The Relationship between Physical and Biological Habitat Conditions and Hermatypic Coral Recruits Abundance within Insular Reefs (colombian Caribbean)

Little evidence exists on the dependence between the presence and abundance of juvenile hermatypic corals and the conditions of their habitats, despite that juveniles contribute with the understanding of the community structure and its reproductive success. To assess this, the abundance of nine species of juvenile corals was correlated with eight macro-habitat (location of the reef on shelf, depth) and micro-habitat (type and inclination of the substrate, exposure to light, texture and amount of sediment accumulated on bottom, potential growth area for juveniles) conditions. Sampling was conducted in four insular coral reefs in the Colombian Caribbean: two oceanic and two continental reefs (influenced by large rivers), covering a total of 600m 2 and the distribution of corals on a vertical gradient. Contingency tables and coefficients (magnitude) and multiple correspondence analyses were used to evaluate the dependency ratios for each species. The results showed that Agaricia tenui-folia displayed the most robust pattern of dependence (two high and two moderate), significant for juveniles present at a high frequency in continental reefs, devoid of potential area for juvenile growth (surrounded by macroalgae), and covering horizontal substrates exposed to light. The juveniles were associated with a habitat of moderate to high bottom accumulation of extremely fine sediment. Porites astreoides presented four moderate dependencies; ocean reefs between 2-16m depths, a high frequency of juveniles on horizontal substrates, exposed to light, non-sedimented and occupied by competitors. Siderastrea siderea displayed three moderate dependences for juveniles in cryptic zones, inclined substrate and devoid of competitors. A. lamarcki, Leptoseris cucullata and A. agaricites presented two moderate dependences; these species share high abundance of juveniles in habitats with no sediment, exposed to light and occupied by competitors (except A. agaricites). The P. porites, Favia fragum and Montastraea cavernosa species had a moderate dependence with high incidence of juveniles in ocean reefs and microhabitats exposed to light. For the nine species, results indicate that the presence (colonization), abundance and survival of juveniles, depend on certain species-specific particularities of the habitat. However, the juveniles show high tolerance and plasticity to a range of habitat variables, given their independence and low dependence observed in over 50% of the variables assessed. Hermatypic coral recruitment, defined as the introduction of new individuals to a population , occurs when a larva settles, undergoes metamorphosis (formation of the calcium car-bonate skeleton) and endures a length of time (Sale et al. 2010). The ecological and evolutionary significance is …

Hermatypic coral recruitment, defined as the introduction of new individuals to a population, occurs when a larva settles, undergoes metamorphosis (formation of the calcium carbonate skeleton) and endures a length of time (Sale et al. 2010).The ecological and evolutionary significance is that recruitment is a key process within the reef's successional cycle, since it determines the structure of the coral community, the renewal and continuity of local populations, favoring an increase of genetic variability, which in turn implies an adaptive advantage to climate change (Porter & Tougas 2001).
Recruitment measurable in situ by the presence of juveniles between 4mm and 2-4cm (depending on species) is considered the result of the synergy of several processes: partial and total mortality, that results in the quantity or coverage of reproductive adults, the total number of gametes and larvae released through sexual reproduction (reproductive output), the dispersal and survival of larvae in the water column, larval macro and micro habitat selection behavior for attachment before loss of competitiveness, larval attachment according to the extent, quality and availability of the most advantageous substrate (Baird et al. 2003, Sale et al. 2010).The process of metamorphosis to form polyps, asexual reproduction for juvenile growth, resistance to the physical environment conditions (Baird et al. 2003), and competition for resources such as substrate and light (McCook et al. 2001, Fabricius 2005) also partake in this synergy.When these processes are favorable it is reflected in the presence, abundance and distribution of juveniles in a particular habitat; this success is mediated by species life history, and the pressure of local natural selection caused by abiotic and biotic factors (Hughes & Jackson 1985).
There are several possible explanations for the spatial distribution of juveniles of a particular coral species (Mundy & Babcock 1998, 2000, Dornelas et al. 2006): 1-settlement and thus recruitment is an entirely random spatial process (not larval selection), in which the physical and biological habitat do not determine the pattern of recruitment (independent); 2-larvae of each species settle on preferential reef habitats according to their resource requirements, and habitat factors do not affect the distribution of recruits; 3-larvae select suitable settlement habitat, and habitat factors cause post-settlement mortality to produce the recruitment pattern; 4-distribution and abundance are independent of larval behavior but dependent on selective factors in the environment.Vermeij et al. (2006) have reported that in the case of environments providing conditions of high stress (drastic changes in salinity, sedimentation, light and competition) larvae do not present specific habitat selection; therefore, settlement becomes opportunistic and random.Contrastingly, coral larvae when under favorable macro-habitat conditions (no pollution), and having different micro-habitat options, is more active in its selection, creating a pattern of settlement and recruitment that varies by species (Harrington et al. 2004).There is no consensus on which of these hypotheses correctly explain the distribution and abundance of juvenile corals in a reef; it is a complex process, where multiple habitat variables interact, at different temporal and spatial scales, with larval settlement behavior.What is evident is that the fitness of any adult coral depends on the conditions of the habitat where the juvenile lived and developed.
