Ecology, 89(7), 2008, pp. 1994–2004 Ó 2008 by the Ecological Society of America DENSITY-DEPENDENT SETTLEMENT AND MORTALITY STRUCTURE THE EARLIEST LIFE PHASES OF A CORAL POPULATION MARK J. A. VERMEIJ1,3 1 AND STUART A. SANDIN2 Cooperative Institute for Marine and Atmospheric Studies, Rosenstiel School for Marine and Atmospheric Sciences, University of Miami, 4600 Rickenbacker Causeway, Miami, Florida 33149 USA 2 Center for Marine Biodiversity and Conservation, Scripps Institution of Oceanography, 9500 Gilman Drive, La Jolla, California 92093-0202 USA Abstract. The local densities of heterospecifics and conspecifics are known to have profound effects on the dynamics of many benthic species, including rates of settlement and early post-settlement survivorship. We described the early life history of the Caribbean coral, Siderastrea radians by tracking the population dynamics from recently settled planulae to juveniles. Through three years of observation, settlement correlated with the abundance of other benthic organisms, principally turf algae (negatively) and crustose coralline algae (positively). In addition, adult density showed independent effects on coral settlement and early post-settlement survivorship. Settlement rates increased across low levels of adult cover and saturated at a maximum around 10% cover. Early post-settlement survivorship decreased with adult cover, revealing structuring density dependence in coral settlers. The earliest life stages of corals are defined by low survivorship, with survivorship increasing appreciably with colony size. However, recent settlers (one-polyp individuals, ,1 year old) are more likely to grow into two-polyp juveniles than older single polyps (.1 year old) that were delayed in their development. The early benthic phase of corals is defined by a severe demographic bottleneck for S. radians, with appreciable density-dependent and density-independent effects on survivorship. For effective management and restoration of globally imperiled coral reefs, we must focus more attention on this little studied, but dynamic, early life history period of corals. Key words: coral reef; crustose coralline algae; Florida keys; invertebrates; mortality; population dynamics; recruitment; settlement; Siderastrea radians; turf algae. INTRODUCTION The early life history period of marine species is perhaps the most critical and dynamic, though least understood, of all life stages. For benthic species, settlement after the pelagic larval period is variable in space and time and therefore is frequently deemed unpredictable. Following arrival to the benthos a severe survivorship bottleneck faces settling larvae, with mortality frequently exceeding 99% especially in marine invertebrates that lack parental care of young and have small, little-developed larvae (Gosselin and Qian 1997, Hunt and Scheibling 1997). Settlement and subsequent mortality of benthicassociated species can be affected by characteristics of the local environment in a density-independent manner. For example, early survivorship of some species has been shown to be affected by water temperature (Pepin 1991), salinity (Vermeij et al. 2006), sedimentation (Gilmour 1999), and availability of refuges (Tupper and Boutilier 1995, Anderson 2001) or preferred Manuscript received 8 August 2007; revised 14 November 2007; accepted 20 November 2007. Corresponding Editor: R. B. Aronson. 3 Present address: Department of Botany, University of Hawaii at Manoa, 3190 Maile Way, Honolulu, Hawaii 96822 USA. E-mail: [email protected] settlement substrata (Harrington et al. 2004). Densitydependent effects also can increase or decrease population growth rates with increasing population size (Caley et al. 1996). In a variety of taxa, increased conspecific density can lead to an increase in rates of settlement. Newly arriving individuals can demonstrate attraction to sites where adults are already present (fish [Sweatman 1988]; barnacles [Chabot and Bourget 1988]; annelids [Minchinton 1997]) or limited dispersal capabilities can result in a positive relationship between the numbers of settlers and reproductive adults at a local spatial scale (Molofsky 1994, Hughes et al. 2000, Johst et al. 2002, Vermeij 2005). In addition to the structuring effects on settlement, increased conspecific density can lead to an increase in per capita mortality probabilities for young settlers (barnacles [Miron et al. 1999]; fish [Hixon and Webster 2002]; algae [Reed 1990, Creed et. al. 1996]). The agents of such direct density dependence can include aggregating predators (Hixon and Carr 1997, Anderson 2001), competition for space (Roughgarden et al. 1985, Carlon 2001), or distance-related disease transmission (Bruckner et al. 1997, Lafferty et al. 2004). Difficulties associated with studying early life history stages in corals Here we aim to dissect demographic patterns within distinct phases of the early life history of a Caribbean 1994 July 2008 EARLY LIFE HISTORY DYNAMICS OF CORALS scleractinian coral in order to provide insight into the likely factors structuring coral population dynamics. The conservation of coral reefs depends on a comprehensive understanding of the earliest life history dynamics of corals. However, three issues have constrained such research. (1) The high topographic complexity for which coral reefs are notorious makes true coral settlers, i.e., planulae first settling to the benthic environment, extremely difficult to find and to quantify in situ. For example, coral recruits frequently are defined operationally as individuals of small size (e.g., ,5 cm2) in one-time censuses across space (see notable exceptions in time-series sampling, e.g., Connell et al. [1997], Hughes and Tanner [2000], Miller et al. [2000]). These small individuals, however, are likely to be more than one year old and hence will not reveal insights into the earliest (and likely most dramatic) demographic bottlenecks (Caley et al. 1996, Vermeij 2006). (2) Very small corals typically lack appreciable morphological patterning to enable reliable taxonomic identification in the field. Therefore repeated censuses through time will reveal survivorship information of the aggregated coral assemblage, but not species-specific demographic information. (3) Coral settlement is notoriously variable through time and across space (e.g., Wallace 1985, Sammarco 1991, Connell et al. 1997, Smith 1997, Dunstan and Johnson 1998, Hughes et al. 2000), thereby complicating statistically rigorous interpretation of early life history dynamics. Variability is especially notable for broadcast spawning species whose pelagic larvae are subject to the ephemeral transport and delivery mechanisms defined by local oceanographic conditions. An