Differential persistence favors habitat preferences that determine the distribution of a reef fish

Differential persistence favors habitat preferences that determine the distribution of a reef fish Abstract A central focus of population ecology is understanding what factors explain the distribution and abundance of organisms within their range. This is a key issue in marine systems, where many organisms produce dispersive larvae that develop offshore before returning to settle on benthic habitat. We investigated the distribution of the neon goby, Elacatinus lori, on sponge habitat and evaluated whether variation in the persistence of recently settled individuals (i.e. settlers) among different sponge types can result in habitat preferences and establish their observed distribution. We found that E. lori settlers were more likely to occur on large yellow tube sponges (Aplysina fistularis) than on small yellow sponges or brown tube sponges (Agelas conifera). An experiment seeding settlers onto multiple species and sizes of sponge habitat revealed that settlers persist longer on large yellow sponges than on small yellow sponges or brown sponges. Habitat preference experiments also indicated that settlers prefer large yellow sponges over small yellow sponges or brown sponges. Settlers achieved these preference behaviors using visual, but not chemical, cues. Finally, new settlers arriving from the water column were more likely to occur on large yellow sponges than on small yellow sponges or brown sponges, indicating that the observed habitat preferences existed independent of prior experience. These results support the hypothesis that E. lori have evolved behavioral preferences for sponge habitats that will maximize their post-settlement persistence, and that decisions at settlement will shape the population level pattern of settler distribution on coral reefs. INTRODUCTION A central question in the field of ecology is what causes the distribution and abundance of organisms throughout their range? For species with dispersive offspring, distribution patterns may be established prior to settlement as a result of variation in the supply of offspring across settlement habitats, or after settlement as a result of variation in the survival of offspring among habitat types (Forrester 1995; Schmitt and Holbrook 1996; Jones 1997; Schmitt and Holbrook 1999). When this variation in survival is consistent, natural selection is expected to favor the evolution of behavioral preferences for the habitat that confers the highest relative fitness (Booth and Wellington 1998). These ecological, evolutionary, and behavioral processes are not mutually exclusive and likely interact to influence distribution patterns. Although previous studies have integrated these processes to explain distribution patterns in terrestrial vertebrates (for a review see: Safran et al. 2007), relatively few studies have integrated these processes in marine fishes (but see: Wellington 1992; Shima 2001; Srinivasan 2003). Most coral reef fishes produce eggs and pelagic larvae that develop offshore before returning to settle on reef habitat. In some species, recently settled individuals (from here on “settlers”) occupy a transitional habitat before recruiting to the adult habitat (Lecchini et al. 2012). Settlers are often nonrandomly distributed, suggesting that ecological, evolutionary, and behavioral processes may influence their distribution (Booth and Wellington 1998). The influence of evolutionary processes on the distribution of settlers is often evaluated by measuring components of their fitness, including variation in the survival or persistence (i.e. a correlate of survival that indicates losses may be caused by mortality and/or movement) of individuals across settlement habitats. This variation occurs as a result of differences among habitats with respect to the risk of predation (Holbrook and Schmitt 2003; Almany 2004a), the availability of resources (Booth and Hixon 1999; Almany 2004b), and/or the intensity of density-dependent interactions (Holbrook and Schmitt 2002; Buston 2003). If survival varies predictably among habitat types, then larvae should prefer to settle on the habitat that maximizes their potential post-settlement survival (Booth and Wellington 1998; Safran et al. 2007). As larvae settle from the water column they are expected to prefer habitat characteristics or social cues that provide the most reliable information regarding their potential for post-settlement survival. Habitat characteristics may provide larvae with information regarding the quality of resources, such as food and shelter, or the risk of predation associated with a settlement habitat (Dixson et al. 2012; Lecchini et al. 2014). In contrast, social cues may provide information regarding the costs and/or benefits of joining groups at settlement (Stamps 1988; Safran et al. 2007). After settlement, individuals may benefit from enhanced growth, survival, and future reproductive opportunities provided by conspecifics (Jones 1988; Adam 2011), or they may experience costs due to antagonism and competition for resources (Schmitt and Holbrook 1999; Buston 2003). Though larvae may use both habitat characteristics and social cues when selecting settlement habitat, the relative importance of pre-settlement preference behaviors in establishing distribution patterns requires further investigation (Gutiérrez 1998). To use habitat characteristics and social cues, larvae must possess sensory systems and swimming abilities that allow them to detect, navigate to, and discriminate among potential settlement habitats (Leis et al. 2011). In laboratory and field experiments, settlers often display a behavioral preference for visual (Igulu et al. 2011), chemical (Atema et al. 2002), or auditory cues (Simpson et al. 2004) that emanate from their preferred settlement habitat. However, they likely use multiple sensory modalities when locating habitat under natural conditions (Lecchini et al. 2005b; Huijbers et al. 2012). Larvae may use these cues to discriminate among settlement habitats at large (e.g. among reefs; Gerlach et al. 2007) and small spatial scales (e.g. among microhabitats; Dixson et al. 2008). At small scales, some larvae discriminate among multiple host species (Elliott and Mariscal 2001), and even among the health of hosts of the same species (Feary et al. 2007). When presented with sensory cues from conspecifics, larvae may be either attracted to or actively avoid cues related to the presence (Sweatman 1985; Elliott and Mariscal 2001) and size (Igulu et al. 2011) of conspecifics. These behaviors suggest that many larvae are equipped to use both habitat characteristics and social cues to make fine-scale decisions when selecting among potential settlement habitats. However, few studies have investigated whether settlement decisions ultimately result in population level patterns of distribution (Booth and Wellington 1998). We used the neon goby Elacatinus lori (family Gobiidae) as a study species to advance our understanding of how behavioral preferences and post-settlement persistence interact to shape population level patterns of distribution. E. lori is an obligate sponge-dwelling goby endemic to the Mesoamerican Barrier Reef (Colin 2002). Like most reef fishes, E. lori has a pelagic larval phase, a settler phase that typically resides on the outer sponge wall, a recruit phase that moves from the outer sponge wall to living inside the sponge tube and a relatively sedentary adult phase that lives and breeds within the sponge (Figure 1; Shima 2001; Buston 2003; Ben-Tzvi et al. 2009). Although E. lori is known to occupy a variety of sponge species, they are most abundant in the yellow tube sponge Aplysina fistularis (Figure 2a, b) and the brown tube sponge Agelas conifera (Figure 2c, d) on the outer fore reef. These sponge species are similar in morphology, but preliminary observations suggest that E. lori are far more abundant in association with A. fistularis (from here on “yellow sponges”) than in A. conifera (from here on “brown sponges”). A study of the distribution of E. lori settlers occupying yellow sponges found that E. lori are more likely to occur in large, multi-tube sponges rather than in sponges with other morphologies (D’Aloia et al. 2011). The behavioral preferences and post-settlement processes that establish the distribution of E. lori across sponge habitats remain to be determined. Figure 1 View largeDownload slide The life cycle of Elacatinus lori in the yellow tube sponge Aplysina fistularis. After hatching, E. lori larvae develop in the water column for 26 ± 3.6 days (mean ± SD), reaching 8–9.5 mm SL (standard length), before settling onto sponge habitat (D’Aloia et al. 2015). Settlers, post-settlement individuals approximately 8–18 mm SL, live on the outer sponge wall before recruiting into a sponge tube. Recruits (>18 mm SL) grow to approximately 28 mm SL before reaching maturity and ultimately breeding within the sponge tube (D’Aloia et al. 2011; Rickborn and Buston 2015). Figure 1 View largeDownload slide The life cycle of Elacatinus lori in the yellow tube sponge Aplysina fistularis. After hatching, E. lori larvae develop in the water column for 26 ± 3.6 days (mean ± SD), reaching 8–9.5 mm SL (standard length), before settling onto sponge habitat (D’Aloia et al. 2015). Settlers, post-settlement individuals approximately 8–18 mm SL, live on the outer sponge wall before recruiting into a sponge tube. Recruits (>18 mm SL) grow to approximately 28 mm SL before reaching maturity and ultimately breeding within the sponge tube (D’Aloia et al. 2011; Rickborn and Buston 2015). Figure 2 View largeDownload slide Photographs of yellow tube sponges (Aplysina fistularis, a–b) and brown tube sponges (Agelas conifera, c–d) on the outer fore reef in Belize. Individuals of both sponge species often occur in single (a, c) and multi-tube (b, d) morphologies. Figure 2 View largeDownload slide Photographs of yellow tube sponges (Aplysina fistularis, a–b) and brown tube sponges (Agelas conifera, c–d) on the outer fore reef in Belize. Individuals of both sponge species often occur in single (a, c) and multi-tube (b, d) morphologies. Here, we build upon these previous studies by demonstrating 1) that the distribution of E. lori settlers across species and morphologies of sponge habitat is stable among reef sites surveyed in different years. We then investigate 2) the persistence of settlers inhabiting different sponge types, 3) the preference behaviors and sensory modalities that settlers use to discriminate among habitat types, and 4) whether observed preferences are based on prior experience. Our results support the hypothesis that E. lori have evolved behavioral preferences for sponge habitats that will maximize their post-settlement persistence, and that decisions at settlement shape the population level pattern of settler distribution on coral reefs. MATERIALS AND METHODS Study system This study was conducted in the South Water Caye Marine Reserve (SWCMR) on the Belizean Barrier Reef (BBR) in 2011, 2015, and 2017. Within the reserve, the barrier reef has several reef zones including a back reef, reef crest, inner fore reef, and outer fore reef (Rützler and Macintyre 1982). In 2011 and 2017, we experimentally tested habitat preferences of E. lori settlers on the back reef, behind Curlew Caye (16° 47’ 23” N, 88° 