Globally, we know that there are key factors of macro and micro-habitat that affect the survival of juvenile coral.At a macro-scale, for example, are the type and geographic location of the reef, depth and the degree of environmental degradation (Richmond 1997).At micro-scale are the availability and complexity of the substrate (Ruiz-Zárate et al. 2000), the type and amount of accumulated sediment and the competition for resources with other sessile organisms (McCook et al. 2001).
At micro-scale, the characteristics of the settlement substrate, is one of the factors that determine habitat preference and the pattern of juvenile survival (Ruiz-Zárate et al. 2000, Sale et al. 2010).The differential location of larvae on the substrate, is related to site selection, the irregularity and substrate availability, resistance to sedimentation, depth and variation of light intensity (Ruiz-Zárate et al. 2000, Baird et al. 2003).There is evidence that suggests that most larvae prefer roughened substrates, such as rocks and dead coral, as they offer countless micro-spaces with binding potential (Smith 1997); however, larvae species such as the Manicina aerolata prefer attachment onto encrusting red algae because they facilitate settlement, metamorphosis and survival in conditions of high sedimentation (Ruiz-Zárate et al. 2000).
Factors such as vulnerability to foraging (sea urchins, fish) and the ability to compete with macroalgae and other sessile organisms for space are involved in juvenile survival, after settlement takes place (Ruiz-Zárate et al. 2000, McCook et al. 2001).It has been shown that some algae not only compete for space and light with adult corals, but also with juveniles (overgrowth and suffocating), especially under conditions of nutrient enrichment, high sedimentation and low herbivores density (Birrell et al. 2008, Elmhirst et al. 2009).
The type and amount of sediment accumulated in the microhabitat is considered as the most significant factor determining larval and juvenile survival worldwide (Edmunds et al. 2004, Fabricius 2005).The proximity to river mouths causes an increase in the amount of particulate organic and inorganic matter entering the reef, causing nutrification, turbidity and sedimentation.This favors the proliferation of macroalgae (Fabricius 2005), the reduction of available surfaces for coral settlement (Edmunds et al. 2004, Hughes et al. 2007); a decrease of the larvae's sensory ability to select microhabitat, and a decline in the rate of juvenile survival and diversity (Gardner et al. 2003).The negative effects of sediment runoff entering the reef have been reported by Fabricius (2005) in even the early reproductive stages (gametogenesis, fertilization and embryogenesis) as well as the late stages (larval survival, settlement, metamorphosis, juvenile growth and survival).
An increase in reef regime alterations since the early seventies has caused a modification in the settlement habitats and the survival of juveniles in the Caribbean (Gardner et al. 2003).Juvenile coral communities are enduring a "community shift" (Aronso et al. 2004, Hughes et al. 2007), caused by the dominance of brooding species such as the Agaricia spp.and Porites spp., which replaced spawning species, such as Montastraea annularis and Siderastrea siderea (Porter & Tougas 2001, Green et al. 2008).
Although it is clear that the habitat factors mentioned above are important for both adult and juvenile corals, the differential effects of each factor in the post-settlement survival by species that determine the frequency, density, juvenile survival and fitness of the species have not been thoroughly examined (Mundy & Babcoock 2000, Victor 2008, Birrell et al. 2008).Consequently, it is unknown whether the recruitment and survival of coral species in the Caribbean, in fact, depends on certain physical and biological habitat factors or if it is entirely random (Mundy & Babcoock 2000).Investigating the consequences (juveniles of the same species present in the same macro or micro-habitat=frequency) could help to infer the causal agents that are affecting the larval settlement (selective or not) and/or the selective factors of the environment (present or not).
The contribution of this study is to demonstrate whether the frequency of juvenile hermatypic corals depends on certain physical factors of the habitat on a macro-scale (location of the reef on the shelf and depth) and on a micro-scale (type and angle of the substrate, exposure to light, texture and amount of sediment accumulated around the juveniles) and micro-biological factors of the micro-habitat (competition for potential juvenile development space).The results will be a key tool to define habitat variables that favor or limit the survival of nine dominant coral species in the Colombian Caribbean.This information is vital for the conservation, rehabilitation (translocation and transplantation of juveniles, larval supply) and protection of reef habitats.