extensive body of research using settlement tiles has successfully navigated a number of these concerns to document spatial and temporal patterns of coral recruitment (e.g., Sammarco 1991, Smith 1992, 1997, Harriott 1999, Carlon 2001, Hughes et al. 2002). However, detailed post-settlement survivorship studies through time remain little addressed. A 60þ-year-old wreck off the coast of the Florida Keys provided an experimental reef enabling the study of the earliest life phases of Siderastrea radians, a common Caribbean scleractinian coral. This study system allowed us to accurately relate settlement and early post-settlement mortality rates to characteristics of the local environment in a Caribbean coral population, namely densities of conspecifics and cover of two dominant benthic algae. MATERIALS AND METHODS Study site The study site was a shipwreck, the MV Benwood, located on a shallow fore-reef in the Florida Keys Natural Marine Sanctuary near Key Largo (Florida, USA; wreck coordinates: 2583 0 700 N, 80819 0 5900 W). The Benwood is a 110-m, metal-hulled freighter that has been submerged since 1942. Its deck (approximately 110 3 10 m) is parallel to the water surface at 10 m depth. The 1995 metal deck is no longer exposed, but is encrusted primarily with a 1–3 cm thick layer of crustose coralline algae (CCA; mainly species of Porolithon and Paragoniolithon) on which other substrate types (predominately turf algae and adult S. radians colonies) can be locally abundant. The surfaces of the wreck offered a virtually twodimensional limestone substrate characterized by smallscale topographical complexity (millimeters to centimeters) upon which to reliably find all size classes of corals, including recently settled planulae. The coral community of the wreck was dominated by S. radians, as the single species comprised over 95% of the colonies present. The remaining coral colonies included large colonies of broadcast-spawning hard corals (.50 cm diameter) and gorgonians (.30 cm height). Coral assemblages dominated by S. radians were observed in Jamaica and Bermuda as long as a century ago (Duerden 1904 and references therein). More recent surveys report a similar coral assemblage in many reef flat environments (i.e., shallow, near-horizontal habitats) throughout the tropical Atlantic (Virgin Islands [Lewis 1989], south Florida [Chiappone and Sullivan 1994], Colombia and Haiti [González-Cortés, unpublished data], Netherlands Antilles [Vermeij et al. 2007], Cape Verde Islands [Moses et al. 2003]). Focal species The Caribbean coral species Siderastrea radians (Pallas 1766; Family Siderastreidae) occurs on shallow limestone pavements and ledges throughout the Tropical Atlantic where it dominates coral communities in naturally disturbed habitats (e.g., areas with high sedimentation, temperature, and/or salinity fluctuations [Lewis 1989, Lirman et al. 2002, Moses et al. 2003, Vermeij et al. 2007]). The species is gonochoric and females release competent planulae year-round (Soong 1991, Vermeij 2005). Colonies are reproductively active at sizes larger than 2.23 cm2 and are thus considered to be adults. Beyond this size, total planular production increases exponentially with colony size (Vermeij 2005). At a 0.25-m2 scale, 46% of the settling planulae were produced by local adults (Vermeij 2005), suggesting that S. radians populations are largely demographically closed at this small spatial scale (sensu Hixon et al. 2002). Planulae of S. radians are relatively large (maximum length 0.7 mm [Vermeij 2005]) and have a ring of zooxanthellae at their oral pole, facing away from the substrate after settlement. Both their large size and the characteristic ring of zooxanthellae make S. radians planulae easily identifiable during underwater surveys (see Plate 1). Furthermore, none of the surviving settlers that we initially identified as S. radians were found to be misidentified after early growth. S. radians is a brooding species with frequent reproduction and a short typical dispersal distance (Vermeij 2005). These life-history characteristics are common to at least eight other Caribbean and several 1996 MARK J. A. VERMEIJ AND STUART A. SANDIN Pacific coral species (Richmond and Hunter 1990) as well as to species from various other invertebrate taxa (Vermeij 2005). Populations of coral juveniles are frequently dominated by brooding species (Smith 1997, Carlon 2001, Edmunds 2004, Vermeij 2006). Moreover, ecosystem changes on many Caribbean reefs are leading to the systematic replacement of spawning species by brooders in the coral assemblage (Knowlton 2001, Aronson et al. 2005). As such, the study of the earliest stages of colony formation and survivorship of a brooding species will provide insights in the population dynamics typical of a growing number of degraded or recovering reef communities worldwide. Survey protocols Benthic community composition was measured in 28 fixed 50 3 50 cm quadrats at approximately three-month intervals from May 2002 until March 2005. Quadrats were haphazardly positioned at the start of the study and spanned the range of adult coral cover on the wreck (;1–30% benthic cover at a 0.25-m2 scale). Beginning on the second survey, settlement rates of S. radians were estimated by counting new arrivals to the quadrats. As such, we define settlement as the initial sighting of a settled planula (no tentacles) or early-metamorphosed polyp (with tentacles) that had not started skeletal formation (diameter ,1 mm) and had arrived at most three months previous. Benthic composition was determined in each quadrat at all intervals for three dominant functional groups: turf algae (height ,2 cm), crustose coralline algae (CCA), and adult S. radians colonies. Settlement rates were estimated from the appearance of newly arrived individuals found from repeated photographic censuses. Although larger coral colonies (.0.20 cm2) can be reliably documented with highresolution photography, smaller colonies are missed frequently. To find these smallest colonies and recent settlers, all sediment was removed from each quadrat prior to sampling by gently waving a plastic sheet above its surface so sediments were blown away by the moving water. If necessary, a turkey baster was used to further remove remaining sediments. Occasionally, algal turfs were dense and dissecting needles were used to search through them for recruits without removing the algae themselves. All settlers could be found with the naked eye and no magnification tools were used. To avoid under-sampling, small individuals (,0.20 cm2) were located and labeled in situ with temporary rings (stainless steel washers) that were color-coded to indicate the size of the encircled settler. A picture of the quadrat