Westneat, 2005), ontogenetic evidence suggests that tapered pectoral fins are a relatively derived trait. All of the juvenile stages of labriform taxa examined to date have a generalized, rounded pectoral fin shape (Fig. 8; Fulton and Bellwood, 2002; Wainwright et al., 2002), with only those species having tapered fins as adults acquiring this fin shape through ontogenetic development (Fig. 8; Fulton and Bellwood, 2002; Wainwright et al., 2002). Current phylogenetic evidence also indicates that tapered fins and the lift-based mode of locomotion are relatively derived traits (Drucker and Lauder, 2002), which appear to have arisen independently in at least two perciform lineages: the Acanthuroidei and Labroidei (Fulton et al., 2005; Nelson, 2006). Moreover, there seems to be multiple origins of lift-based swimming within these lineages, with up to three in the Pomacentridae (Abudefduf, Chromis-Dascyllus and Neopomacentrus-Pomacentrus; Quenouille et al., 2004), six within the Labridae (Cirrhilabrus, Chlorurus-Scarus, Halichoeres, Labroides, Thalassoma-Gomphosus and Stethojulis; Wainwright et al., 2002; Thorsen and Westneat, 2005; Westneat and Alfaro, 2005), and at least two in the Acanthuroidei (Acanthurus-Ctenochaetus and Zanclus; Clements et al., 2003). 13

14 Despite these multiple origins, all the lift-based lineages seem to have evolved within the tropics. Although temperate assemblages largely consist of relatively basal fish taxa (Westneat and Alfaro, 2005), it is unclear why lift-based lineages have not expanded into the seemingly empty wave-swept habitats found on temperate reefs. It is possible that the advantages associated with lift-based locomotion do not apply in cold temperate waters. Alternatively, the generally larger body sizes found in temperate fish taxa may make body size a more readily accessible trait for these fishes to attain the swimming speeds needed to access wave-swept habitats (Fulton and Bellwood, 2004; Denny, 2005). An examination of a sub-tropical fish assemblage, where tropical fish species occur in a relatively cool climate, may provide some more clarity to these issues. At least in tropical systems, the acquisition of a tapered fin appears to have been a key innovation in the evolution of reef fishes; an innovation that may ultimately have facilitated the ecological expansion of labriform fishes into wave-swept reef habitats. Dominance of wave-swept habitats by relatively few unique fish species could have profound implications for the health and function of reef ecosystems. Waveswept habitat zones such as the shallow crest and flat are often areas of high benthic diversity and live coral cover (Done, 1983; Steneck and Dethier, 1994; Hughes and Connell, 1999). Reef fishes play an important part in preserving this high diversity in the shallows via their role as predators of benthic organisms, helping to maintain a competitive balance among algae, corals and other benthic invertebrates (Hughes, 1994; Bellwood et al., 2004; Mumby, 2006; Hughes et al., 2007). In fact, the two dominant fish genera in wave-swept reef habitats, Thalassoma and Scarus, are tapered-fin fishes that feed upon a range of benthic invertebrates, dead coral and algae (Bellwood et al., 2006; Hoey and Bellwood, 2008). Due to their extremely high abundances, these fishes have a substantial impact on the presence and abundance of benthic organisms in shallow reef habitats (Hoey and Bellwood, 2008). Given that access to such wave-swept areas is dependent on possession of a tapered fin shape and high speed performance, the loss of any fish species in these areas may not be easily replaced certainly not by rounded-fin species coming from more sheltered areas. Consequently, if we were to lose our unique tapered-fin fishes from wave-swept reef habitats, we may see a breakdown in reef ecosystem function that results in a severe loss of biodiversity and habitat degradation (Mumby, 2006; Bellwood et al., 2004; Hughes et al., 2007). 14