Study area:
The study was conducted in four insular reefs in the Colombian Caribbean (Fig. 1), two on the continental shelf (Isla Fuerte and Isla Grande) subject to different levels of disturbance caused by freshwater input from large rivers, and two on the ocean platform (San Andres and Providencia).(Garay 2001) and is the main affluent of the archipelago.The profile of the windward terraces (10º10'21'' N -75º42'36'' W) is characterized by a reef crest less than 10m wide and a slope with coral cover up to 25m.The reefs feature a predominance of M. annularis, Agaricia tenuifolia and Porites astreoides, along with deteriorated coral patches of Acropora palmata and extensive coverage of macroalgae such as Dictyota spp.and Halimeda spp.(Alvarado et al. 1989).

Continental shelf reefs:
Isla Fuerte is part of the Bajo Bushnell and Bajo Burbujas reef complex (Díaz et al. 1996); it is influenced by the following rivers: the Sinú (average flow=290.8m 3/s, Garay 2001), the Atrato (average flow=2 366.13m 3 /s, Garay 2001) and the Magdalena (7 149.53m 3 /s, Garay 2001).The windward zone (9º23'15.6''N -76º10'10.1''W) has a terrace formed by a fringing reef up to a depth of 20m, followed by a sandy bottom, with a coral zonation typical of the Caribbean (Díaz et al. 2000), with high algal growth (Dictyota spp.) and the accumulation of extremely fine sediment (siltclay).The annual loss rate by species or genus in continental reefs, caused by sedimentation and pollution has been estimated at 0.4-2.2%(Acosta & Martínez 2006).
Ocean platform reefs: Sampling was conducted west of San Andres and Providencia, at coordinates 12º30'01'' N -81º45'56'' W and 13º23'54.6'' N -81º23'57'' W, respectively.The best coral formations in San Andres are found on the leeward terraces (fringing reef).The shallow terrace features a calcareous pavement with areas exposed or covered with algae, corals, octocorals and scattered sponges.The submerged terrace is a sandy plain where at a depth of 12m we find a coral carpet of Dendrogyra cylindrus, Diploria labyrinthiformis, D. strigosa, Colpophyllia natans and Montastraea (Díaz et al. 1995).The latter terrace ends abruptly at 20-22m with an inclined coralcovered slope and sediment to 30-35m (Díaz et al. 1995, Acosta & Martínez 2006).Leeward in Providence has a broad lagoon terrace covered with sediment and sea grass areas.The shallow zone is strewn with skeletons of Acropora palmata, Millepora complanata, zoanthids, and fleshy and coralline algae.The fore-reef terrace (12-15m) has an assortment of corals, predominantly of the Diploria spp., M. annularis, Dichocoenia stokesi, P. astreoides species, as well as octocorals, while in deeper waters, the coverage of Montastraea cavernosa, M. annularis, D. labyrinthiformis and C. natans increases, as well as the presence of sponges and octocorals (Díaz et al. 2000).Coral coverage on the slope decreases rapidly (Acosta & Martínez 2006).Sampling: A line was traced from the deepest point (maximum 30m), where the reef ends or the sandy bottom begins, to the coast, in areas of the most thriving coral reefs (richness, coral coverage).Perpendicular to this line and at increasing depths of 2m, 12 quadrants of 1m 2 were measured, six on each side of the line and 1m from each other (following Vidal et al. 2005); this procedure was repeated along the entire length of the reef depth gradient.The transects were grouped by depth in four categories (Acosta et al. 2011): 1-Shallow (2, 4, 6 and 8m), 2-Medium (10, 12, 14 and 16m), 20,22 and 24m) and 4-Very deep (>26m), the latter only found in oceanic reefs, totaling 600m 2 sampled.Juveniles were defined as colonies of less than 4cm in diameter for large species (diameter>15cm) such as Montastraea spp., Diploria spp.and S. siderea (Bak & Engel 1979, Richmond & Hunter 1990, Dueñas et al. 2010); and 2cm in diameter for small species (diameter<15cm), such as P. astreoides and Favia fragum.Taxonomic identification was done according to the Dueñas et al. (2010) guide.
As the number of species and juveniles per quadrant was recorded, the physical and biological characteristics of the habitat in which each one of the juveniles was found were described as follows: 1-Location of the reef on the platform; 2-Range of depth; 3-Type of substrate colonized by the juvenile; 4-Substrate inclination on which the juvenile was found; 5-Juvenile's exposure to light; 6-Texture of the sediment accumulated on the substrate around the juvenile; 7-Amount of accumulated sediment on the bottom, around the juvenile; 8-Potential space around the juvenile for growth.See table 1 for further details on classification and measurement criteria for each variable.