was taken recording all small colonies (labeled by rings) and larger colonies present. Colonies on all photographs were traced manually and analyzed using spatial analysis software (Scion Corporation, Frederick, Maryland, USA), providing information of the size and location of coral individuals of all sizes present in the quadrat. The success rate of finding all individuals inside Ecology, Vol. 89, No. 7 a quadrat, including the smallest of settlers, was .99.7% (Vermeij 2005). Note that, because settlers were sampled at threemonth intervals, we likely underestimated true settlement as individuals may have settled and died between successive survey periods. Additionally, size variation in settlers might arise as they could have been present on the bottom between one day and three months before being noted. Nevertheless, all settlers considered in this study were less than three months old when first observed and thus represent the earliest stage of coral recruitment, i.e., when a planula first attaches to the benthic substrate. For this stage, post-settlement survival (or mortality) was determined by tracking the survival (or mortality) of an individual through successive intervals. The maximum length of time that a known settler could be observed was 27 months, i.e., the time from the second survey until the end of the study. We define juvenile colonies as those composed of at least two polyps that have started skeletal formation and share a skeletal wall. This definition reflects the significant decline in instantaneous mortality probability for two- (or more) polyp individuals (minimum size of 0.20 cm2) relative to one-polyp settlers (see Results). The abundance of turf algae and CCA was estimated by overlaying each quadrat with a Plexiglas sheet labeled with 200 randomly placed circles measuring 1 mm in diameter. The benthic type under each circle was recorded, in sum providing an estimate of benthic composition (reported as percent cover of the quadrat). For more details on the sampling methodology, see Vermeij (2005). Analyses Data on settlement and post-settlement dynamics were collected from repeated measures in permanent quadrats through time. Thus, from each quadrat we have quantitative estimates of settlement rates as well as survivorship and growth histories for each individual coral settler and for colonies already present at the commencement of the study. Temporal patterns of settlement.—We tested the relationship between the abundance of each of two dominant benthic algal types (i.e., cover of CCA and cover of turf algae) and coral settlement rates through time. These analyses are founded on the observations that each algal type affects coral settlement in strong and distinct ways. CCA have been shown to promote coral settlement by releasing substances that attract coral planulae (Morse et al. 1988, Steneck and Testa 1997, Heyward and Negri 1999) and by creating suitable settlement habitat on which early post-settlement mortality is relatively low (Harrington et al. 2004, Vermeij 2006). In contrast, turf algae can negatively affect recruitment by trapping settling planulae (Birrell et al. 2005, Kuffner et al. 2006, Vermeij 2006) and, indirectly, by displacing CCA (Airoldi 2000). To investigate temporal patterns of both CCA and turf July 2008 EARLY LIFE HISTORY DYNAMICS OF CORALS algae in relation to coral settlement rates, we performed Pearson’s correlation analyses comparing means across quadrats for each algal type to mean coral recruitment rate for each of the 11 time intervals. Spatial patterns of early post-settlement survivorship.— To gain further insights into the early life history dynamics of S. radians, we investigated patterns of settlement and survivorship across quadrats. This approach was motivated by two observations. First, temporal fluctuations of settlement rate and algal cover revealed strong annual periodicity (see Results). By comparing variability across space, we avoided potentially confounding seasonal correlations in the data. Second, adult density (as measured by total coral cover of all colonies greater than 2.23 cm2) varied little through time in each quadrat (,1%). As such, investigation of the effects of adult density on settlement and survivorship could only be realized through comparisons among quadrats. The effect of adult density on the probability of settler survivorship was estimated directly from repeated censuses of each quadrat. Because of the colonial nature of corals, we define ‘‘density’’ here as the total cover of adult corals (loosely scaling with the total number of polyps), not the number of individual colonies, in the quadrat. Nearby adult density can have a negative effect on early survival through intraspecific competition (Roughgarden et al. 1985, Vermeij 2005) or through transmission of pathogens (Bruckner et al. 1997). To understand the pattern of settler survivorship, three models of early post-settlement survival probability, qA, as functions of adult coral cover, A, were generated: i) constant (qA ¼ s0), (ii) negative linear (qA ¼ s0  s1A), and (iii) inverse (qA ¼ s0/[1 þ s1A]), where s0 is the density-independent term, s1 is the density-dependent term, and their values are constrained such that 0  qA  1 for all observed A. As such, model (i) describes density-independent survivorship and models (ii) and (iii) describe linear and slower-than-linear, respectively, density-dependent survivorship. Survivorship of settlers was estimated directly by tracking the fates of known settlers found in one sampling interval until the subsequent interval. For each quadrat we recorded a maximum of 10 survivorship periods, i.e., the maximum number of inter-census intervals surveyed in this study, with which to test the relative fits of the survivorship models (with fewer than 10 events if no settlers arrived to a quadrat in the preceding time interval). Because settlers were spatially nonoverlapping, we assumed that the survivorship of each settler within a quadrat was independent of the fate of neighboring settlers, with a quadrat-specific probability of survivorship defined by models (i–iii). As such, a binomial error distribution was used to confront models of survivorship probabilities with observed settler data. The summed log likelihood of a particular model describing the observed data given values of parameters s0 and s1 was computed as folows: Lðs0 ; s1 Þ ¼ ln " 28 Y 10 Y i¼1 j¼1 1997 binomðNi; j ; Si; j jqA;x Þ # ð1Þ where binom is the binomial probability mass function, Ni, j and Si, j are the numbers of settlers arriving and surviving, respectively, for the subsequent three-month interval in quadrat i at time period j, and qA,x is the per capita probability of settlers surviving given