04’ 33” W); and in 2015, we investigated the distribution, persistence, and habitat preferences of E. lori settlers on the outer fore reef off South Water Caye (16° 48’ 92” N, 88° 04’ 89” W). E. lori settlers New E. lori settlers arriving from the water column lack pigment throughout most of their body, but can be identified by a small blue line on the snout and black spot on the caudal peduncle. Thus, we use the term “new settlers” to describe individuals ≥ 8 mm SL but < 10 mm SL that have minimal pigmentation along the body (Figure 3a). After 1–2 days, settlers ≥ 10 mm SL develop dark pigmentation and light blue lines that originate on the head and spread along the body as settlers increase in size (Figure 3b–d). These differences in size and pigmentation allow observers to discriminate among settlers at approximately ±1 mm SL size increments. Each component of this study was performed using either the complete size distribution of settlers (8–18 mm SL) or a subset of the size distribution, which specifically enabled us to: 1) observe the distribution of E. lori settlers on sponge habitat; 2) experimentally determine variation in settler persistence across sponge habitat types; 3) experimentally test the habitat preferences of E. lori settlers; and 4) determine the influence of prior experience on habitat preference behaviors (Table 1). Table 1 Summary of the size range of settlers that were used in the observational and experimental components of this study Experimental Design  Standard Length (SL)  8 ≤ × < 10  10 ≤ × ≤ 18 mm  Observational Survey of Natural Settler Distribution  Yes  Yes  Experimental Test of Settler Persistence  No  Yes  Experimental Test of Habitat Preferences  Yes  Yes  Effects of Prior Habitat Experience      a) Habitat preferences of settlers from brown sponges  Yes  Yes  b) Observational survey of new settler distribution  Yes  No  Experimental Design  Standard Length (SL)  8 ≤ × < 10  10 ≤ × ≤ 18 mm  Observational Survey of Natural Settler Distribution  Yes  Yes  Experimental Test of Settler Persistence  No  Yes  Experimental Test of Habitat Preferences  Yes  Yes  Effects of Prior Habitat Experience      a) Habitat preferences of settlers from brown sponges  Yes  Yes  b) Observational survey of new settler distribution  Yes  No  View Large Establishment of a sponge transect A transect of 60 tagged yellow tube sponges and 60 tagged brown tube sponges was established on the outer fore reef off South Water Caye for use in both observational and experimental components of this study. The size and number of sponge tubes per individual varies among sponge species and among individuals of the same species. Therefore, divers identified and tagged pairs of adjacent (≤5 m apart) yellow and brown sponges of similar size and tube number to control for potential spatial variation in the arrival of larvae to these sponges. Sponge pairs were chosen to represent the range of maximum tube lengths and number of tubes among individuals of each sponge species. The average maximum tube length of tagged yellow sponges was 39 ± 15.7 cm (mean ± SD) and average number of sponge tubes was 1.8 ± 1.3 tubes; the average maximum tube length of tagged brown sponges was 33 ± 10.6 cm (mean ± SD) and average number of sponge tubes was 2.8 ± 2.0 tubes. We excluded sponges less than 10 cm in maximum tube length from the study because E. lori are rarely observed in such small sponges (D’Aloia et al. 2011). We also excluded sponges that occurred deeper than 20 m, because dive time is limited below this depth. At each tagged sponge, divers recorded “habitat” variables including species, maximum tube length, number of tubes greater than 10 cm, and water depth at the base of the sponge, as well as “social” variables including the presence or absence of E. lori residents, density of residents in each sponge (i.e. the number of residents / the total number of sponge tubes) and the presence or absence of E. lori settlers on the outer sponge wall. The 120 tagged sponges in this transect were used to determine the influence of these variables on the: 1) distribution, 2) persistence, and 3) arrival of new E. lori settlers in subsequent experiments (Table 1). Experiment 1: observational survey of natural settler distribution Divers surveyed each of the 120 tagged sponges for the presence or absence of E. lori settlers to test the hypothesis that habitat and/or social variables are related to the natural settler distribution (8–18 mm SL; Table 1). We constructed a set of generalized linear models (distribution= binomial; link = logit) in R 3.2.3 (R Core Team 2015) to investigate the relationship between the presence or absence of an E. lori settler on a sponge (1 or 0) and all habitat and social variables (as defined above). Each variable was treated as an alternative hypothesis for the factors that predict the distribution of settlers. Experiment 2: experimental test of settler persistence Settlers were seeded onto sponges along the transect to test the hypothesis that the distribution of E. lori settlers is the result of variation in their persistence (i.e. defined here as the time a settler spent on a sponge as a result of mortality and/or movement) across settlement habitats. First, naturally occurring settlers were cleared from each of the tagged sponges, while adult gobies were left in place. Next, settlers were collected from yellow sponges on the fore reef outside of the transect. They were collected exclusively from yellow sponges because settlers were exceedingly rare on the brown sponge species, thus preventing the implementation of a fully cross-factored experimental design. Further, settled individuals were used because, 1) naïve larvae could not be effectively collected from the water column, and 2) at the time of this study we had not yet developed techniques for rearing larvae in captivity. To specifically address these issues, additional experiments were conducted to determine the effects of prior experience on yellow sponges, and to observe the habitat preference of new settlers arriving from the water column (see Experiment 4 below). We used only individuals 10–18 mm SL (SLmean ± SD = 12.37 mm ± 1.63; Table 1) so that experimental settlers seeded onto sponges could be distinguished from new settlers arriving naturally from the water column based on size (< 10 mm SL) and differences in pigmentation (Figure 3). Using a random stratified approach, settlers were placed on 4 different sponge habitats (mean ± SD; large yellow: 55.6 ± 11.7 cm; large brown: 43.3 ± 6.7 cm; small yellow: 28.1 ± 7.1 cm, small brown: 26.9 ± 5.8 cm), such that each of the habitats were seeded with the range of settler sizes that had been collected. We seeded one settler per sponge (n = 120 fish) because instances of 2 or more E. lori settling to the same sponge are rare (<10 % of settlement events). For each of the 120 tagged sponges, divers recorded the presence or absence of the seeded settler every other day for 2 weeks (n = 7 observations/settler). New settlers that arrived from the water column and individuals from elsewhere that moved to tagged sponges were identified using differences in size and pigmentation (Figure 3), removed from the sponge, and measured to confirm size (SL). Following completion of the first 2-week trial, a second trial was carried out using the same sponges, but with a new group of 120 E. lori settlers. Figure 3 View largeDownload slide Photographs of wild-caught Elacatinus lori settlers. New settlers (a) arriving on sponges range from 8 ≤ × < 10 mm SL and have minimal pigmentation along the body. Larger settlers ≥ 10 mm SL (b–d) gradually develop blue stripes and dark pigmentation along the body. Figure 3 View largeDownload slide Photographs of wild-caught Elacatinus lori settlers. New settlers (a) arriving on sponges range from 8 ≤ × < 10 mm SL and have minimal pigmentation along the body. Larger settlers ≥ 10 mm SL (b–d) gradually develop blue stripes and dark pigmentation along the body. We fit a mixed-effects Cox proportional hazards regression (Mills 2011) using the “coxme” package in R (Therneau 2015) to determine which habitat and social variables influence the persistence of settlers on sponge habitat. This approach allowed us to statistically control for changes in resident density by including resident density as a time dependent covariate. It also allowed us to statistically control for the initial size of the experimental settler by including settler SL as a covariate. Finally, we included sponge ID (i.e. individual sponge identity) as a random effect to control for repeated measures using the same set of 120 sponges in trials 1 and 2. Kaplan-Meier survival curves were plotted for the proportion of settlers remaining on a sponge as a function of time, stratified by sponge size and sponge species, and controlling for density of residents and the size of settlers. Experiment 3: experimental tests of habitat preferences We conducted habitat preference tests between May and August of 2011 using E. lori settlers to experimentally test the hypothesis that the distribution of settlers is established by preferences for sponge species and morphologies. A total of 344 settlers were collected from yellow sponges on the outer fore reef off Curlew Caye. As in the post-settlement persistence experiment (above), settlers were collected exclusively from yellow sponges. To determine the influence of prior experience on habitat preference behavior, an additional preference experiment was conducted using settlers collected from brown sponges (see Experiment 4 below). Settler size (SL) was measured prior to use, and spanned the full size range from 8 to 18 mm SL (SLmean ± SD = 11.75 mm ± 2.34; Table 1). A circular arena with a 6-m diameter was established in a shallow (<2 m deep) sand patch on the leeward side of Curlew Caye to test alternative hypotheses concerning the habitat characteristics and social cues that E. lori settlers might use to choose sponge habitat. Dye tests were conducted each day to measure current speed, direction, and to observe mixing. For each experiment, 2 habitat types were placed in alternating positions at 60-degree intervals along the arena’s perimeter (e.g. 3 yellow sponges × 3 PVC pipes; Figure 4a) and the position of habitat types was rotated 180° midway through each experiment. Following a 2 min acclimation period, individual settlers were released from a glass jar onto the sand in the center of the arena and allowed to choose from among the habitat types being tested. Settlers that did not move from the center of the arena within 5 min of release were excluded from the experiment (n = 82 of 344 fish). Preliminary observations showed that settlers remained on their first habitat choice for >24 h. Thus, “preference” was recorded by a snorkeling observer (for similar methodology see: Lecchini et al. 2005b) as the first habitat with which a settler made contact. A test ended once the settler made contact with the outer surface of either a sponge or PVC pipe. Data were analyzed using a chi-square goodness of fit test (P = 0.05). Figure 4 View largeDownload slide Diagram of the in situ arena configured for testing E. lori preferences for sponge vs. PVC pipe habitat, under three sensory conditions: (a) no cues manipulated, (b) chemical cues manipulated—optically clear plastic covers sealed to eliminate water flow, (c) visual