The 3 000 juveniles and 45 species presenting recruitment were used to conduct an exploratory analysis of the absolute frequencies of all juveniles in the four coral reefs; the aim was to identify the most common species that had a sampling error less than or equal to 10%, and 95% confidence, using the formula of Pita (1996): E=(Zα 2 )(pq)/n, where Zα=1.96 at 95% confidence, p=percentage of the community expected to be sampled (75%), q=1-p, n=sample size (total number of individuals per species).These species were Agaricia agaricites, Leptoseris cucullata, P. astreoides, S. siderea, F. fragum, Scolymia spp., A. lamarcki, M. cavernosa, A. fragilis, P. porites, S. radians and A. tenuifolia.These 12 species were used to analyze the pattern of dependencies between juveniles and the characteristics of their habitat, by employing the chi-square (X 2 ) test for contingency tables (CT).The Yates continuity correction was used on 2x2 CT.When at maximum 20% of the table cells had an expected frequency under five, we applied Fisher's exact test.However, the proper application of the X 2 test to Scolymia spp., A. fragilis and S. radians was impossible, as more than 20% of the cells in the contingency tables continuously resulted in frequencies equal to zero; consequently, they were excluded from the analysis.For the nine remaining species, in cases where H 0 was rejected (independence of variables), we calculated the contingency coefficient (C) to determine the extent of dependence.This coefficient ranged between zero and a maximum value of association (C maximum ) which corresponds to the number of categories of the variables that were evaluated.Values closer to zero were considered of low association between the variables and those that came closest to its C maximum as a high association (Table 2).Not all combinations of variables were tested; when crossing certain categories of independent variables in the CT some had frequencies of zero and expected frequencies under five, this precluded the application of the test.Subsequently, a multiple correspondence analysis (MCA) was used to reduce variable information to two dimensions on a Cartesian plane (Díaz 2002), graphically representing the association of the variables with the species.

RESULTS
The X 2 test (p<0.05)and the MCA (Fig. 2) evidenced the most robust pattern of dependency in juveniles of A. tenuifolia.Only this  species showed a significantly high dependence (CT 2x2, C=0.59; Fig. 3A), between the frequency of juveniles and habitats located in continental reefs and juveniles surrounded by competitors (78.7% continental reefs and 82.5% with macroalgae and corals).High dependence was also found among juveniles on substrates exposed to light and those on horizontal substrates (CT 3x2, C=0.55; Fig. 3A).Additionally, moderate dependence was found, between juveniles in continental reefs, and juveniles found on substrates exposed to light, and in turn, of those located on substrates exposed to light with those sharing space with other competitors (CT 2x2, C=0.25-0.3;Fig. 3A).Graphically, the MCA illustrates how A. tenuifolia settled apart from other species, beside the habitat with moderate to high silt-clay sediment in continental reefs (10-16m).It is the only species to have 75.2% of its juveniles under these conditions (Table 3).
The frequency of juvenile P. astreoides, moderately depended (CT 2x2, 3x2 y 4x2, C=0.3-0.5;Fig. 3B), on oceanic reefs (61.7%) with substrates exposed to light (72.6%), the latter with horizontal substrates (48.4 %) and non accumulated sediment (56%); in the same way, habitats with non accumulated sediment, but with competitors around the juveniles.The dependence on substrates exposed to light and substrates without accumulated sediment presented the highest value for the species (C=0.5).
It was also established that the frequency of juvenile S. siderea on cryptic substratum (60.2%) depended on its location at a depth of 10-16m (31.7%), on an inclined substrate (60.7%), as well as juveniles on an inclined substrate devoid of competitors (71%); all of these moderate dependences (CT 3x2 and 4x2, C=0.28-0.33;Fig. 3C).Graphically (by the MCA) it was established that the higher frequency of juveniles of the species was on rubble-type substrates (78.5%), cryptic (60.2%), of coarse and fine sediment (47.3% and 39.8% respectively), accumulated in amounts greater than 1cm (69.9%).See Appendix 1 for further details on species frequencies for all measured variables.
The highest frequency of A. lamarcki juveniles depended on substrates occupied by competitors (63.5%), with no sediment accumulated (91.5%) on substrate exposed to light (62.8%).The magnitude of the dependence between variables ranged from low to moderate (CT 2x2 and 4x2; C=0.24-0.34;Fig. 4A).Graphically it was confirmed that the factor closest to A. lamarcki and which determined the frequency of its juveniles, was the absence of sediment on the substrate since, through the MCA, the species was located apart from the other species.However, statistical independence (X 2 , p>0.05) was proved between light exposure and inclination of the substrate, and between the inclination of the substrate and space for growth around the juveniles (Fig. 4A).Abundance was lowest in continental reefs (0.8%) and at depths under 10m (2.3%).