quadratspecific coral cover A and survivorship model x described as models (i–iii) above. A simulated annealing algorithm was used to identify values of s0 and s1 (if appropriate) to maximize the likelihood described in Eq. 1 for each model of survivorship x. The statistical significance of the density-dependent term s1 from each model ii and iii was determined using likelihood ratio tests relative to the nested model (model i), i.e., explicitly testing whether the value of s1 is significantly different than 0. Note that if s1 is not different than zero, then each model ii and iii reduces to model i and, as such, model i is nested within each of the competing models. The relative fit to the data of model ii vs. iii was determined by assuming equal Bayesian prior probabilities for the models. With two alternative hypotheses represented by two models, M1 and M2, the posterior probability that the poorer fitting model M2 is correct rather than the better-fitting model M1 is 2(exp[MLM2]exp[MLM2]/[exp(MLM1) þ exp(MLM2)]), where MLMj is the maximum log likelihood for model j. Multiplying by two accounts for the fact that a priori we did not know which model was better, which is analogous to using a two-tailed statistical test. We say that model M1 is better than model M2 if this posterior probability is ,0.05 (for further details, see Hilborn and Mangel 1997). Spatial patterns of settlement rates.—Insights into the combined effects of adult density on settlement (i.e., each in producing settlers and affecting post-settlement mortality) were afforded by data on settlement rate across quadrats. The number of settlers observed at each time step is the product of the number of settlers arriving in the quadrat and the probability of each settler surviving until being counted. Because upward of 50% of settlers can be produced by nearby adults (Vermeij 2005), the total number of settlers arriving in a quadrat may approximate a positive linear function with adult density (i.e., total cover of colonies larger than 2.23 cm2). Adult density, in contrast, can have a negative effect on post-settlement survival, as explored above. We tested for density-dependent effects on settlement by comparing six models of settlement rate, xA , as functions of adult density, A. Functional forms were generated factorially from the products of two functions of settler production, mA (constant [mA ¼ p0] and positive linear [mA ¼ p0 þ p1A], where p0 and p1 are positive, estimated constants such that mA  0 for all observed A), and three functions of survival (i–iii), as above. Assuming constant planular production with increasing adult density, the functions simplify to the following: 1998 MARK J. A. VERMEIJ AND STUART A. SANDIN ðivÞ xA ¼ x0 where x0 ¼ s0 p0 ðvÞ xA ¼ x0  x1 A where x0 ¼ s0 p0 and x1 ¼ s1 p0 ðviÞ xA ¼ x0 =ð1 þ x1 AÞ where x0 ¼ s0 p0 and x 1 ¼ s1 and assuming positive linear production with increasing adult density, the functions simplify to ðviiÞ xA ¼ x0 þ x1 A where x0 ¼ s0 p0 and x1 ¼ s0 p1 ðviiiÞ xA ¼ x0 þ x1 A  x2 A2 where x0 ¼ s0 p0 ; x1 ¼ s0 p1  s1 p0 ; and x2 ¼ s1 p1 ðixÞ xA ¼ ðx0 þ x1 AÞ=ð1 þ x2 AÞ where x0 ¼ s0 p0 ; x1 ¼ s0 p1 ; and x2 ¼ s1 : The estimated constants x0, x1, and x2 represent the intercept, linear term, and nonlinear term, respectively, of the settlement rate functions relative to adult density, and are constrained such that xA . 0 for all observed A. Settlement rate was calculated for each quadrat as the mean number of settlers found per census, thereby averaging over annual fluctuations. The effects of adult density on settlement rate were estimated by comparing the relative fits of models (iv–ix) to the observed settlement rates. Based on the observations that settlement generally increased with adult density and that models describing a negative relationship between settlement and adult density fit poorly, we do not present analyses for models (v and vi) in the Results. To account for possible confounding effects of abundance of algae on realized settlement rates, we tested all combinations of additive, linear functions of quadratspecific mean cover of each turf algae and CCA on each settlement model. Site-specific cover data did not significantly violate assumptions of normality. As such, data were assumed to be normally distributed around expectations, and the most likely values of parameters x0, x1, and x2 (as needed) for each model (iv–ix) were found by maximum likelihood approaches analogous to ordinary least squares minimization. The statistical significance of the density-dependent linear and nonlinear terms (x1 and x2) and of the additive algal weightings were determined using likelihood ratio tests relative to the nested models. Non-nested models with equivalent numbers of parameters were compared statistically using the assumption of equal Bayesian prior probabilities, as in the analysis of the post-settlement survivorship data above. Size-specific mortality probability.—The dependence of mortality probability on the size of an individual Ecology, Vol. 89, No. 7 colony was estimated from mortality data. A Poisson distribution described the probability of a coral dying at a given interval, with parameter kc describing the mean time until mortality occurred for a colony in size class c. Two functional forms for the rate parameter of a Poisson mortality model were constructed: constant mortality probability with size, kc ¼ m0, and log-linear mortality probability with size, kc ¼ m0 þ m1ln(c). Estimated constants m0 and m1 describe the sizeindependent and size-dependent terms, respectively, of the mortality probability and are constrained such that kc . 0 for all observed c. The time until mortality was recorded for all coral colonies present in each quadrat at the second census (July 2002; 986 total colonies). Colonies were binned into 13 logarithmic size classes (i.e., with the geometric mean of colony area being 0.02, 0.04, 0.08, . . ., 40.96, .40.96 cm2). If the colony died during one of the subsequent 10 sampling intervals of the study, the duration was noted. Otherwise, the time until mortality occurred was recorded as greater than 29 months. The summed log likelihood of a particular model describing the observed data given values of parameters m0 and m1 was computed as follows: Lðm0 ; m1 Þ 8 > kci ;x  ki lnðkci ;x Þ þ lnðki !Þ 986 > < X 12 X ¼ >1  ½kci ;x  kj lnðkci ;x Þ þ lnðkj !Þ i¼1 > : for ki  10 for ki . 