cues manipulated—optically opaque mesh covers transparent to water flow. For clarity, we only illustrate water (wavy lines) flowing through the top of mesh covers, but in practice, water flows through both the top and sides. (Note: diagram is not drawn to scale). Figure 4 View largeDownload slide Diagram of the in situ arena configured for testing E. lori preferences for sponge vs. PVC pipe habitat, under three sensory conditions: (a) no cues manipulated, (b) chemical cues manipulated—optically clear plastic covers sealed to eliminate water flow, (c) visual cues manipulated—optically opaque mesh covers transparent to water flow. For clarity, we only illustrate water (wavy lines) flowing through the top of mesh covers, but in practice, water flows through both the top and sides. (Note: diagram is not drawn to scale). Sensory cue manipulations were conducted using the arena to determine which sensory cues E. lori settlers use when choosing settlement habitat. Initial trials were conducted without manipulation of sensory cues (Figure 4a). To manipulate chemical cues, a clear plastic cylinder (PETG, Visipak, USA) was placed over each habitat choice in the arena, sealed with plastic film and secured approximately 5 cm into the sand to eliminate chemical cues carried by water flow emanating from sponges (Figure 4b). During a preliminary trial, dye was released within the cylinder to verify that chemical cues could not escape the cylinder. To manipulate visual cues, a mesh cylinder was placed over each habitat choice in the arena (Figure 4c). A dye test was conducted to verify that water potentially carrying chemical cues could escape through the top and sides of the mesh. Settler habitat preferences were evaluated using a chi-square goodness of fit test (P = 0.05). Five preference experiments, each comparing 2 habitat types, were conducted to explain the observed distribution of E. lori settlers on sponge habitat. Preference experiments included comparisons of yellow sponges versus grey PVC pipes, yellow sponges versus brown sponges, large (>35 cm) versus small (10 cm) yellow sponges, multi-tube versus single-tube yellow sponges, and E. lori resident-occupied versus unoccupied yellow sponges. Experiments were conducted under different sensory cue conditions for each arena in which settlers displayed a significant habitat preference (no manipulation, manipulation of chemical cues, manipulation of visual cues; Figure 4a–c). The yellow sponge versus gray PVC pipe combination was used to demonstrate that E. lori could choose between habitat types. Each pair of habitats, except large versus small yellow sponges, were size-matched by tube length to control for the effect of sponge size on preference behavior. In multi-tube versus single-tube preference tests, 2 single-tube sponges of similar tube length (~35 cm) were placed side-by-side to create repeatable multi-tube sponges, thereby controlling for tube size variation that occurs naturally among multi-tube sponges. Finally, to allow time for resident E. lori to establish themselves and generate chemical cues, residents collected from yellow sponges were relocated to single-tube sponges in the arena 2 h before the start of resident-occupied vs. unoccupied preference tests. Experiment 4: evaluating the effects of prior habitat experience Due to constraints imposed by the study system, settlers used in the experimental components of this study were not naïve with respect to habitat. They were collected on yellow sponges, where they most commonly occur. Therefore, the preceding experiments cannot discriminate between a preference for a particular habitat type and a habitat with which the fish has had prior experience. To specifically address this issue, 1) we tested the habitat preferences of E. lori settlers collected from brown sponges and 2) conducted an observational study to determine the distribution of new settlers arriving from the water column (<10 mm SL with minimal pigmentation; Table 1, Figure 3a). To test the hypothesis that settlers prefer habitat on which they have had prior experience, settlers collected from brown sponges were provided a choice between yellow versus brown sponges in the arena, as described above in Experiment 3) the experimental test of habitat preferences. To observe the distribution of new settlers arriving from the water column, the 120 tagged sponges were cleared of settlers and then surveyed for new settlers every 24–48 h throughout 2 lunar cycles (28 May–25 July 2015). We constructed a generalized linear mixed-effects model (GLMM; distribution = binomial; link = logit) using the “lme4” package in R (Bates et al. 2015) to evaluate how habitat and social variables influence the distribution of new settlers on sponge habitat. The arrival of multiple new settlers on an individual sponge was rare. Therefore, we investigated the relationship between the presence or absence of an E. lori settler (0 or 1, respectively) and all habitat and social variables. Sponge ID was included as a random effect to control for repeated observations of the same 120 tagged sponges. Model selection We constructed statistical models to determine the effects of habitat and social variables on the distribution and persistence of E. lori settlers using a forward stepwise approach in an information theoretic framework. In this framework, a model with an additional variable was retained if the corrected Akaike information criteria (AICc) score was lower than other candidate models by ≥ 2 ∆AICc units. When multiple candidate models were within 2 ∆AICc either 1) the most parsimonious model was selected, or 2) in the case that candidate models contained the same number of variables, the model with the lowest AICc score was chosen. Candidate models within 2 ∆AICc were compared to the reduced model from the previous step using a likelihood-ratio test. If the reduced model and candidate models were not significantly different, then the reduced model was selected as the best-fit model based on parsimony. RESULTS Observational survey of natural settler distribution Logistic regression analyses revealed that sponge species and maximum tube length predicted the presence or absence of settlers on sponge habitat. In the best-fit model, the probability of a settler occurring on a sponge increases with sponge size and is higher on yellow than on brown sponges (Table 2; Figure 5). The results of this analysis indicate that E. lori settlers are more likely to be found on large yellow sponges rather than on small yellow sponges or brown sponges. Table 2 Summary of the best-fit logistic regression model evaluating the association between multiple habitat and social variables and the presence of E. lori settlers on sponge habitat Predictor  Estimate (OR)  SE  z value  P value  Intercept  −3.39 (0.03)  0.711  −4.766  <0.001***  Maximum Tube Length (cm)  0.04 (1.04)  0.016  2.665  0.008**  Sponge Species (yellow)  1.05 (2.88)  0.485  2.177  0.029*  Predictor  Estimate (OR)  SE  z value  P value  Intercept  −3.39 (0.03)  0.711  −4.766  <0.001***  Maximum Tube Length (cm)  0.04 (1.04)  0.016  2.665  0.008**  Sponge Species (yellow)  1.05 (2.88)  0.485  2.177  0.029*  For sponge species, yellow sponges are compared to the reference group brown sponges; (OR), odds ratios. *P < 0.05, **P < 0.001, ***P < 0.001. View Large Figure 5 View largeDownload slide Predictors of the natural distribution of Elacatinus lori settlers on sponge habitat. Lines represent the probability of settler occurrence as a function of maximum tube length and sponge species estimated from the best-fit logistic regression (dashed, yellow sponges; solid, brown sponges). Shaded regions represent the 95% confidence intervals. Figure 5 View largeDownload slide Predictors of the natural distribution of Elacatinus lori settlers on sponge habitat. Lines represent the probability of settler occurrence as a function of maximum tube length and sponge species estimated from the best-fit logistic regression (dashed, yellow sponges; solid, brown sponges). Shaded regions represent the 95% confidence intervals. Experimental test of settler persistence Of 240 settlers that were seeded onto sponges, 44 persisted throughout the entire two-week experimental period. A mixed-effects Cox proportional hazards regression showed that multiple factors predict settler persistence on sponge habitat. The model including sponge species, maximum tube length, resident density, and the starting size of the seeded settlers was the best predictor of settler persistence on sponge habitat ( χ52 = 65.45, P < 0.001). The hazard of disappearance was lower on yellow than on brown sponges and decreased with increasing sponge size (Table 3; Figure 6). In contrast, the hazard of disappearance increased with increasing resident density and was higher for larger (presumably older) settlers (Table 3). Thus, settlers persist longer on large yellow sponges than they do on small yellow sponges or on brown sponges. Further, settlers persist longer on sponges with low densities of residents than they do on sponges with high densities of residents. Table 3 Summary of the best-fit mixed-effects Cox proportional hazards regression evaluating the association between multiple habitat and social variables and E. lori settler persistence on sponge habitat Predictor  Estimate (exp.)  S.E.  χ2  df  P value  Sponge Species (yellow)  −0.815 (0.44)  0.185  38.624  1  <0.001***  Maximum Tube Length (cm)  −0.030 (0.97)  0.007  9.303  1  0.002**  Resident Density  0.809 (2.25)  0.193  14.080  1  <0.001***  Settler Size (mm)  0.125 (1.13)  0.049  6.299  1  0.012*  Predictor  Estimate (exp.)  S.E.  χ2  df  P value  Sponge Species (yellow)  −0.815 (0.44)  0.185  38.624  1  <0.001***  Maximum Tube Length (cm)  −0.030 (0.97)  0.007  9.303  1  0.002**  Resident Density  0.809 (2.25)  0.193  14.080  1  <0.001***  Settler Size (mm)  0.125 (1.13)  0.049  6.299  1  0.012*  For sponge species, yellow sponges are compared to the reference group brown sponges; (exp.), exponentiated coefficient, *P < 0.05, **P < 0.001, ***P < 0.001. View Large Figure 6 View largeDownload slide Kaplan–Meier survival curves for the proportion of settlers remaining on a yellow or brown sponge as a function of time, stratified by sponge size and species, and controlling for resident density and the size of settlers. Lines (Kaplan–Meier survival curves) represent the relationship between proportion of settlers remaining and the independent variables included in the model (see Table 3). Figure 6 View largeDownload slide Kaplan–Meier survival curves for the proportion of settlers remaining on a yellow or brown sponge as a function of time, stratified by sponge size and species, and controlling for resident density and the size of settlers. Lines (Kaplan–Meier survival curves) represent the relationship between proportion of settlers remaining and the independent variables included in the model (see Table 3). Experimental tests of habitat preference Yellow sponges versus PVC pipe We tested the hypothesis that settlers prefer yellow sponges over gray PVC pipe in each sensory cue treatment. Settlers displayed a preference for yellow sponges over gray PVC pipe habitat when sensory cues were unmanipulated (χ2 = 19, df = 1, P < 0.001, n = 19) and when chemical cues were manipulated using clear plastic covers (χ2 = 13, df = 1, P < 0.001, n = 13), but not when visual