Like P. astreoides, juveniles of L. cucullata and A. agaricites showed that despite being in a variety of habitat conditions, as shown in the figures representing dependencies between variables (Fig. 4B and 4C) and the MCA's two-dimensional plane, the strongest dependence (moderate) was presented by juveniles in substrate devoid of accumulated sediment and substrate exposed to light; with 71.8% and 59.6% of L. cucullata juveniles under these conditions (CT 4x2, C=0.43; Fig. 4B).Additionally, moderate dependence was proven (CT 4x2, C=0.33; Fig. 4B) for L. cucullata between the frequency of juveniles on non-sedimented substrates (71.8%) and substrates occupied by competitors (55.4%).   A. agaricites also showed moderate dependence (CT 4x2, C=0.42; Fig. 4C) between non-sedimented substrates (49.4%) and light availability (51.3%), and those without sediment and no competitors (55%).
The juveniles of P. porites showed dependence on oceanic environments (78.4%) and microhabitats without interaction or the presence of other sessile organisms (54.5%), of moderate dependence (CT 2x2; C=0.42; Fig. 4D).Independence between its location in the reef and light exposure (X 2 , p>0.05) was proven for this species.
As for F. fragum, we found that the abundance of juveniles depended on ocean reefs (89.6%) and substrates exposed to light (92.5%) of moderate intensity (CT 2x2; C=0.32; Fig. 4E).The species was abundant at depths under 16m and on rubble, but lower in habitats with coarse sediment (Appendix 1).However, statistical independence was confirmed between the location of the reef and the potential space available for juvenile growth, and the latter with light exposure (X 2 , p>0.05).
The juveniles of M. cavernosa showed moderate dependence (CT 4x2; C=0.35; Fig. 4F) on substrates exposed to light (56.8%), at depths of 10-16m (47.2%).In addition, the independence between light exposure and the potential space to grow was established, and the latter with substrate inclination (X 2 , p>0.05).
While juveniles of species such as A. tenuifolia and S. siderea, were most abundant in environments with accumulated sediment, the strongest relationships (higher frequency of juveniles) for A. lamarcki, P. astreoides, L. cucullata and A. agaricites juveniles, were present in micro-habitats exposed to light and without sediment accumulation around juveniles; this evidenced that the frequency of juveniles of some species increased in environments of low accumulated sediment.
L. cucullata, A. agaricites, P. porites, F. fragum and M. cavernosa were grouped within the MCA plane, amid many categories of the different physical and biological variables of the habitat and the dependences tested were mostly of low magnitude (none of these species showed high dependence).Homogeneous frequencies between the different categories of each factor confirm that juveniles of these species can be effectively found under diverse habitat conditions (tolerant).

DISCUSSION
These results suggest that the frequency, abundance and survival of juvenile hermatypic corals in insular reefs of the Colombian Caribbean are affected by physical and biological features of their environment and that they depend on certain peculiarities of the habitat, which are species-specific.According to the characteristics of relationships and the number of dependences of moderate to high magnitude, the following two groups of species were recognized: 1-species with the highest number of dependences (3-4) with regard to the macro and micro-habitat variables (A.tenuifolia, P. astreoides and S. siderea) and species that displayed the least number of dependences (1-2) and recruit in a wide range of habitat conditions (A. lamarcki, L. cucullata, A. agaricites, P. porites, F. fragum and M. cavernosa).
The mechanism that describes the dependence of juveniles on certain habitat categories and variables could be simply explained as the balance between larval input where the larvae perform or not a selection within the spectrum of habitat conditions and provide a new individual to the population, and the output to the system generated by the differential mortality of juveniles (natural selection), where ultimately, one of the habitat categories evaluated is favored (disproportionate number of juveniles).Following are two scenarios where this would happen: 1.The combination of high habitat selection by the larva, that is, it requires a specific resource or habitat condition (population's habitat preference), and low post-settlement differential mortality in that specific habitat where juveniles frequently occur (Mundy & Babcock 1998, Ruiz-Zárate et al. 2000, Harrington et al. 2004), 2. No larval selection, random or similar colonization in all habitat categories evaluated (Mundy & Babcock 2000, Vermeij et al. 2006) and strong selection pressure in all but one of the habitat categories, which would explain the dependence identified.In turn, larval substrate selection depends on habitat characteristics (quality, quantity and resource availability) and conditions of the environment, which have not been quantified for any species, in the reefs assessed.Meanwhile, the number of larvae reaching a system depends on population size, number of reproductive individuals, reproductive effort (Alvarado & Acosta 2009); strategy and mode of sexual reproduction of each species (Miller & Barimo 2001), and type of larva and dispersion rate vs. self-seeding (Sale et al. 2010).Also, the natural selection that occurs in larval stages, affects the settlement and metamorphosis (Mundy & Babcock 2000, Baird et al. 2003) and later post-settlement (Vermeij et al. 2006, Victor 2008).The complexity of the variables involved and the lack of understanding regarding the early life stages of the species make discussion on the findings difficult.