10 j¼1 ð2Þ where kci ;x is the expected mortality probability based on function x (i.e., either constant or linear with colony size, ci) and ki is the number of three-month time intervals until mortality for individual i. The equations of (2) correspond to the conditions of the colony dying during the study or the colony surviving through the duration of the entire study. As above, maximum likelihood estimates were generated for each mortality model. The significance of the size-dependent term of the log-linear model was determined by a likelihood ratio test. RESULTS Through time, the rate of coral settlement was coupled strongly to the abundance of other benthic organisms (Fig. 1). Settlement rates were correlated positively with cover of CCA (Pearson’s r ¼ 0.75, P , 0.01, n ¼ 11 time intervals) and negatively with cover of turf algae (Pearson’s r ¼ 0.63, P , 0.04, n ¼ 11). The temporal fluctuations had a distinct annual cycle with settlement high during the boreal fall, tracking annual fluctuations in the cover of CCA and turf algae. Following settlement to the benthos, adult density had a pronounced, negative effect on early postsettlement survivorship. The probability of a known new settler (zero to three months post settlement) to July 2008 EARLY LIFE HISTORY DYNAMICS OF CORALS 1999 dependent effects of adult cover on settler survival, yet both analyses fail to discriminate between the negative linear and inverse models (ii and iii, respectively) of survivorship. In addition to density-dependent effects of adult cover on settlement and early post-settlement survival, the best-fitting model describing settlement rate included significant effects of CCA cover (positive) and turf algal cover (negative; Table 1). The independent effects of each algal type in the saturating model are consistent with the results from the univariate correlations through time (Fig. 1). After successful recruitment to the benthos, the mortality probabilities of juvenile S. radians remained high, with over 75% of coral colonies up to 0.08 cm2 dying within the first year (Fig. 3). In contrast, larger FIG. 1. Seasonal fluctuations in benthic organisms. (A) Settlement (recruitment) rates of the Caribbean coral Siderastrea radians (number of newly observed settlers per 0.25 m2 per 3 months) between July 2002 and March 2005. (B) Temporal fluctuations in benthic habitat composition, notably of percent cover of crustose coralline algae (CCA; solid circles) and turf algae (open circles). Data at each time are shown as the mean of all 28 quadrats surveyed; error bars are 6 SE. survive for the subsequent three months was best fit by a negative linear function of conspecific adult cover (qA ¼ 0.802  0.006A; n ¼ 28 quadrats; P , 0.001 relative to constant model [MLi ¼ 1219.5, MLii ¼ 1187.7, where MLx is the maximum log likelihood value of model x]; Fig. 2A). Despite the superior fit of the linear model (ii), the fit could not be statistically distinguished from the inverse model (iii) (MLiii ¼ 1189.2, P ¼ 0.35). Adult density showed a general, positive effect on local settlement rates (Fig. 2B). At increased adult densities (.10% cover) the positive effect weakened to essentially no effect, as indicated by the best fit of a nonlinear saturating model relating observed settlement to adult density (Table 1; Fig. 2B). The saturating model (ix) is consistent with adult density leading to a positive, linear production of settlers, each with a probability of surviving the short period until the next census inversely related to adult density. Notably, model ix is only marginally better supported than the model (viii) with linear production and negative linear survival (P ¼ 0.053 assuming equal Bayesian priors; Table 1). As such, each the direct survivorship analyses (Fig. 2A) and the indirect survivorship analyses (i.e., the product of settlement and survivorship patterns as inferred from models (viii–ix); Table 1) confirm the negative, density- FIG. 2. Effects of adult coral cover on survivorship probability and settlement rate (number of newly observed settlers per 0.25 m2 per 3 months) of S. radians. (A) Observed (solid circles) and predicted (line) survivorship probability for early post-settlement corals for each quadrat. Survivorship is defined here as the mean proportion of newly observed individuals (beginning at 0–3 months post-settlement) that survive for the following three months. Note that each data point is the mean survivorship from the 10 surveys in one quadrat. (B) Observed settlement rate is presented (solid circles) with expected settlement rate (þ symbols) based on prediction for best-fit model (ix). Model expectation includes quadratspecific effects: adult, CCA, and turf algal cover (see Table 1). 2000 MARK J. A. VERMEIJ AND STUART A. SANDIN Ecology, Vol. 89, No. 7 TABLE 1. Best-fit parameter estimates and maximum likelihood value (ML) for competing models of mean coral settlement rate. Parameter estimate Model and components x0 x1 x2 iv) Constant, no adult effects, xA ¼ x0 No algae 21.97 CCA 11.28 Turf 35.73 CCA þ turf 33.63 vii) Linear adult effects, xA ¼ x0 þ x1A No algae 18.93 CCA 12.78 Turf 31.96 CCA þ turf 55.30 xCCA xturf ML 0.62 0.59 99.82 98.65 96.87à 96.84 0.55 0.73 98.51 91.99ààà 95.97à 90.41ààà 0.56 0.41 95.36* 89.81ààà 92.10à 89.35àà 0.46 0.03 91.75*** 88.66à 88.80à 86.28àà 0.19 0.03 0.34 0.84 0.26 1.36 viii) Quadratic adult effects, xA ¼ x0 þ x1A  x2A2 No algae 12.32 1.97 CCA 14.40 2.01 Turf 25.26 1.95 CCA þ turf 36.19 1.99 ix) Saturating adult effects, xA ¼ (x0 þ x1A)/(1 þ x2A2) No algae 24.93 46.03 CCA 63.18 44.82 Turf 9.04 60.23 CCA þ turf 20.59 6.52 0.49 0.89 0.06 0.04 0.06 0.03 1.62 3.00 1.58 0.41 0.44 0.65 0.23 0.33 Notes: Coral settlement rate was estimated in units of number of settlers per 0.25 m2 per 3 months, and adult coral, crustose coralline algae (CCA), and turf algal cover are all measured as percent cover. The estimated constants x0, x1, and x2 represent the intercept, linear term, and nonlinear term, respectively, of the settlement rate functions relative to adult density. The four competing models of adult effects, f (x0, x1, x2, A), are outlined in Materials and methods: Analsyses: Spatial patterns of settlement rates, with additive effects of two algal types as xA ¼ f (x0, x1, x2, A) þ (xCCA 3 [CCA cover]) þ (xturf 3 [turf algal cover]). The best supported model is in boldface. Asterisks indicate significance of density-dependent model parameters (x1 and x2) relative to the nested density-independent model (iv): * P , 0.05; ** P , 0.01; *** P , 0.001. Significant, additive effects of algae on model fit, repeated for each model (iv, vii–ix), are indicated as follows: à P , 0.05; àà P , 0.01; ààà P , 0.001. FIG. 3. Survival durations for all individuals present in July 2002. The maximum life span was quantified as the total number of months a recruit could have been present on the reef before its disappearance was noticed. The graph shows the fraction of each size class that survived for a certain period of time as indicated by the key. Individual