cues were manipulated using opaque mesh covers (χ2 = 0.53, df = 1, P = 0.47, n = 17). Yellow sponges versus brown sponges We then tested the hypothesis that settlers collected from yellow sponges prefer yellow sponges over brown sponges. Settlers preferred yellow over brown sponges when sensory cues were unmanipulated (χ2 = 13.33, df = 1, P < 0.001, n = 30; Figure 7) and when chemical cues were manipulated using clear plastic covers (χ2 = 4.5, df = 1, P = 0.034, n = 32; Figure 7), but not when visual cues were manipulated using opaque mesh covers (χ2 = 0.2, df = 1, P = 0.65, n = 20; Figure 7). Figure 7 View largeDownload slide Proportion of E. lori settlers collected from yellow tube sponges that chose yellow versus brown tube sponges during trials with no cues manipulated, chemical cues manipulated, or visual cues manipulated. Significant habitat preferences indicated by * (chi-square test; P < 0.05). Figure 7 View largeDownload slide Proportion of E. lori settlers collected from yellow tube sponges that chose yellow versus brown tube sponges during trials with no cues manipulated, chemical cues manipulated, or visual cues manipulated. Significant habitat preferences indicated by * (chi-square test; P < 0.05). Large versus small yellow sponges We tested the hypothesis that settlers would prefer large (>35 cm tube length) over small (10 cm tube length) yellow sponges. Settlers displayed a preference for large over small yellow sponges when sensory cues were unmanipulated (χ2 = 16.2, df= 1, P < 0.001, n = 20; Figure 8) and when chemical cues were manipulated using clear plastic covers (χ2 = 16.2, df= 1, P < 0.001, n = 20; Figure 8), but not when visual cues were manipulated using opaque mesh covers (χ2 = 1.64, df = 1, P = 0.20, n = 22; Figure 8). Figure 8 View largeDownload slide Proportion of settlers collected from yellow tube sponges that chose large vs. small yellow sponges during trials with no cues manipulated, chemical cues manipulated, or visual cues manipulated. Significant habitat preferences indicated by * (chi-square test; P < 0.05). Figure 8 View largeDownload slide Proportion of settlers collected from yellow tube sponges that chose large vs. small yellow sponges during trials with no cues manipulated, chemical cues manipulated, or visual cues manipulated. Significant habitat preferences indicated by * (chi-square test; P < 0.05). Multi versus single-tube yellow sponges We tested the hypothesis that settlers would prefer multi-tube over single-tube yellow sponges in each sensory cue treatment. Settlers displayed no preference for multi vs. single-tube sponges when sensory cues were unmanipulated (χ2 = 0.53, df = 1, P = 0.53, n = 23). Thus, we did not conduct additional experiments manipulating chemical or visual sensory cues. Resident-occupied versus unoccupied yellow sponges Finally, we tested the hypothesis that settlers would prefer yellow sponges already occupied by E. lori residents over unoccupied sponges. Settlers had no preference for resident-occupied vs. unoccupied yellow sponges when cues were unmanipulated (χ2 = 0, df = 1, P = 1, n = 16). Thus, we did not conduct additional experiments manipulating chemical or visual sensory cues. Evaluating the effects of prior habitat experience Despite having been collected from brown sponges, settlers preferred yellow over brown sponges in the arena (χ2 = 18.24, df = 1, P < 0.0001, n = 29; Figure 9a). Considering the distribution of new settlers arriving to sponges, maximum tube length, sponge species, and sponge depth predicted the presence or absence of a new settler (n = 142 new settlers). In the best-fit model, the probability of a new settler occurring on a sponge increases with increasing sponge size (Table 4; Figure 9b), is higher on yellow than on brown sponges (Table 4; Figure 9b), and increases with the depth at which a sponge was located (Table 4). Table 4 Summary of the best-fit mixed-effects logistic regression model evaluating the association between multiple habitat and social variables and the arrival of E. lori settlers on sponge habitat Predictor  Estimate (OR)  SE  z value  P value  (Intercept)  −14.119 (0)  1.386  −10.188  <0.001***  Maximum Tube Length (cm)  0.045 (1.05)  0.007  6.102  <0.001***  Sponge Species (yellow)  1.723 (5.60)  0.307  5.608  <0.001***  Depth  0.477 (1.61)  0.084  5.706  <0.001***  Predictor  Estimate (OR)  SE  z value  P value  (Intercept)  −14.119 (0)  1.386  −10.188  <0.001***  Maximum Tube Length (cm)  0.045 (1.05)  0.007  6.102  <0.001***  Sponge Species (yellow)  1.723 (5.60)  0.307  5.608  <0.001***  Depth  0.477 (1.61)  0.084  5.706  <0.001***  For sponge species, yellow sponges are compared to the reference group brown sponges; (OR), odds ratio; *P < 0.05, **P < 0.001, ***P < 0.001. View Large Figure 9 View largeDownload slide Evaluating the effects of prior habitat experience on preference behaviors of E. lori settlers. (a) Proportion of E. lori settlers collected from brown tube sponges that chose yellow vs. brown sponges during arena trials with no cues manipulated. Significant habitat preferences indicated by * (chi-square test; P < 0.05). (b) Probability of settlement to a sponge as a function of maximum tube length and sponge species, with sponge depth held at its species mean. Lines represent the relationship between the probability of settlement to a sponge and the independent variables estimated from the parameters of the logistic model (dashed, yellow sponges; solid, brown sponges); Shaded regions represent the 95% confidence intervals. Figure 9 View largeDownload slide Evaluating the effects of prior habitat experience on preference behaviors of E. lori settlers. (a) Proportion of E. lori settlers collected from brown tube sponges that chose yellow vs. brown sponges during arena trials with no cues manipulated. Significant habitat preferences indicated by * (chi-square test; P < 0.05). (b) Probability of settlement to a sponge as a function of maximum tube length and sponge species, with sponge depth held at its species mean. Lines represent the relationship between the probability of settlement to a sponge and the independent variables estimated from the parameters of the logistic model (dashed, yellow sponges; solid, brown sponges); Shaded regions represent the 95% confidence intervals. DISCUSSION In organisms with dispersive offspring, distribution patterns may be established by variation among habitats in offspring supply, pre-settlement preference behaviors, and post-settlement mortality (Booth and Wellington 1998; Jenkins 2005; Bohn et al. 2013). Here we demonstrate that the distribution, persistence, and preferences of Elacatinus lori settlers are primarily influenced by characteristics of their settlement habitat: settlers occur more often on (Table 2; Figure 5), persist longer on (Table 3; Figure 6) and prefer large yellow sponges (Table 4; Figures 7–9) rather than small yellow sponges or brown sponges. Taken together, these results suggest that variation in post-settlement persistence selects for habitat preferences that ultimately determine the distribution of E. lori settlers on the reef. Distribution of Elacatinus lori Species-specific microhabitat associations are relatively common in marine systems (Thiel et al. 2003; Baeza 2008), especially in coral reef fishes (Munday et al. 1997; Elliott and Mariscal 2001; Bonin 2011). Microhabitat characteristics such as host species, location, size, and morphology are often linked with their quality as settlement habitat (Connell and Jones 1991; Munday 2001; Harrington et al. 2004). Here, we found that the occurrence of E. lori settlers is predicted by: 1) the sponge species and 2) maximum sponge tube length (Table 2; Figure 5), suggesting that the quality of sponge habitat varies by sponge species and size. These results confirm and extend the conclusions of previous surveys conducted within the South Water Caye Marine Reserve (Carrie Bow Cay, D’Aloia, et al. 2011; Curlew Cay, Majoris J.E., unpublished data). Although these surveys were completed in different locations and years, the variables that predict settler distribution were consistent, suggesting that settler distribution patterns were spatially and temporally stable over this time frame. Stable distributions are often attributed to consistency in settlement behavior, habitat availability, and post-settlement mortality (Holbrook et al. 2000; Jenkins 2005; Bohn et al. 2013). In E. lori, the stable spatiotemporal distribution of settlers suggests that the processes regulating their distribution may also be spatially and temporally stable. Variation in settler persistence Determining the evolutionary consequences associated with alternate settlement habitats is necessary to understand the processes that establish population level distribution patterns (Booth and Wellington 1998; Safran et al. 2007). For E. lori, settler persistence was related to sponge species and size, suggesting that there is predictable variation in settler mortality and/or movement among sponge habitats (Table 3; Figure 6). Similar observations have been made in coral-dwelling gobies, where individuals benefit from increased growth and survival when associating with their preferred species of coral, and at a finer scale, from corals with smaller inter-branch spacing (Munday 2000; Munday 2001). In this system, the outer surface of large yellow sponges are typically highly rugose and may shelter E. lori settlers from predation, compared to the smooth surface of brown sponges (Figure 2), which leaves them more exposed. We also found that resident density had a negative effect on settler persistence (Table 3). In other fishes, antagonistic interactions with resident conspecifics often result in a higher risk of eviction and mortality for settlers (Holbrook and Schmitt 2002; Almany 2003). In the clown anemonefish Amphiprion percula, residents evict (Buston 2003) and may cannibalize (Elliott et al. 1995) incoming settlers when anemone saturation is high. Similarly, E. lori residents evict settlers from occupied sponge tubes, forcing them to the outer sponge wall and, on occasion, have been observed cannibalizing settlers. As sponge tubes become saturated with residents, higher rates of eviction are expected to result in increased mortality or movement of settlers. As settlers were too small to tag in this study, it was not possible to determine the relative contributions of mortality and movement to variation in settler persistence among sponge habitats. However, previous studies have shown that microhabitat specialist reef fishes display strong site fidelity and are unlikely to move among habitats spaced greater than 1–2 m apart (Buston 2003; Feary 2007). In addition, we have observed that E. lori settlers suffer high predation rates when they attempt to leave their sponge (Majoris, personal observation). This observation suggests that mortality plays an important role in determining patterns of settler persistence, while successful movement among sponges may be low. Additional work is necessary to determine the relative influence of mortality and movement on persistence in this species. Habitat preferences As individuals select settlement habitat, they are expected to rely on habitat characteristics or social cues that are positively correlated with habitat quality (Booth and Wellington 1998; Safran et al. 2007). For example, Harrington et