The independence between habitat variables and the abundance of juveniles could be explained by the combination of the following: 1. Absence of larval selection or opportunistic behavior at the time of settlement (Vermeij et al. 2006), concurrently with low natural selection of juveniles within the categories of the variable; 2. High preference for one habitat category (Mundy & Babcock 1998, 2000) and natural selection directed on that category presenting most frequency of juveniles, which would generate an equilibrium in frequency among the categories of the variable compared, and therefore, independence.The results reflect the adaptation and high tolerance of juvenile coral to numerous habitat factors, which allows for their survival (Baird et al. 2003) in a system with high spatial heterogeneity and where resources and conditions are distributed in patches or follow a gradient.Some of the traits most relevant to the higher survival rate of coral species are: high phenotypic plasticity, which confers them tolerance to various conditions of the environment (Miller & Barimo 2001, Green et al. 2008) and resistance to changes in habitat conditions (Lirman & Manzello 2009), efficiency in the removal of accumulated sediment (mostly by massive or foliar growth, Huitric & McField 2001), a high energy investment to protect and reduce possible inter-specific competition for potential growth space (Green et al. 2008); rapid colonization of empty substrates as a result of a high investment in sexual reproduction and periodic release of planulae resistant to limiting resources (Smith 1997, Kramer 2003).These strategies are likely to provide them predominance as an adult and even as a youth in many regions of the world (Smith 1997).The graphic association (on the MCA) of species of the second group with various habitat factors confirms the almost homogeneous distribution and frequency of juveniles under different conditions of the environment and its evidence of tolerance.
The robust pattern of high and moderate dependences displayed by A. tenuifolia could be accounted by their life history traits and their ability to withstand extreme conditions.Its vertical growth could provide an advantage by minimizing the effects of continental sediment and capturing sufficient light in areas presenting algal growth (Aronso et al. 2004).
The combination of turbidity, high sedimentation rate and increased nutrient concentrations, typical of continental reefs in this region (Aronso et al. 2004, Gardner et al. 2003, Acosta & Martínez 2006) especially of Isla Fuerte, favors the proliferation of macroalgae and decreases substrate availability for coral recruitment.Because it is a brooder and has a high level of reproductive effort and tolerance to adverse mediums, A. tenuifolia has been able to rapidly colonize empty substrates and increase their recruitment (Hughes & Jackson 1985).As a result, the presence of A. tenuifolia juveniles and adults has been favored by conditions that predominate in Caribbean reefs, where it is the most abundant and has replaced Porites sp. in recent decades (Aronso et al. 2004).The low number of A. tenuifolia juveniles in ocean reefs, where there is adequate numbers of adult reproductive individuals and where sedimentation is lower, may suggest that the species is a better competitor in stressful conditions, such as in areas with high silt-clay sediment (Miller & Barimo 2001), or that this ocean reef habitat type was not available at the time of larval selection and settlement (Edmunds 2004).
P. astreoides' light requirements are consistent with the high frequency of juveniles in horizontal locations, exposed to light, in clear water ocean reefs and recruitment at shallow depths (Vidal et al. 2005).Cofforth (1985) and Miller et al. (2000) assert that P. astreoides invest high levels of energy in protection and competition, producing mucus and presenting a gregarious behavior, which reduces possible cases of inter-specific competition and aggression by algae, giving them a higher rate of survival (Green et al. 2008).This strategy and the colonization of substrates devoid of sediment accumulation or with strong hydrodynamics, produces a lower energy expenditure for cleaning, energy that it can invest in competition with macroalgae such as Lobophora sp. and Halimeda sp., which occupy 12-15% of the substrate (Alvarado & Acosta 2009).Although Miller & Barimo (2001) and Green et al. (2008) state that P. astreoides is tolerant to different physical and biological habitats, juveniles appear to show specificity to habitat factors.
The favorable relationship of S. siderea juveniles and cryptic locations, inclined substrates and the lack of competitors could reveal the strategy employed by the larvae to colonize crevices, clefts and holes.This occurs when the larvae are faced with limitations of settlement substrate (e.g.due to high algal cover).This type of environment is abundant in ocean reefs at depths between 10-16m, where the reef has higher rugosity and presents a 3D structure.The amount of coarse and fine sediment accumulated in holes (greater than 1cm) presents a disadvantage for juveniles in this type of habitat as it may causes larvae to settle and recruit on inclined sections, precisely where the highest frequency of juveniles was observed.Smith (1997) proposes that S. siderea boasts a resistance to sedimentation, which probably gives it an advantage to settle in holes or rubble, as shown in the MCA in this study.Another disadvantage caused by the crevices is the reduction of direct light necessary for juvenile growth, which in turn is advantageous since it repels macroalgae and prevents the expenditure of energy on competition.Barrios (2000) stresses the high sensitivity of S. siderea juvenile to the aggression of other sessile organisms.However, although it has been proposed that S. siderea is an indicator species for pollution and sedimentation (Guzmán & Jiménez 1992), juveniles were scarce in continental environments where these factors are present, this may be due to low roughness and 3D structure of the system to recruit, a small number of adults and hence low reproductive effort of this spawner species, as Alvarado & Acosta (2009) found with lower number of gamets for M. annularis in Isla Grande, Colombian Caribbean, or due to lower larval survival in the water column by factors such as low salinity.