polyps start first polyp divisions at a colony size around 0.2 cm2. Note the x-axis log scale. July 2008 EARLY LIFE HISTORY DYNAMICS OF CORALS colonies experienced much reduced mortality probabilities, with less than 5% of colonies larger than 2.5 cm2 dying within two years. This effect of size class on mortality probability was described by strong statistical support for a log-linear function with colony size (kc ¼ 10.98 þ 2.24ln(c), P , 0.001 compared to sizeindependent mortality model). Successful recruitment of S. radians settlers into the juvenile and adult classes depends on the combination of survivorship and colony growth. Fig. 4A combines these two processes, revealing that on average less than 20% of S. radians settlers successfully enter the two-polyp juvenile class. Although growth into a two-polyp individual can occur for upward of 27 months after settlement, the probability of initial growth (i.e., from one-polyp individuals to individuals of two or more polyps) decreased with time since settlement (Fig. 4B). For the first year after settlement, approximately 21% of surviving settlers grew into two-polyp juveniles. In contrast, less than 6% surviving as one-polyp individuals for over one year made this growth step (74 of 350 new [,3 months old] settlers vs. five of 84 one-year-old, onepolyp individuals; Fisher’s exact test, two-tailed, P , 0.004). Because both age and size strongly determined the fate of coral settlers, and because single polyps can be present on the reef for over two years, it is important to highlight both (1) the presence of one-polyp individuals per se is not indicative of recent or effective recruitment, and (2) processes structuring the early life history of S. radians populations operate predominantly on the smallest individuals, as mortality rates quickly drop after the start of polyp divisions. DISCUSSION Our results show that the early life phase of scleractinian corals is a dynamic period. Characteristics of the local habitat, especially conspecific density, greatly affect the rates of settlement and subsequent survival of settled planulae. Settlement rates of S. radians showed a saturating relationship with adult cover. Across low levels of adult cover (;0–10%) settlement increased rapidly, but settlement saturated to a fairly constant rate at higher adult cover (Fig. 2B). Importantly, the saturating relationship between total settlement and coral cover translates to a negative function of settlement per unit local coral cover (analogous to per capita measures across adult polyp densities) across all levels of cover, consistent with models of density-dependent settlement (Caley et al. 1996). Additionally, post-settlement survivorship of recently settled planulae was negatively related to adult cover (Fig. 2A). Such conspecific density dependence could act to limit the population growth rates of S. radians at high density. Density-dependent recruitment likely is effected by the combination of local production (a positive function of adult density) and very early post-settlement mortality (a negative function of adult density), occurring 2001 FIG. 4. Survival and growth of S. radians settlers. (A) The overall survival rates of all recruits observed in this study and the cumulative fraction of recruits that divided into two- (or more) polyp individuals, as a function of time after initial recruitment (i.e., age). (B) The effect of age on growth into juvenile colonies. Bars represent the fraction of surviving settlers of a particular age that had divided into two- (or more) polyp colonies. Each fraction was computed based on approximately 100 haphazardly selected individuals per age class. within days to weeks after settlement. The positive effects of density are well described by the short dispersal distances typical of S. radians planulae (Vermeij 2005). As a species adapted to physically disturbed habitats, S. radians appears to maintain the capacity for rapid population expansion at small spatial scales given arrival to an uncolonized environment. In complement, negative density dependence sets a limit to the local population growth. Vermeij (2005) proposed a mechanism leading to negative effects of adult density on early survivorship, namely Janzen-Connell effects in which the survivorship of settlers increases with increasing distance from conspecific adult colonies (Janzen 1970, Connell 1971). Microbial associates on coral surfaces are candidate pathogenic agents driving this form of density dependence (Knowlton and Rohwer 2003), such that settlers close to adult colonies harboring potential pathogens are more prone to infection than those at greater distances. Independently of mechanism, the negative density dependence inferred from settle- 2002 MARK J. A. VERMEIJ AND STUART A. SANDIN Ecology, Vol. 89, No. 7 PLATE 1. Overview of various life stages of Siderastrea radians that were considered in this study. The two metal rings in the top-left corner were used to indicate the position of recently settled planulae and non-metamorphosed settlers in our overview pictures. Photo credit: M. Vermeij. ment models and directly quantified in early postsettlement survivorship is an important dynamic affecting the regulation and ultimate density of S. radians populations (Fig. 2; Hixon et al. 2002). The patterns of both positive and negative density dependence documented here are accentuated by the high planular production rates typical of S. radians and other particularly opportunistic species. However, the mechanisms of such density dependence likely are extensible to many other coral species, especially to other brooding species with short dispersal distances. Complementing conspecific density dependence, other benthic organisms provided strong density-independent effects on patterns of settlement. We observed that increased cover of CCA increased local settlement rates of S. radians, while increased cover of turf algae reduced settlement (Fig. 1, Table 1). This result is consistent with the independent roles of each algal type; while CCA have been suggested to facilitate coral settlement by providing suitable substrate (Harrington et al. 2004, Vermeij 2005), turf algae are considered to be compet- itors for space with settling corals (Birrell et al. 2005, Vermeij 2005, 2006, Kuffner et al. 2006). The densityindependent effects of these two algal types reveal a notable pathology relating coral population growth and human disturbance. Human activities are known to increase turf algal cover while reducing available CCA cover through additions of nutrients and reductions of grazing pressure through overfishing (Hatcher and Larkum 1983, Belliveau and Paul 2002). As such, our data suggest that the efficacy of coral