al. (2004) found that coral larvae evolved behavioral preferences for the species and growth form of crustose coralline algae that enhance their post-settlement survival. In this study, we found that E. lori settlers prefer (Figures 6 and 7) and persist longer on (Figure 6) large yellow sponges over small yellow sponges or brown sponges. This result suggests that E. lori settlers have evolved a preference for sponge characteristics that are positively correlated with post-settlement persistence. These habitat preferences, in combination with differential post-settlement persistence, can explain their abundance on large yellow sponges. The use of social cues from conspecifics for assessing habitat quality is common across taxa (Sweatman 1983; Stamps 1991; Muller 1998; Fletcher 2007). Despite this, E. lori settlers did not discriminate between resident occupied and unoccupied sponges, suggesting that they did not use social cues which could have improved their post-settlement persistence (Table 3). There are several potential explanations for this mismatch: 1) settlers may not be able to assess resident density—residents are, after all, hidden within sponge tubes, 2) movement of residents between sponges could weaken selection for avoidance behaviors, and 3) the benefits of settling on large yellow sponges may outweigh the costs of settling on occupied sponges. Our results suggest that E. lori use habitat characteristics, rather than social cues, to choose settlement habitat. The role of sensory cues in habitat preference Though individuals may use visual, chemical, and/or auditory cues at settlement, this study showed that E. lori settlers rely on visual cues when selecting sponge habitat over small spatial scales (Figures 7 and 8). This was an unexpected result, as sponges are known to emit a range of chemicals that were expected to provide settlers with cues for habitat selection (Pawlik 2011), and previous research has demonstrated the importance of chemosensory cues during habitat selection by some marine species (Krug and Manzi 1999; Forward et al. 2001; Lecchini and Nakamura 2013). However, previous studies have shown that other reef fishes also rely on visual, rather than chemical, cues for habitat recognition (Lecchini et al. 2005a; Igulu et al. 2011; Lecchini et al. 2014). The large size and bright coloration of yellow sponges make them conspicuous on the reef. These characteristics may allow settling E. lori to quickly locate sponge habitat using visual cues. Effects of prior habitat experience A limitation of this study was the inability to collect naïve larvae for use in habitat preference experiments, or to collect sufficient quantities of settlers from brown sponges to implement a fully cross-factored experimental design. Given this limitation, all settlers were collected from and had prior experience on yellow sponges. Sale (1971) found that, when given a choice between two coral species as settlement habitat, recently-settled Dascyllus aruanus preferred the species of coral from which they were collected. In contrast, Danilowicz (1996) demonstrated that both naïve Dascyllus albisella settlers and wild-caught settlers collected from several species of coral preferred a single coral species. Given these mixed results for other species, we considered the potential influence of prior experience on the results of this study. We found that settlers collected from brown sponges preferred yellow sponges over the habitat on which they had prior experience (Figure 9a), indicating that prior experience did not influence their preference behaviors. Further, by observing the distribution of new settlers arriving on both sponge species, we found that new E. lori settlers were more likely to occur on large yellow sponges than small yellow sponges or brown sponges (Table 4; Figure 9b). Since persistence was similar on yellow and brown sponges within the first 24–48 h (Figure 6), these results suggest that E. lori larvae chose large yellow sponges at settlement. Settlers also prefer (Figures 7 and 8) and ultimately persist longer on (Table 4; Figure 6) large yellow sponges under experimental conditions. Taken together, these results support the hypothesis that E. lori have evolved behavioral preferences for sponge habitats that will maximize their post-settlement persistence, and that decisions at settlement shape the population level pattern of settler distribution on coral reefs. Evolution of habitat preferences In light of these results, it is interesting to consider under what conditions habitat preferences evolve. For preference behaviors to evolve, 1) sensory capabilities and habitat cues must allow individuals to reliably discriminate among habitat types, 2) the habitat cues must be strongly correlated with post-settlement mortality, and 3) individuals must be able to sample and choose freely among habitat types (Fretwell and Lucas 1969). In this study, E. lori settlers prefer habitat characteristics that are strongly associated with their post-settlement persistence, irrespective of their prior experience. While we cannot quantify the relative contributions of post-settlement mortality and movement in establishing persistence, we predict that variation in mortality among sponges has selected for behavioral preferences for the habitat on which settlers are most likely to persist (i.e. large yellow sponges). Given these preferences, individuals that move post-settlement will likely migrate from less preferred sponges toward large yellow sponges when possible, reinforcing the distribution pattern established by habitat preferences at settlement. However, there are many species in which variation in mortality across settlement habitats is not associated with pre-settlement preference behaviors (Amphiprion percula: Buston 2003, 2004; Stegastes leucostictus: Almany 2003). In these species, the presence of reliable sensory cues for discriminating between habitat types and assumption of free choice between settlement habitats may be violated (Levin et al. 2000). Therefore, individuals would benefit from settling on the first habitat that they encounter rather than continuing to search for an optimal habitat (Buston 2004; Stamps et al. 2005). CONCLUSION Habitat preferences and post-settlement processes are often invoked to explain the distribution and abundance of organisms throughout their range. Here, we demonstrate that settling E. lori select sponge habitat using characteristics that are associated with their potential for post-settlement persistence, and that decisions at settlement shape the population level pattern of distribution on coral reefs. Although this study focuses on a coral reef fish, pre- and post-settlement processes can play an important role in establishing the distribution patterns of diverse taxa. FUNDING This work was funded in part by a startup award from the Trustees of Boston University and National Science Foundation grants (grant numbers OCE-1260424, OCE-1459546) to PMB. Additional funding was provided by a Lerner-Gray Grant awarded by the American Museum of Natural History, a Warren McLeod Summer Research Scholarship awarded by Boston University and a Doctoral Dissertation Improvement Grant (grant number IOS-1501651) awarded by the National Science Foundation to JEM. Data accessibility: Analyses reported in this article can be reproduced using the data provided by Majoris et al. (2017). We would like to thank the Belizean government and Fisheries Department for permission to conduct this research. Thank you to the staff at Wee Wee Caye Marine Station and the International Zoological Expeditions field station for the use of their facilities. Special thanks to Kevin David, Earl David, Udel Forman, Romain Chaput, Emma Shlatter, James Garner, and Alissa Rickborn for assistance in the field, and Kate Langwig for assistance with statistical analyses. Jelle Atema, Karen Warkentin, Jacqueline Webb, John Crawford, Kristina Cohen and two anonymous reviewers provided helpful comments on this manuscript. This study represents a portion of the dissertation work of JEM. All work was approved by the Belize Fisheries Department and the Boston University IACUC (protocol # 10–036). REFERENCES Adam TC. 2011. High-quality habitat and facilitation ameliorate competitive effects of prior residents on new settlers. Oecologia . 166: 121– 130. Google Scholar CrossRef Search ADS PubMed  Almany GR. 2003. Priority effects in coral reef fish communities. Ecology . 84: 1920– 1935. Google Scholar CrossRef Search ADS   Almany GR. 2004a. Differential effects of habitat complexity, predators and competitors on abundance of juvenile and adult coral reef fishes. Oecologia . 141: 105– 113. Google Scholar CrossRef Search ADS   Almany GR. 2004b. Does increased habitat complexity reduce predation and competition in coral reef fish assemblages? Oikos . 106: 275– 284. Google Scholar CrossRef Search ADS   Atema J, Kingsford MJ, Gerlach G. 2002. Larval reef fish could use odour for detection, retention and orientation to reefs. Mar Ecol Prog Ser . 241: 151– 160. Google Scholar CrossRef Search ADS   Baeza JA. 2008. Social monogamy in the shrimp Pontonia margarita, a symbiont of Pinctada mazatlanica, off the Pacific coast of Panama. Mar Biol . 153: 387– 395. Google Scholar CrossRef Search ADS   Bates D, Maechler M, Bolker B, Walker S. 2015. Fitting Linear Mixed-Effects Models Using lme4. J Stat Softw . 67: 1– 48. Google Scholar CrossRef Search ADS   Ben-Tzvi O, Kiflawi M, Polak O, Abelson A. 2009. The effect of adult aggression on habitat selection by settlers of two coral-dwelling damselfishes. PLoS One . 4: e5511. Google Scholar CrossRef Search ADS PubMed  Bohn K, Richardson CA, Jenkins SR. 2013. The importance of larval supply, larval habitat selection and post-settlement mortality in determining intertidal adult abundance of the invasive gastropod Crepidula fornicata. J Exp Mar Biol Ecol . 440: 132– 140. Google Scholar CrossRef Search ADS   Bonin MC. 2011. Specializing on vulnerable habitat: Acropora selectivity among damselfish recruits and the risk of bleaching-induced habitat loss. Coral Reefs . 31: 287– 297. Google Scholar CrossRef Search ADS   Booth DJ, Hixon MA. 1999. Food ration and condition affect early survival of the coral reef damselfish, Stegastes partitus. Oecologia . 121: 364– 368. Google Scholar CrossRef Search ADS PubMed  Booth DJ, Wellington G. 1998. Settlement preferences in coral-reef fishes: Effects on patterns of adult and juvenile distributions, individual fitness and population structure. Aust J Ecol . 23: 274– 279. Google Scholar CrossRef Search ADS   Buston PM. 2003. Forcible eviction and prevention of recruitment in the clown anemonefish. Behav Ecol . 14: 576– 582. Google Scholar CrossRef Search ADS   Buston PM. 2004. Territory inheritance in clownfish. Proc Biol Sci . 271( suppl 4): S252– S254. Google Scholar CrossRef Search ADS PubMed  Colin PL. 2002. A new species of sponge-dwelling Elacatinus (Pisces: Gobiidae) from the western Caribbean. Zootaxa . 106: 1– 7. Google Scholar CrossRef Search ADS   Connell SD, Jones GP. 1991. The influence of habitat complexity on postrecruitment processes in a temperate reef fish population. J Exp Mar Biol Ecol . 151: 271– 294. Google Scholar CrossRef Search ADS   D’Aloia CC, Bogdanowicz SM, Francis RK, Majoris JE, Harrison RG, Buston PM. 2015. Patterns, causes, and consequences of marine larval dispersal. Proc Natl Acad Sci U S A . 