For its part, the dependence of juvenile A. lamarcki, L. cucullata and A. agaricites on a non-sedimented bottom, exposed to light and ocean reef habitats, reveals that recruitment and survival is lower in degraded areas (Hughes 1985), influenced by rivers.These species may secure the energy investment made by the larva during settlement, by locating itself specifically in areas sheltered from sediment (Hughes & Tanner 2000).Despite the slow colonization of new areas by A. lamarcki larvae, this species has high survival rate (Hughes 1988), higher than even A. agaricites and L. cucullata, the dominant species of juveniles in this study (593 and 354 juveniles, respectively) and in most Caribbean reefs (Hughes 1988, Kramer 2003).This potentially allows the species to maintain stable populations in areas with low sedimentation rate in the Colombian Caribbean, in areas like reef terraces presenting high hydrodynamic or a high gradient slope.The horizontal form in which these species grow explains their dependence on habitats exposed to light and for A. lamarcki, the monopolization of substrate in deep oceanic reefs the interaction with corals of the same species.Among all the species, A. agaricites is the most proficient at recruitment; it colonizes different microhabitats at depths ranging from 2-30m and can endure a broad spectrum of conditions in the environment (Vidal et al. 2005), in agreement with Chiappone & Sullivan (1996) and the study by Edmunds (2000) in Florida and St. John -Virgin Islands, respectively.
Juvenile P. porites, F. fragum and M. cavernosa had a dependence on ocean reefs, a microhabitat exposed to light and unoccupied by competitors.The susceptibility of P. porites to competition with macroalgae may explain its requirement to colonize spaces without competitors and with direct light.River & Edmunds (2001) assert that the growth rate of P. porites decreases up to 80% when surrounded by macroalgae.Contrary to Lewis (1974), who found that most larvae colonize cryptic sites, displaying a gregarious behavior by juveniles and adults, the presence of F. fragum juveniles was detected on coral rubble, in shallow terraces exposed to light and unoccupied by competitors.In turn, juveniles of M. cavernosa were associated with environments presenting high sediment load (60.8%).Their massive shape makes it more resistant to wave action and more efficient in the removal of sediments (Huitric & McField 2001, Martínez & Acosta 2005).Its relatively low recruitment rate (Smith 1997) is offset by an increase in its inter and intra-specific aggression (Hughes & Jackson 1985); this allows it to compete for potential growth-space.These survival strategies provide it dominance in many regions, as an adult (Martinez & Acosta 2005) and even as a juvenile.
Although the variables assessed here are those considered as critical to the recruitment of hermatypic corals in Caribbean reefs, exploring and evaluating other biotic (foraging) and abiotic factors of the habitat is essential, given the dependence of the species on specific factors.Moreover, the diverse causes of the observed patterns of dependency and the synergy between larval selection behavior and natural selection factors should also be investigated.This study presents an examination of the entire juvenile coral community; however, more detailed studies by species are necessary, particularly those that are rare, vulnerable and spawners with low fertility/fecundity rates and growth.In order to find possible dependences and associations not detected in this study and apply statistical independence analyses to other species, it is recommended a minimum sampling size of 100-150 juveniles/species.
The density of juveniles found is not considered sufficient to replace the dying adults in the reefs studied; this makes juvenile richness and density very low when compared to that of adult corals (Acosta et al. 2011).As the fitness of coral species decreases because of recruitment failure, so does the resilience of the reef, its recovery and populations' viability (Sale et al. 2010).Seven of the most active species in recruitment were brooders, with the exception of M. cavernosa and S. siderea, which as Carlon (2002) proposes, may be an indication of a potential change in dominance of the coral assemblage from building species for shortlived brooding species (juvenile A. agaricites, L. cucullata and P. astreoides).The low recruitment and the poor recovery of populations in the system also facilitate a community shift in continental reefs (algae for corals), as is the case in shallow areas of Isla Grande (Alvarado & Acosta 2009, Acosta et al. 2011).