settlement is particularly prone to the systematic shifts in the algal community (i.e., from CCA to turf algae) commonly induced by human disturbance. Given settlement to the benthos, we have shown that one-polyp settlers have a brief temporal window during which to grow and recruit to the juvenile two-polyp stage. Mortality of settlers is high, especially while individuals are in the smallest size classes (Fig. 3). In addition to the effect of size on mortality, the probability of growing into larger size classes is affected by the age of the settler. Surviving one-polyp individuals July 2008 EARLY LIFE HISTORY DYNAMICS OF CORALS can be present on the benthos without growing into a two-polyp individual for more than two years. However, those that are less than one year old are at least three times more likely to grow into a two-polyp juvenile than their older counterparts (Fig. 4B). The combination of these mortality and maturation processes indicates the critical importance of early growth to ensure successful recruitment to the adult population. We have shown that density-dependent processes able to structure populations act on the youngest age classes of corals, i.e., before individuals become two-polyp juveniles. By the time that a coral settler reaches one year of age, most of the severe and structuring mortality and demographic bottlenecks have passed. Notably, most survey-based studies exploring early dynamics of coral populations operationally define a ‘‘recruit’’ or ‘‘settler’’ as an individual ,3–5 cm2. In comparison, the typical S. radians individual surviving one year is ,0.1 cm2 in size. Depending on species-specific growth rates, tracking the fate of .1 cm2 colonies likely omits reference to the most critical of early life history dynamics. Because structuring mortality, both densityindependent and density-dependent, occurs well before these larger colonies are counted, we may expect that studies based on such definitions will identify tight coupling between density of operational ‘‘recruits’’ and future contribution to the adult population (see similar concern for coral reef fish in Doherty and Fowler [1994] and subsequent comments in Caley et al. [1996] and others). Although this predictive ability is powerful for some applications (e.g., monitoring programs and basic inventories), focused investigations of the factors dynamically affecting coral settlement and recruitment will not be successful without specific reference to these smallest size and age classes and the incorporation of time. ACKNOWLEDGMENTS This work was carried out under permit no. FKNMS 2002032. C. Fasano, as well as many volunteers from RSMAS and NOAA are thanked for their assistance during the fieldwork. Funding was provided by NOAA Fisheries’ Coral Reef Initiative and E. W. Scripps, Jr., Rolf P. M. Bak, Scott Lilliputian Hamilton, and three anonymous reviewers provided invaluable comments on earlier drafts. LITERATURE CITED Airoldi, L. 2000. Effects of disturbance, life histories, and overgrowth on coexistence of algal crusts and turfs. Ecology 81:798–814. Anderson, T. W. 2001. Predator responses, prey refuges, and density-dependent mortality of a marine fish. Ecology 82: 245–257. Aronson, R. B., I. G. Macintyre, S. A. Lewis, and N. L. Hilbun. 2005. Emergent zonation and geographic convergence of coral reefs. Ecology 86:2586–2600. Belliveau, S. A., and V. J. Paul. 2002. Effects of herbivory and nutrients on the early colonization of crustose coralline and fleshy algae. Marine Ecology Progress Series 232:105–114. Birrell, C. L., L. J. McCook, and B. L. Willis. 2005. Effects of algal turfs and sediment on coral settlement. Marine Pollution Bulletin 51:408–414. 2003 Bruckner, A. W., R. J. Bruckner, and E. H. Williams. 1997. Spread of a black-band disease epizootic through the coral reef system in St Ann’s Bay, Jamaica. Bulletin of Marine Science 61:919–928. Caley, M. J., M. H. Carr, M. A. Hixon, T. P. Hughes, G. P. Jones, and B. A. Menge. 1996. Recruitment and the local dynamics of open marine populations. Annual Review of Ecology and Systematics 27:477–500. Carlon, D. B. 2001. Depth-related patterns of coral recruitment and cryptic suspension-feeding invertebrates on Guana Island, British Virgin Islands. Bulletin of Marine Science 68:525–541. Chabot, R., and E. Bourget. 1988. Influence of substratum heterogeneity and settled barnacle density on the settlement of cypris larvae. Marine Biology 97:45–56. Chiappone, M., and K. M. Sullivan. 1994. Patterns of coral abundance defining nearshore hardbottom communities of the Florida Keys. Florida Scientist 57:108–125. Connell, H. H. 1971. On the role of natural enemies in preventing competitive exclusion in some marine animals and in rain forest trees. Pages 298–312 in P. J. Den Boer and G. Gradwell, editors. Dynamics of populations. Center for Agricultural Publication and Documentation, Wageningen, The Netherlands. Connell, J. H., T. P. Hughes, and C. C. Wallace. 1997. A 30year study of coral abundance, recruitment and disturbance at several scales of space and time. Ecological Monographs 67:461–488. Creed, J. C., T. A. Norton, and J. M. Kain. 1996. Are neighbours harmful or helpful in Fucus vesiculosus populations? Marine Ecology Progress Series 133:191–201. Doherty, P. J., and A. J. Fowler. 1994. An empirical test of recruitment limitation in a coral reef fish. Science 263:935– 939. Duerden, J. E. 1904. The coral Siderastrea radians and its postlarval development. The Carnegie Institution, Washington, D.C., USA. Dunstan, P. K., and C. R. Johnson. 1998. Spatio-temporal variation in coral recruitment at different scales on Heron reef, southern Great Barrier Reef. Coral Reefs 17:17–81. Edmunds, P. J. 2004. Juvenile coral population dynamics track rising seawater temperature on a Caribbean reef. Marine Ecology Progress Series 269:111–119. Gilmour, J. 1999. Experimental investigation into the effects of suspended sediment on fertilization, larval survival and settlement in a scleractinian coral. Marine Biology 135:451– 462. Gosselin, L. A., and P. Y. Qian. 1997. Juvenile mortality in benthic marine invertebrates. Marine Ecology Progress Series 146:265–282. Harrington, L. M., K. Fabricius, G. Dea’th, and A. Negri. 2004. Habitat selection of settlement substrata determines post-settlement survival in corals. Ecology 85:3428–3437. Harriott, V. J. 1999. Coral recruitment at a high latitude Pacific site: a comparison with Atlantic reefs. Bulletin of Marine Science 65:881–891. Hatcher, B. G., and A. W. D. Larkum. 1983. An experimental analysis of factors controlling the standing crop of the epilithic algal community on a coral reef. Journal Experimental Marine Biology and Ecology 69:61–84. Heyward, A. J., and A. P. Negri. 