112: 13940– 13945. Google Scholar CrossRef Search ADS PubMed  D’Aloia CC, Majoris JE, Buston PM. 2011. Predictors of the distribution and abundance of a tube sponge and its resident goby. Coral Reefs . 30: 777– 786. Google Scholar CrossRef Search ADS   Danilowicz BS. 1996. Choice of Coral Species by Naive and Field-Caught Damselfish. Copeia . 1996: 735– 739. Google Scholar CrossRef Search ADS   Dixson D, Pratchett M, Munday P. 2012. Reef fishes innately distinguish predators based on olfactory cues associated with recent prey items rather than individual species. Anim Behav . 84: 45– 51. Google Scholar CrossRef Search ADS   Dixson DL, Jones GP, Munday PL, Planes S, Pratchett MS, Srinivasan M, Syms C, Thorrold SR. 2008. Coral reef fish smell leaves to find island homes. Proc Biol Sci . 275: 2831– 2839. Google Scholar CrossRef Search ADS PubMed  Elliott JK, Elliott JM, Mariscal RN. 1995. Host selection, location, and association behaviors of anemonefishes in field settlement experiments. Mar Biol . 122: 377– 389. Google Scholar CrossRef Search ADS   Elliott JK, Mariscal RN. 2001. Coexistence of nine anemonefish species: differential host and habitat utilization, size and recruitment. Mar Biol . 138: 23– 36. Google Scholar CrossRef Search ADS   Feary DA. 2007. The influence of resource specialization on the response of reef fish to coral disturbance. Mar Biol . 153: 153– 161. Google Scholar CrossRef Search ADS   Feary DA, Almany GR, McCormick MI, Jones GP. 2007. Habitat choice, recruitment and the response of coral reef fishes to coral degradation. Oecologia . 153: 727– 737. Google Scholar CrossRef Search ADS PubMed  Fletcher RJJr. 2007. Species interactions and population density mediate the use of social cues for habitat selection. J Anim Ecol . 76: 598– 606. Google Scholar CrossRef Search ADS PubMed  Forrester GE. 1995. Strong density-dependent survival and recruitment regulate the abundance of a coral reef fish. Oecologia . 103: 275– 282. Google Scholar CrossRef Search ADS PubMed  Forward RB, Tankersley RA, Rittschof D. 2001. Cues for metamorphosis of Brachyuran crabs: an overview. Am Zool . 41: 1108– 1122. Fretwell SD, Lucas H. 1969. On territorial behavior and other factors influencing habitat distribution in birds. Acta Biotheor . 19: 16– 36. Google Scholar CrossRef Search ADS   Gerlach G, Atema J, Kingsford MJ, Black KP, Miller-Sims V. 2007. Smelling home can prevent dispersal of reef fish larvae. Proc Natl Acad Sci U S A . 104: 858– 863. Google Scholar CrossRef Search ADS PubMed  Gutiérrez L. 1998. Habitat selection by recruits establishes local patterns of adult distribution in two species of damselfishes: Stegastes dorsopunicans and S. planifrons. Oecologia . 115: 268– 277. Google Scholar CrossRef Search ADS PubMed  Harrington L, Fabricius K, De’ath G, Negri A. 2004. Recognition and Selection of Settlement Substrata Determine Post-Settlement Survival in Corals. Ecology . 85: 3428– 3437. Google Scholar CrossRef Search ADS   Holbrook SJ, Forrester GE, Schmitt RJ. 2000. Spatial patterns in abundance of a damselfish reflect availability of suitable habitat. Oecologia . 122: 109– 120. Google Scholar CrossRef Search ADS PubMed  Holbrook SJ, Schmitt RJ. 2002. Competition for Shelter Space Causes Density-Dependent Predation Mortality in Damselfishes. Ecology . 83: 2855. Google Scholar CrossRef Search ADS   Holbrook SJ, Schmitt RJ. 2003. Spatial and temporal variation in mortality of newly settled damselfish: patterns, causes and co-variation with settlement. Oecologia . 135: 532– 541. Google Scholar CrossRef Search ADS PubMed  Huijbers CM, Nagelkerken I, Lössbroek PA, Schulten IE, Siegenthaler A, Holderied MW, Simpson SD. 2012. A test of the senses: fish select novel habitats by responding to multiple cues. Ecology . 93: 46– 55. Google Scholar CrossRef Search ADS PubMed  Igulu MM, Nagelkerken I, Fraaije R, van Hintum R, Ligtenberg H, Mgaya YD. 2011. The potential role of visual cues for microhabitat selection during the early life phase of a coral reef fish (Lutjanus fulviflamma). J Exp Mar Biol Ecol . 401: 118– 125. Google Scholar CrossRef Search ADS   Jenkins SR. 2005. Larval habitat selection, not larval supply, determines settlement patterns and adult distribution in two chthamalid barnacles. J Anim Ecol . 74: 893– 904. Google Scholar CrossRef Search ADS   Jones GP. 1988. Some interactions between residents and recruits in two coral reef fishes. J Exp Mar Biol Ecol . 114: 169– 182. Google Scholar CrossRef Search ADS   Jones GP. 1997. Relationships between recruitment and postrecruitment processes in lagoonal populations of two coral reef fishes. J Exp Mar Biol Ecol . 213: 231– 246. Google Scholar CrossRef Search ADS   Krug PJ, Manzi AE. 1999. Waterborne and Surface-Associated Carbohydrates as Settlement Cues for Larvae of the Specialist Marine Herbivore Alderia modesta. Biol Bull . 197: 94– 103. Google Scholar CrossRef Search ADS PubMed  Lecchini D, Carassou L, Frederich B, Nakamura Y, Mills SC, Galzin R. 2012. Effects of alternate reef states on coral reef fish habitat associations. Environ Biol Fishes . 94: 421– 429. Google Scholar CrossRef Search ADS   Lecchini D, Nakamura Y. 2013. Use of chemical cues by coral reef animal larvae for habitat selection. Aquat Biol . 19: 231– 238. Google Scholar CrossRef Search ADS   Lecchini D, Peyrusse K, Lanyon RG, Lecellier G. 2014. Importance of visual cues of conspecifics and predators during the habitat selection of coral reef fish larvae. C R Biol . 337: 345– 351. Google Scholar CrossRef Search ADS PubMed  Lecchini D, Planes S, Galzin R. 2005a. Experimental assessment of sensory modalities of coral-reef fish larvae in the recognition of their settlement habitat. Behav Ecol Sociobiol . 58: 18– 26. Google Scholar CrossRef Search ADS   Lecchini D, Shima J, Banaigs B, Galzin R. 2005b. Larval sensory abilities and mechanisms of habitat selection of a coral reef fish during settlement. Oecologia . 143: 326– 334. Google Scholar CrossRef Search ADS   Leis JM, Siebeck U, Dixson DL. 2011. How Nemo finds home: the neuroecology of dispersal and of population connectivity in larvae of marine fishes. Integr Comp Biol . 51: 826– 843. Google Scholar CrossRef Search ADS PubMed  Levin PS, Tolimieri N, Nicklin M, Sale PF. 2000. Integrating individual behavior and population ecology: the potential for habitat-dependent population regulation in a reef fish. Behav Ecol . 11: 565– 571. Google Scholar CrossRef Search ADS   Majoris JE, D’Aloia CC, Francis RK, Buston PM. 2017. Data from: differential persistence favors habitat preferences that determine the distribution of a reef fish. Dryad Digital Repository . http://dx.doi.org/10.5061/dryad.vn712. Mills M. 2011. Introducing Survival and Event History Analysis . London: SAGE Publications. Google Scholar CrossRef Search ADS   Muller KL. 1998. The role of conspecifics in habitat settlement in a territorial grasshopper. Anim Behav . 56: 479– 485. Google Scholar CrossRef Search ADS PubMed  Munday PL. 2001. Fitness consequences of habitat use and competition among coral-dwelling fishes. Oecologia . 128: 585– 593. Google Scholar CrossRef Search ADS PubMed  Munday PL. 2000. Interactions Between Habitat Use and Patterns of Abundance in Coral-dwelling Fishes of the Genus Gobiodon. Environ Biol Fishes . 58: 355– 369. Google Scholar CrossRef Search ADS   Munday PL, Jones GP, Caley MJ. 1997. Habitat specialisation and the distribution and abundance of coral-dwelling gobies. Mar Ecol Prog Ser . 152: 227– 239. Google Scholar CrossRef Search ADS   Pawlik JR. 2011. The Chemical Ecology of Sponges on Caribbean Reefs: Natural Products Shape Natural Systems. BioScience . 61: 888– 898. Google Scholar CrossRef Search ADS   R Core Team. 2015. R: A language and environment for statistical computing . Vienna (Austria): R Foundation for Statistical Computing. Rickborn AJ, Buston PM. 2015. Life-history transitions of the coral-reef fish Elacatinus lori. J Fish Biol . 86: 637– 650. Google Scholar CrossRef Search ADS   Rützler K, Macintyre IG. 1982. The habitat distribution and community structure of the barrier reef complex at Carrie Bow Cay, Belize. In: The Atlantic barrier reef ecosystem at Carrie Bow Cay, Belize, I. structure and communities . Washington, DC: Smithsonian Institution Press. (Smithsonian contributions to the marine sciences). p. 9– 45. Safran RJ, Doerr VA, Sherman PW, Doerr ED, Flaxman SM, Winkler DW. 2007. Group breeding in vertebrates: linking individual-and population-level approaches. Evol Ecol Res . 9: 1163. Sale PF. 1971. Apparent effect of prior experience on a habitat preference exhibited by the reef fish, Dascyllus aruanus (Pisces: Pomacentridae). Anim. Behav . 19: 251– 256. Google Scholar CrossRef Search ADS   Schmitt R, Holbrook S. 1996. Local-scale Patterns of Larval Settlement in a Planktivorous Damselfish—Do They Predict Recruitment? Mar. Freshw Res . 47: 449– 463. Google Scholar CrossRef Search ADS   Schmitt RJ, Holbrook SJ. 1999. Settlement and recruitment of three damselfish species: larval delivery and competition for shelter space. Oecologia . 118: 76– 86. Google Scholar CrossRef Search ADS PubMed  Shima JS. 2001. Recruitment of a coral reef fish: roles of settlement, habitat, and postsettlement losses. Ecology . 82: 2190– 2199. Google Scholar CrossRef Search ADS   Simpson SD, Meekan MG, McCauley RD, Jeffs A. 2004. Attraction of settlement-stage coral reef fishes to reef noise. Mar Ecol Prog Ser . 276: 263– 268. Google Scholar CrossRef Search ADS   Srinivasan M. 2003. Depth distributions of coral reef fishes: the influence of microhabitat structure, settlement, and post-settlement processes. Oecologia . 137: 76– 84. Google Scholar CrossRef Search ADS PubMed  Stamps JA. 1988. Conspecific Attraction and Aggregation in Territorial Species. Am Nat . 131: 329– 347. Google Scholar CrossRef Search ADS   Stamps JA. 1991. The effect of conspecifics on habitat selection in territorial species. Behav. Ecol Sociobiol . 28: 29– 36. Google Scholar CrossRef Search ADS   Stamps JA, Krishnan VV, Reid ML. 2005. Search costs and habitat selection by dispersers. Ecology . 86: 510– 518. Google Scholar CrossRef Search ADS   Sweatman HP. 1985. The influence of adults of some coral reef fishes on larval recruitment. Ecol Monogr . 55: 469– 485. Google Scholar CrossRef Search ADS   Sweatman HPA. 1983. Influence of conspecifics on choice of settlement sites by larvae of two pomacentrid fishes (Dascyllus aruanus and D. reticulatus) on coral reefs. Mar Biol . 75: 225– 229. Google Scholar CrossRef Search ADS   Therneau T. 2015. coxme: Mixed Effects Cox Models . (R package version 2.2–5). https://cran.r-project.org/package=coxme Thiel M, Zander A, Valdivia N, Baeza JA, Rueffler C. 2003. Host fidelity of a symbiotic porcellanid crab: the importance of host characteristics. J Zool . 261: 353– 362. Google Scholar CrossRef Search ADS   Wellington GM. 1992. Habitat selection and juvenile persistence control the distribution of two closely related Caribbean damselfishes. Oecologia . 90: 500– 508. Google Scholar CrossRef Search ADS PubMed  © The Author(s) 2018. Published by Oxford University Press on behalf of the International Society for Behavioral Ecology. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Behavioral Ecology Oxford University Press

Differential persistence favors habitat preferences that determine the distribution of a reef fish

Loading next page...