15 Of course, the severity of this issue is greatly dependent on whether fin shape and swimming performance is fixed within each species according to their phylogenetic history. Can reef fishes express flexibility in their fin shape and performance? Locomotion has been found to be a highly plastic trait in many vertebrates, including frogs (Parichy and Kaplan, 1995), snakes (Jayne and Bennett, 1990; Aubret, 2005), lizards (Losos et al., 1997, 2000) and some freshwater fishes (Nelson et al., 2003; Peres-Neto and Magnan, 2004). Preliminary evidence in reef fishes also suggests the possibility of locomotor flexibility. In one widely distributed species, Acanthochromis polyacanthus (Spiny Damselfish), we find that juveniles display a similarly rounded fin morphology, regardless of their wave energy location (average fin aspect-ratio = 1.24 ± 0.03 SE). However, fully-grown adults inhabiting different wave energy environments display markedly different fin shapes (Fig. 9) in a pattern that matches their local environment. Specifically, adult fish of this species display more tapered fins (and potential for higher swimming speeds) in habitats of higher wave energy (Fig. 9). Presumably, these intraspecific variations in fin shape arise from variations in the fin ontogeny of individuals living in different habitats (Fulton and Bellwood, 2002). However, the question remains as to whether this may be due to phenotypic plasticity, or genetic differentiation among separate populations in each wave energy environment (i.e. local adaptation; Via, 2001; Bradshaw, 2006). These alternative scenarios of phenotypic plasticity versus local adaptation are highly contingent on dispersal mode and extent of gene flow across the spatial scale of different wave energy environments. Theory suggests that if there is poor dispersal and restricted gene flow, local adaptation may be the most likely explanation (Via 2001; Coyne and Orr, 2004). However, if spatial heterogeneity in the selective environments is sufficiently smaller than the dispersal capabilities of species, phenotypic plasticity can arise as an advantageous trait (West-Eberhard, 1989; Sultan and Spencer, 2002; Bradshaw, 2006). Indeed, A. polyacanthus displays relatively high levels of gene flow at the within-reef scale where these intraspecific variations in fin shape have been detected (Bay, 2007). While this suggests that phenotypic plasticity may be the most likely mechanism behind intraspecific variations in fin phenotype, this would need to be verified by a suite of common-garden experiments. If confirmed, such flexibility may provide further explanation for why labriformswimming fishes occupy such a wide diversity of reef habitats in high abundance. 15

16 IV. Behavioural mechanisms: fishes swimming beyond their limits Matching morphology to patterns of ecology has provided some important insights into the forces structuring patterns of resource use in fishes (review in Wainwright and Reilly, 1994). However, the impact of behaviour must not be overlooked when examining the relationship between swimming and habitat-use in reef fishes. While morphology may set overriding limits on the range of resources a species can exploit, behaviour ultimately determines how this morphology is used to interact with their environment (Arnold, 1983; Wainwright and Reilly, 1994). As such, behavioural mechanisms can sometimes produce patterns of resource use that significantly depart from predictions based solely on morphology or physiology alone. For example, many labroid fishes have the pharyngeal jaw strength needed to crush hard-bodied prey, but behavioural selection of prey types often results in soft-bodied organisms being the dominant prey item consumed (Wainwright and Richard, 1995; Bellwood et al., 2006). Similarly, small fishes may have relatively poor swimming capabilities when compared to their larger predators, but the use of anti-predator behaviours can substantially increase their likelihood of survival (Langerhans et al., 2004; Lethiniemi, 2005; Masuda, 2006). Behaviour can also be used to reduce the environmental demands placed on fishes, helping them to overcome barriers that may have otherwise curtailed their patterns of ecology. Schooling is one such behaviour, where fish significantly increase their travel time/distance by staying close to conspecifics and exploiting downstream vortices to reduce costs of locomotion (Parker, 1973; Ross and Backman, 1992; Herskin and Steffensen, 1998; Liao et al., 2003). Likewise, flow avoidance behaviour such as refuging can reduce the flows experienced by a fish to substantially decrease the energy expenditure needed to station-hold within fast-flowing habitats (Gerstner, 1998; Gerstner and Webb, 1998; Webb, 2006). Such behavioural mechanisms can provide a powerful means of shaping the boundaries of resource use, seemingly pushing fishes beyond the limits imposed by their morphology and physiology. Behaviour is undoubtedly an important factor in shaping patterns of fish habitat-use across wave energy gradients. Despite a general trend that only faster swimming fishes occupy the most wave-swept reef habitats (section II and III above), we do find several species occupying wave-swept habitats which have a morphology and body size that suggests very poor swimming abilities. While these slow swimmers 16