1999. Natural inducers for coral larval metamorphosis. Coral Reefs 18:273–279. Hilborn, R., and M. Mangel. 1997. The ecological detective: confronting models with data. Princeton University Press, Princeton, New Jersey, USA. Hixon, M. A., and M. H. Carr. 1997. Synergistic predation, density dependence, and population regulation in marine fish. Science 277:946–949. Hixon, M. A., S. W. Pacala, and S. A. Sandin. 2002. Population regulation: historical context and contemporary challenges of open vs. closed systems. Ecology 83:1490–1508. 2004 MARK J. A. VERMEIJ AND STUART A. SANDIN Hixon, M. A., and M. S. Webster. 2002. Density dependence in reef fishes: coral-reef populations as model systems. Pages 303–325 in P. F. Sale, editor. Coral reef fishes: dynamics and diversity in a complex ecosystem. Academic Press, San Diego, California, USA. Hughes, T. P., A. H. Baird, E. A. Dinsdale, N. A. Moltschaniwskyj, M. S. Pratchett, J. E. Tanner, and B. L. Willis. 2000. Supply-side ecology works both ways: the link between benthic adults, fecundity and larval recruits. Ecology 81:2241–2249. Hughes, T. P., A. H. Baird, E. A. Dinsdale, V. J. Harriott, N. A. Moltschaniwskyj, M. S. Pratchett, J. E. Tanner, and B. L. Willis. 2002. Detecting regional variation using metaanalysis and large-scale sampling: latitudinal patterns in recruitment. Ecology 83:436–451. Hughes, T. P., and J. E. Tanner. 2000. Recruitment failure, life histories and long-term decline of Caribbean corals. Ecology 81:2250–2263. Hunt, H. L., and R. E. Scheibling. 1997. Role of early postsettlement mortality in recruitment of benthic marine invertebrates. Marine Ecology Progress Series 155:269–301. Janzen, D. H. 1970. Herbivores and the number of tree species in tropical forests. American Naturalist 104:501–528. Johst, K., R. Brandl, and S. Eber. 2002. Metapopulation persistence in dynamic landscapes: the role of dispersal distance. Oikos 98:263–270. Knowlton, N. 2001. The future of coral reefs. Proceedings of the National Academy of Sciences (USA) 98:5419–5425. Knowlton, N., and F. Rohwer. 2003. Microbial mutualisms on coral reefs: the host as a habitat. American Naturalist 162: 51–62. Kuffner, I. B., L. J. Walters, M. A. Becerro, V. J. Paul, R. Ritson-Williams, and K. S. Beach. 2006. Inhibition of coral recruitment by macroalgae and cyanobacteria. Marine Ecology Progress Series 323:107–117. Lafferty, K. D., J. W. Porter, and S. E. Ford. 2004. Are diseases increasing in the ocean? Annual Review of Ecology, Evolution and Systematics 35:31–54. Lewis, J. B. 1989. Spherical growth in the Caribbean coral Siderastrea radians (Pallas) and its survival in disturbed habitats. Coral Reefs 7:161–167. Lirman, D., D. Manzello, and S. Macia. 2002. Back from the dead: the resilience of Siderastrea radians to severe stress. Coral Reefs 21:291–292. Miller, M. W., E. Weil, and A. Szmant. 2000. Coral recruitment and juvenile mortality as structuring factors for reef benthic communities in Biscayne National Park. Coral Reefs 19:115– 123. Minchinton, T. E. 1997. Life on the edge: conspecific attraction and recruitment of populations to disturbed habitats. Oecologia 111:45–52. Miron, G., B. Boudreau, and E. Bourget. 1999. Intertidal barnacle distribution: a case study using multiple working hypotheses. Marine Ecology Progress Series 189:205–219. Molofsky, J. 1994. Population dynamics and pattern formation in theoretical populations. Ecology 75:30–39. Morse, D. E., N. Hooker, A. N. C. Morse, and R. A. Jensen. 1988. Control of larval metamorphosis and recruitment in sympatric agariciid corals. Journal of Experimental Marine Biology and Ecology 116:193–217. Ecology, Vol. 89, No. 7 Moses, C. S., P. K. Swart, K. P. Helmle, R. E. Dodge, and S. E. Merino. 2003. Pavements of Siderastrea radians on Cape Verde reefs. Coral Reefs 22:506–506. Pepin, P. 1991. Effect of temperature and size on development, mortality, and survival rates of the pelagic early life history stages of marine fish. Canadian Journal of Fisheries and Aquatic Sciences 48:503–518. Reed, D. C. 1990. An experimental evaluation of density dependence in a subtidal algal population. Ecology 71:2286–2296. Richmond, R. H., and C. L. Hunter. 1990. Reproduction and recruitment of corals: comparisons among the Caribbean, the tropical Pacific, and the Red Sea. Marine Ecology Progress Series 60:185–203. Roughgarden, J., Y. Iwasa, and C. Baxter. 1985. Demographic theory for an open marine population with space-limited recruitment. Ecology 66:54–57. Sammarco, P. W. 1991. Geographically specific recruitment and post-settlement mortality as influences on coral communities: the cross-shelf transplant experiment. Limnology and Oceanography 36:469–514. Smith, S. R. 1992. Patterns of coral recruitment and postsettlement mortality on Bermuda reef: comparisons to Caribbean and Pacific reefs. American Zoologist 32:663–764. Smith, S. R. 1997. Patterns of coral settlement, recruitment and juvenile mortality with depth at Conch Reef, Florida. Proceedings of the Eighth International Coral Reef Symposium 2:1197–1202. Soong, K. 1991. Sexual reproductive patterns of shallow-water reef corals in Panama. Bulletin of Marine Science 49:832–846. Steneck, R. S., and V. Testa. 1997. Are calcareous algae important to reefs today or in the past? Symposium summary. Proceedings of the Eighth International Coral Reef Symposium 1:685–688. Sweatman, H. 1988. Field evidence that settling coral reef fish larvae detect resident fishes using dissolved chemical cues. Journal of Experimental Marine Biology and Ecology 124: 163–174. Tupper, M., and R. Boutilier. 1995. Effects of conspecific density on settlement, growth and post-settlement mortality of a temperate reef fish. Journal of Experimental Marine Biology and Ecology 191:209–222. Vermeij, M. J. A. 2005. Substrate composition and adult distribution determine recruitment patterns in a Caribbean brooding coral. Marine Ecology Progress Series 295:123–133. Vermeij, M. J. A. 2006. Early life-history dynamics of Caribbean coral species on artificial substratum: the importance of competition, growth and variation in life-history strategy. Coral Reefs 25:59–71. Vermeij, M. J. A., N. D. Fogarty, and M. W. Miller. 2006. Pelagic conditions affect larval behavior, survival, and settlement patterns in the Caribbean coral Montastraea faveolata. Marine Ecology Progress Series 310:119–128. Vermeij, M. J. A., P. R. Frade, R. Jacinto, A. O. Debrot, and R. P. M. Bak. 2007. Effect of reproductive mode on habitatrelated differences in population structure of eight Caribbean coral species. Marine Ecology Progress Series 351:91–102. Wallace, C. C. 1985. Seasonal peaks and annual fluctuations in recruitment of juvenile scleractinian corals. Marine Ecology Progress Series 21:289–298.