 
/lp/ou_press/differential-persistence-favors-habitat-preferences-that-determine-the-ijWtx4w9oZ
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of the International Society for Behavioral Ecology. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com
ISSN
1045-2249
eISSN
1465-7279
D.O.I.
10.1093/beheco/arx189
Publisher site
See Article on Publisher Site

Abstract

Abstract A central focus of population ecology is understanding what factors explain the distribution and abundance of organisms within their range. This is a key issue in marine systems, where many organisms produce dispersive larvae that develop offshore before returning to settle on benthic habitat. We investigated the distribution of the neon goby, Elacatinus lori, on sponge habitat and evaluated whether variation in the persistence of recently settled individuals (i.e. settlers) among different sponge types can result in habitat preferences and establish their observed distribution. We found that E. lori settlers were more likely to occur on large yellow tube sponges (Aplysina fistularis) than on small yellow sponges or brown tube sponges (Agelas conifera). An experiment seeding settlers onto multiple species and sizes of sponge habitat revealed that settlers persist longer on large yellow sponges than on small yellow sponges or brown sponges. Habitat preference experiments also indicated that settlers prefer large yellow sponges over small yellow sponges or brown sponges. Settlers achieved these preference behaviors using visual, but not chemical, cues. Finally, new settlers arriving from the water column were more likely to occur on large yellow sponges than on small yellow sponges or brown sponges, indicating that the observed habitat preferences existed independent of prior experience. These results support the hypothesis that E. lori have evolved behavioral preferences for sponge habitats that will maximize their post-settlement persistence, and that decisions at settlement will shape the population level pattern of settler distribution on coral reefs. INTRODUCTION A central question in the field of ecology is what causes the distribution and abundance of organisms throughout their range? For species with dispersive offspring, distribution patterns may be established prior to settlement as a result of variation in the supply of offspring across settlement habitats, or after settlement as a result of variation in the survival of offspring among habitat types (Forrester 1995; Schmitt and Holbrook 1996; Jones 1997; Schmitt and Holbrook 1999). When this variation in survival is consistent, natural selection is expected to favor the evolution of behavioral preferences for the habitat that confers the highest relative fitness (Booth and Wellington 1998). These ecological, evolutionary, and behavioral processes are not mutually exclusive and likely interact to influence distribution patterns. Although previous studies have integrated these processes to explain distribution patterns in terrestrial vertebrates (for a review see: Safran et al. 2007), relatively few studies have integrated these processes in marine fishes (but see: Wellington 1992; Shima 2001; Srinivasan 2003). Most coral reef fishes produce eggs and pelagic larvae that develop offshore before returning to settle on reef habitat. In some species, recently settled individuals (from here on “settlers”) occupy a transitional habitat before recruiting to the adult habitat (Lecchini et al. 2012). Settlers are often nonrandomly distributed, suggesting that ecological, evolutionary, and behavioral processes may influence their distribution (Booth and Wellington 1998). The influence of evolutionary processes on the distribution of settlers is often evaluated by measuring components of their fitness, including variation in the survival or persistence (i.e. a correlate of survival that indicates losses may be caused by mortality and/or movement) of individuals across settlement habitats. This variation occurs as a result of differences among habitats with respect to the risk of predation (Holbrook and Schmitt 2003; Almany 2004a), the availability of resources (Booth and Hixon 1999; Almany 2004b), and/or the intensity of density-dependent interactions (Holbrook and Schmitt 2002; Buston 2003). If survival varies predictably among habitat types, then larvae should prefer to settle on the habitat that maximizes their potential post-settlement survival (Booth and Wellington 1998; Safran et al. 2007). As larvae settle from the water column they are expected to prefer habitat characteristics or social cues that provide the most reliable information regarding their potential for post-settlement survival. Habitat characteristics may provide larvae with information regarding the quality of resources, such as food and shelter, or the risk of predation associated with a settlement habitat (Dixson et al. 2012; Lecchini et al. 2014). In contrast, social cues may provide information regarding the costs and/or benefits of joining groups at settlement (Stamps 1988; Safran et al. 2007). After settlement, individuals may benefit from enhanced growth, survival, and future reproductive opportunities provided by conspecifics (Jones 1988; Adam 2011), or they may experience costs due to antagonism and competition for resources (Schmitt and Holbrook 1999; Buston 2003). Though larvae may use both habitat characteristics and social cues when selecting settlement habitat, the relative importance of pre-settlement preference behaviors in establishing distribution patterns requires further investigation (Gutiérrez 1998). To use habitat characteristics and social cues, larvae must possess sensory systems and swimming abilities that allow them to detect, navigate to, and discriminate among potential settlement habitats (Leis et al. 2011). In laboratory and field experiments, settlers often display a behavioral preference for visual (Igulu et al. 2011), chemical (Atema et al. 2002), or auditory cues (Simpson et al. 2004) that emanate from their preferred settlement habitat. However, they likely use multiple sensory modalities when locating habitat under natural conditions (Lecchini et al. 2005b; Huijbers et al. 2012). Larvae may use these cues to discriminate among settlement habitats at large (e.g. among reefs; Gerlach et al. 2007) and small spatial scales (e.g. among microhabitats; Dixson et al. 2008). At small scales, some larvae discriminate among multiple host species (Elliott and Mariscal 2001), and even among the health of hosts of the same species (Feary et al. 2007). When presented with sensory cues from conspecifics, larvae may be either attracted to or actively avoid cues related to the presence (Sweatman 1985; Elliott and Mariscal 2001) and size (Igulu et al. 2011) of conspecifics. These behaviors suggest that many larvae are equipped to use both habitat characteristics and social cues to make fine-scale decisions when selecting among potential settlement habitats. However, few studies have investigated whether settlement decisions ultimately result in population level patterns of distribution (Booth and Wellington 1998). We used the neon goby Elacatinus lori (family Gobiidae) as a study species to advance our understanding of how behavioral preferences and post-settlement persistence interact to shape population level patterns of distribution. E. lori is an obligate sponge-dwelling goby endemic to the Mesoamerican Barrier Reef (Colin 2002). Like most reef fishes, E. lori has a pelagic larval phase, a settler phase that typically resides on the outer sponge wall, a recruit phase that moves from the outer sponge wall to living inside the sponge tube and a relatively sedentary adult phase that lives and breeds within the sponge (Figure 1; Shima 2001; Buston 2003; Ben-Tzvi et al. 2009). Although E. lori is known to occupy a variety of sponge species, they are most abundant in the yellow tube sponge Aplysina fistularis (Figure 2a, b) and the brown tube sponge Agelas conifera (Figure 2c, d) on the outer fore reef. These sponge species are similar in morphology, but preliminary observations suggest that E. lori are far more abundant in association with A. fistularis (from here on “yellow sponges”) than in A. conifera (from here on “brown sponges”). A study of the distribution of E. lori settlers occupying yellow sponges found that E. lori are more likely to occur in large, multi-tube sponges rather than in sponges with other morphologies (D’Aloia et al. 2011). The behavioral preferences and post-settlement processes that establish the distribution of E. lori across sponge habitats remain to be determined. Figure 1 View largeDownload slide The life cycle of Elacatinus lori in the yellow tube sponge Aplysina fistularis. After hatching, E. lori larvae develop in the water column for 26 ± 3.6 days (mean ± SD), reaching 8–9.5 mm SL (standard length), before settling onto sponge habitat (D’Aloia et al. 2015). Settlers, post-settlement individuals approximately 8–18 mm SL, live on the outer sponge wall before recruiting into a sponge tube. Recruits (>18 mm SL) grow to approximately 28 mm SL before reaching maturity and ultimately breeding within the sponge tube (D’Aloia et al. 2011; Rickborn and Buston 2015). Figure 1 View largeDownload slide The life cycle of Elacatinus lori in the yellow tube sponge Aplysina fistularis. After hatching, E. lori larvae develop in the water column for 26 ± 3.6 days (mean ± SD), reaching 8–9.5 mm SL (standard length), before settling onto sponge habitat (D’Aloia et al. 2015). Settlers, post-settlement individuals approximately 8–18 mm SL, live on the outer sponge wall before recruiting into a sponge tube. Recruits (>18 mm SL) grow to approximately 28 mm SL before reaching maturity and ultimately breeding within the sponge tube (D’Aloia et al. 2011; Rickborn and Buston 2015). Figure 2 View largeDownload slide Photographs of yellow tube sponges (Aplysina fistularis, a–b) and brown tube sponges (Agelas conifera, c–d) on the outer fore reef in Belize. Individuals of both sponge species often occur in single (a, c) and multi-tube (b, d) morphologies. Figure 2 View largeDownload slide Photographs of yellow tube sponges (Aplysina fistularis, a–b) and brown tube sponges (Agelas conifera, c–d) on the outer fore reef in Belize. Individuals of both sponge species often occur in single (a, c) and multi-tube (b, d) morphologies. Here, we build upon these previous studies by demonstrating 1) that the distribution of E. lori settlers across species and morphologies of sponge habitat is stable among reef sites