17 tend to occur in wave-swept habitats in low abundance, we might question how they can persist in such high flow habitats with a seemingly inadequate swimming ability? An answer may be found in their behavioural choice of water-column position, with all of the slow-swimming taxa in wave-swept habitats tending to stay very close to the substratum (Fig. 5a; Fulton et al., 2001; Fulton and Bellwood, 2002; Johansen et al., 2007b). By maintaining close proximity to the reef, these fishes would experience reduced flow speeds due to a phenomenon known as the boundary-layer effect. The boundary-layer effect is where friction between the water and the substratum reduces wave-induced water movements so that flows immediately adjacent to the reef are reduced by around 20-35% below ambient (Fig. 5b; Symonds et al., 1995; Kench, 1998; Gourlay and Colleter, 2005; Reidenbach et al., 2006). Such variations in watercolumn use and flow with swimming ability are not unique to reef fish assemblages, and have been found in a wide range of marine and freshwater systems (e.g. Chipps et al., 1994; Gerstner and Webb, 1998; Gatwicke and Speight, 2005; Webb, 2006). Flow avoidance, however, may not be the only reason for refuging behaviour in reef fishes. An alternative hypothesis is that refuging is primarily a predatoravoidance mechanism. Certainly, most refuging behaviour in reef fishes has been attributed to predation risk, with any reduction in flow stress considered to be adjunct and/or insignificant (Hixon and Beets, 1993; Beukers-Stewart and Jones, 1997; Lethiniemi, 2005). However, recent experimental work has found that not only do substratum refuges substantially reduce the environmental demands placed on the swimming abilities of fishes, such refuging may be the sole reason why slowswimming fishes can occupy high-flow habitats (Fulton et al., 2001; Fulton and Bellwood, 2002; Webb, 2006; Johansen et al., 2007b, in press). Even relatively simple structures such as sand ripples can significantly reduce the flow speeds experienced by fishes and increase their station-holding capabilities ten-fold (Gerstner, 1998; Gerstner and Webb, 1998). More complex benthic structures, such as the holes and coral heads commonly found in a coral reef matrix, have also been found to dramatically reduce flow speeds, with typically flow speed reductions of 60 80% below ambient (Fig. 10a; Johansen et al., 2007b, in press). Rather than maintaining a swimming speed of 60+ cm s -1 in the ambient flows of a wave-swept habitat (Fulton and Bellwood, 2005), fishes occupying such benthic refuges would only need to maintain a speed of 24 cm s -1 or less to hold their position. This substantial reduction, and the associated energy savings, may mean the difference 17