The reality of Caribbean reefs is the great decline caused by anthropic disturbance and inland waters that increase the direct input of sediment and nutrients.This according to Alvarado & Acosta (2009) creates stress, population mortality, reduction in colony size, low number of breeding individuals, low reproductive output, lower quality and limited availability of habitat for larval selection (Montenegro & Acosta 2008, 2010), random settlement (Mundy & Babcock 2000, Vermeij et al. 2006), recruitment in sub-optimal locations and postsettlement mortality (Fabricius 2005, Vermeij et al. 2006, Victor 2008) and consequently, low recruitment (Acosta et al. 2011).To achieve successful coral recruitment, the favorable or ideal environment would occur in oceanic islands or areas not exposed to the effect of rivers, with good visibility, availability of dead coral for settlement, in a wide range of depths to set up the assembly of coral, inclined microhabitats that receive plenty of direct light, with not accumulated sediment and with low numbers of macroalgae or competitors.These habitat requirements should be considered by the decision makers in conservation efforts.
In conclusion, the presence, abundance and survival of juveniles depend on certain features of the habitat, which are species-specific.However, the juveniles of six species, mostly brooders, display specificity to one or two factors and high tolerance to a wide range of variables of the habitat evaluated.Most juveniles recruit on dead coral (except S. siderea and F. fragum on rubble), on inclined areas, exposed to direct light (except S. siderea and species of the Agaricia genus, cryptic), where F. fragum and the Porites genus are highly dependent on light (abundant in shallow and medium zones).This indicates the importance of clear water with low turbidity (low concentration of dissolved and particulate organic matter).Few species used rocky or encrusted red algae substrates; this reflects their low number and availability.Similarly, juveniles seem to prefer substrates where sediment does not accumulate, particularly A. lamarcki and L. cucullata, in contrast to what was observed for S. siderea, A. tenuifolia and M. cavernosa that tolerate high sediment loads.Juveniles recruit and survive on substrates occupied, primarily by macroalgae; this is evident for A. tenuifolia and in a lesser degree for A. lamarcki and P. astreoides, while S. siderea displays the lowest number of interactions.Most species use a wide range of depths (two or three categories), or specific ranges depending on their requirements and life history; juveniles of A. tenuifolia are favored, specifically between 10-16m and F. fragum between 2-8m.All species recruit disproportionately, in higher numbers in oceanic reefs except for A. tenuifolia, which prefers continental reefs, indicating that oceanic islands offer the best conditions for recruitment.

ACKNOWLEDGMENTS
To the Academic Vice-rectory of the Pontificia Universidad Javeriana that financed this project (0417; 4803), to Flavia Cárdenas, Mónica Sepúlveda, Claudia Villamil and Andrés Vidal for the field work that yielded the data used in this study, as well as to Margarita Ordoñez and Miguel Pinzón for their statistical advice.To Gypsy Espanol for the translation into the English version.(Traducciones Técnicas T y T; www.traduccionestyt.com).

Fig. 3 .
Fig. 3. Dependency relationship between habitat variables and the frequency of juveniles by species for the first group of species with the most dependences in respect to the macro and micro-habitat variables.(A) A. tenuifolia.n=80.(B) P. astreoides, n=248.(C) S. siderea, n=186.p=X 2 .C= contingency coefficient when p<0.05.*=statistical dependence.Y =Yates's correction.F =Fisher's exact test.=high dependence.=moderate dependence.=low dependence.Dotted line indicates independence.Italics indicate a biological variable.
(between the frequency of juveniles of nine coral species and physical and biological habitat conditions) determined by the MCA.+= Variables determined by the X 2 and MCA.Variables without an accompanying symbol were determined by X 2. N/A: insufficient data for some categories.

Fig. 4 .
Fig. 4. Dependency relationship between habitat variables and the frequency of juveniles by species that displayed the least number of dependences and recruit in a wide range of habitat conditions.(A) A. lamarcki, n=129.(B) L. cucullata, n=354.(C) A. agaricites, n=593.p=X 2 .C= contingency coefficient when p<0.05.*=statistical dependence.Y =Yates's correction.F =Fisher's exact test.=moderate dependence.=low dependence.Dotted line indicates independence.Italics indicate a biological variable.

Fig. 4 .
Fig. 4. Dependency relationship between habitat variables and the frequency of juveniles by species that displayed the least number of dependences and recruit in a wide range of habitat conditions.(D) P. porites, n=88.(E) F. fragum, n=173.(F) M. cavernosa, n=125.p=X 2 .C= contingency coefficient when p<0.05*=statistical dependence.Y =Yates's correction.F =Fisher's exact test.=moderate dependence.=low dependence.Dotted line indicates independence.Italics indicate a biological variable.

TABLE 1
Classification of habitat variables assessed for hermatypic coral recruits

TABLE 2
Maximum coefficient of association between variables (C max ) K: Rock, L: Encrusted red algae, M: Exposed to light, N: Cryptic, O: Horizontal substrate, P: Inclined substrate, Q: Vertical substrate, R: No sediment, S: Very fine, silt-clay sediment, T: Fine sediment, U: Coarse sediment, V: Dispersed sediment accumulated on the bottom, W: Moderate sediment accumulated, X: High sediment accumulated (See Table1).