18 between whether or not a slow-swimming species can persist in such high-flow locations. Indeed, this certainly seems the case in at least two reef fish species whose field swimming speed capabilities are significantly slower than the ambient flows found on the coral reef flats where they commonly reside. These two species, Halichoeres margaritaceus (average routine field speed of 48 cm s -1 ) and Pomacentrus chrysurus (27 cm s -1 ; Fulton, 2007), are incapable of holding station for longer than 1 hour when placed in an experimental flow of 60 cm s -1 without access to a refuge (Fig. 10b; Johansen et al., 2007b). However, when presented with a hole refuge, these species were able to increase their station-holding time by more than 400% and maintain their swimming activity for 35+ hours (Fig. 10b; Johansen et al., 2007b). Conducted in the absence of predators, these flow tank experiments suggest that such refuging behaviour is largely due to the need for flow-avoidance in order to hold their position in high flows (Johansen et al., 2007b, in press). Without such flowrefuging behaviour, it appears that these two species would simply be unable to occupy wave-swept habitats over the long term. Ultimately, behaviour has allowed these reef fish species to expand their ecological limits beyond what we may otherwise have predicted from their swimming speed capabilities. V. Conclusions Reefs are both challenging and rewarding environments for reef fishes. Waveinduced water motion creates gradients of environmental stress that directly interact with the swimming capabilities of species to shape their patterns of distribution and abundance on reefs. Swimming speed performance appears to be the critical trait that determines these patterns of habitat-use, with clear and consistent trends between habitat-specific levels of water motion and the swimming speed performance of resident fishes. Given the generality of this relationship in fishes from multiple perciform lineages and locations around the world, wave energy seems to impose a strong and overriding force on reef fishes, which applies regardless of any differences in lifestyle or taxonomy. As a consequence, reef fishes are both adapted, and restricted, to a particular suite of habitats according to their inherent swimming capabilities. Of course, this assertion would depend on the extent to which swimming morphology and performance is fixed within a species. While there is some evidence to suggest that reef fishes may show some flexibility in their swimming morphology 18

19 and performance, the specific mechanism remains to be determined. Determining whether fishes can use the relatively rapid mechanism of phenotypic plasticity to change their swimming morphology and performance, or whether such changes have been acquired through longer-term genetic differentiation and local adaptation should be a primary focus for future research. One group of reef fishes in particular have emerged as the dominant occupants of reefs. Using solely their pectoral fins to produce thrust, these so-called labriformswimming fishes occupy the full range of reef habitats available to fishes, from the deepest reef slopes to the shallowest wave-swept areas of coral and rocky reefs alike. In particular, one sub-group of labriform-swimming fishes those with highly tapered pectoral fins seem most suited to the challenges of a wave-swept existence. Using the equivalent of underwater flight, these species use their tapered pectoral fins to exploit lift-based thrust and maintain very high speeds with little energetic cost. In a pattern that is repeated in reef fish assemblages throughout the globe, these taperedfin reef fishes completely dominate wave-swept habitats and attain the largest population sizes. Combining high speed performance with efficiency and flexibility, labriform swimming appears to provide a highly versatile mode of swimming that is ideally suited to the challenges of a reef-associated lifestyle. Undoubtedly, further research into this unique group, particularly in the area of locomotor energetics, should provide further insights into the mechanisms that have shaped the evolution and ecology of fishes on reefs. Acknowledgments Many people have contributed to the vast body of research on the biology and ecology of swimming in reef fishes. In particular, I have benefited greatly from interactions with D. Bellwood, R. Blake, R. Fisher, J. Johansen, J. Steffensen, P. Wainwright, J. Walker, P. Webb, M. Westneat and their published research in this field. This work was supported by funding from the Australian Research Council, Lizard Island Reef Research Foundation and the Australian National University. 19

20 Table 1. Diversity of swimming modes employed during the daily swimming activities of reef fishes. Based on field observations of fin and body use following Fulton (2007). Swimming mode nomenclature and definitions follow Webb (1994), images adapted from Randall et al. (1997). Swimming mode Source of propulsion Reef fish Family or Genera Labriform Pectoral fins Acanthuridae, Labridae, Pomacentridae, Pomacanthus, Scaridae, Zanclidae Subcarangiform Body-caudal fin Haemulidae, Kyphosidae, Lethrinidae, Lutjanidae, Naso, Serranidae, Siganidae, Mullidae, Scorpaenidae Chaetodontiform Pectoral-caudal fins Apogonidae, Chaetodontidae, Centropyge, Dischistodus, Ephippidae, Nemipteridae, Pygoplites, Stegastes Carangiform Caudal fin Carangidae, Caesionidae, Balastiform Dorsal-ventral fins (broad) Balistidae, Monacanthidae, Syngnathidae Tetraodontiform Dorsal-ventral fins (narrow) Tetraodontidae, Ostraciidae Anguilliform Body undulation Muraenidae, Plotosidae 20