surveyed in different years. We then investigate 2) the persistence of settlers inhabiting different sponge types, 3) the preference behaviors and sensory modalities that settlers use to discriminate among habitat types, and 4) whether observed preferences are based on prior experience. Our results support the hypothesis that E. lori have evolved behavioral preferences for sponge habitats that will maximize their post-settlement persistence, and that decisions at settlement shape the population level pattern of settler distribution on coral reefs. MATERIALS AND METHODS Study system This study was conducted in the South Water Caye Marine Reserve (SWCMR) on the Belizean Barrier Reef (BBR) in 2011, 2015, and 2017. Within the reserve, the barrier reef has several reef zones including a back reef, reef crest, inner fore reef, and outer fore reef (Rützler and Macintyre 1982). In 2011 and 2017, we experimentally tested habitat preferences of E. lori settlers on the back reef, behind Curlew Caye (16° 47’ 23” N, 88° 04’ 33” W); and in 2015, we investigated the distribution, persistence, and habitat preferences of E. lori settlers on the outer fore reef off South Water Caye (16° 48’ 92” N, 88° 04’ 89” W). E. lori settlers New E. lori settlers arriving from the water column lack pigment throughout most of their body, but can be identified by a small blue line on the snout and black spot on the caudal peduncle. Thus, we use the term “new settlers” to describe individuals ≥ 8 mm SL but < 10 mm SL that have minimal pigmentation along the body (Figure 3a). After 1–2 days, settlers ≥ 10 mm SL develop dark pigmentation and light blue lines that originate on the head and spread along the body as settlers increase in size (Figure 3b–d). These differences in size and pigmentation allow observers to discriminate among settlers at approximately ±1 mm SL size increments. Each component of this study was performed using either the complete size distribution of settlers (8–18 mm SL) or a subset of the size distribution, which specifically enabled us to: 1) observe the distribution of E. lori settlers on sponge habitat; 2) experimentally determine variation in settler persistence across sponge habitat types; 3) experimentally test the habitat preferences of E. lori settlers; and 4) determine the influence of prior experience on habitat preference behaviors (Table 1). Table 1 Summary of the size range of settlers that were used in the observational and experimental components of this study Experimental Design  Standard Length (SL)  8 ≤ × < 10  10 ≤ × ≤ 18 mm  Observational Survey of Natural Settler Distribution  Yes  Yes  Experimental Test of Settler Persistence  No  Yes  Experimental Test of Habitat Preferences  Yes  Yes  Effects of Prior Habitat Experience      a) Habitat preferences of settlers from brown sponges  Yes  Yes  b) Observational survey of new settler distribution  Yes  No  Experimental Design  Standard Length (SL)  8 ≤ × < 10  10 ≤ × ≤ 18 mm  Observational Survey of Natural Settler Distribution  Yes  Yes  Experimental Test of Settler Persistence  No  Yes  Experimental Test of Habitat Preferences  Yes  Yes  Effects of Prior Habitat Experience      a) Habitat preferences of settlers from brown sponges  Yes  Yes  b) Observational survey of new settler distribution  Yes  No  View Large Establishment of a sponge transect A transect of 60 tagged yellow tube sponges and 60 tagged brown tube sponges was established on the outer fore reef off South Water Caye for use in both observational and experimental components of this study. The size and number of sponge tubes per individual varies among sponge species and among individuals of the same species. Therefore, divers identified and tagged pairs of adjacent (≤5 m apart) yellow and brown sponges of similar size and tube number to control for potential spatial variation in the arrival of larvae to these sponges. Sponge pairs were chosen to represent the range of maximum tube lengths and number of tubes among individuals of each sponge species. The average maximum tube length of tagged yellow sponges was 39 ± 15.7 cm (mean ± SD) and average number of sponge tubes was 1.8 ± 1.3 tubes; the average maximum tube length of tagged brown sponges was 33 ± 10.6 cm (mean ± SD) and average number of sponge tubes was 2.8 ± 2.0 tubes. We excluded sponges less than 10 cm in maximum tube length from the study because E. lori are rarely observed in such small sponges (D’Aloia et al. 2011). We also excluded sponges that occurred deeper than 20 m, because dive time is limited below this depth. At each tagged sponge, divers recorded “habitat” variables including species, maximum tube length, number of tubes greater than 10 cm, and water depth at the base of the sponge, as well as “social” variables including the presence or absence of E. lori residents, density of residents in each sponge (i.e. the number of residents / the total number of sponge tubes) and the presence or absence of E. lori settlers on the outer sponge wall. The 120 tagged sponges in this transect were used to determine the influence of these variables on the: 1) distribution, 2) persistence, and 3) arrival of new E. lori settlers in subsequent experiments (Table 1). Experiment 1: observational survey of natural settler distribution Divers surveyed each of the 120 tagged sponges for the presence or absence of E. lori settlers to test the hypothesis that habitat and/or social variables are related to the natural settler distribution (8–18 mm SL; Table 1). We constructed a set of generalized linear models (distribution= binomial; link = logit) in R 3.2.3 (R Core Team 2015) to investigate the relationship between the presence or absence of an E. lori settler on a sponge (1 or 0) and all habitat and social variables (as defined above). Each variable was treated as an alternative hypothesis for the factors that predict the distribution of settlers. Experiment 2: experimental test of settler persistence Settlers were seeded onto sponges along the transect to test the hypothesis that the distribution of E. lori settlers is the result of variation in their persistence (i.e. defined here as the time a settler spent on a sponge as a result of mortality and/or movement) across settlement habitats. First, naturally occurring settlers were cleared from each of the tagged sponges, while adult gobies were left in place. Next, settlers were collected from yellow sponges on the fore reef outside of the transect. They were collected exclusively from yellow sponges because settlers were exceedingly rare on the brown sponge species, thus preventing the implementation of a fully cross-factored experimental design. Further, settled individuals were used because, 1) naïve larvae could not be effectively collected from the water column, and 2) at the time of this study we had not yet developed techniques for rearing larvae in captivity. To specifically address these issues, additional experiments were conducted to determine the effects of prior experience on yellow sponges, and to observe the habitat preference of new settlers arriving from the water column (see Experiment 4 below). We used only individuals 10–18 mm SL (SLmean ± SD = 12.37 mm ± 1.63; Table 1) so that experimental settlers seeded onto sponges could be distinguished from new settlers arriving naturally from the water column based on size (< 10 mm SL) and differences in pigmentation (Figure 3). Using a random stratified approach, settlers were placed on 4 different sponge habitats (mean ± SD; large yellow: 55.6 ± 11.7 cm; large brown: 43.3 ± 6.7 cm; small yellow: 28.1 ± 7.1 cm, small brown: 26.9 ± 5.8 cm), such that each of the habitats were seeded with the range of settler sizes that had been collected. We seeded one settler per sponge (n = 120 fish) because instances of 2 or more E. lori settling to the same sponge are rare (<10 % of settlement events). For each of the 120 tagged sponges, divers recorded the presence or absence of the seeded settler every other day for 2 weeks (n = 7 observations/settler). New settlers that arrived from the water column and individuals from elsewhere that moved to tagged sponges were identified using differences in size and pigmentation (Figure 3), removed from the sponge, and measured to confirm size (SL). Following completion of the first 2-week trial, a second trial was carried out using the same sponges, but with a new group of 120 E. lori settlers. Figure 3 View largeDownload slide Photographs of wild-caught Elacatinus lori settlers. New settlers (a) arriving on sponges range from 8 ≤ × < 10 mm SL and have minimal pigmentation along the body. Larger settlers ≥ 10 mm SL (b–d) gradually develop blue stripes and dark pigmentation along the body. Figure 3 View largeDownload slide Photographs of wild-caught Elacatinus lori settlers. New settlers (a) arriving on sponges range from 8 ≤ × < 10 mm SL and have minimal pigmentation along the body. Larger settlers ≥ 10 mm SL (b–d) gradually develop blue stripes and dark pigmentation along the body. We fit a mixed-effects Cox proportional hazards regression (Mills 2011) using the “coxme” package in R (Therneau 2015) to determine which habitat and social variables influence the persistence of settlers on sponge habitat. This approach allowed us to statistically control for changes in resident density by including resident density as a time dependent covariate. It also allowed us to statistically control for the initial size of the experimental settler by including settler SL as a covariate. Finally, we included sponge ID (i.e. individual sponge identity) as a random effect to control for repeated measures using the same set of 120 sponges in trials 1 and 2. Kaplan-Meier survival curves were plotted for the proportion of settlers remaining on a sponge as a function of time, stratified by sponge size and sponge species, and controlling for density of residents and the size of settlers. Experiment 3: experimental tests of habitat preferences We conducted habitat preference tests between May and August of 2011 using E. lori settlers to experimentally test the hypothesis that the distribution of settlers is established by preferences for sponge species and morphologies. A total of 344 settlers were collected from yellow sponges on the outer fore reef off Curlew Caye. As in the post-settlement persistence experiment (above), settlers were collected exclusively from yellow sponges. To determine the influence of prior experience on habitat preference behavior, an additional preference experiment was conducted using settlers collected from brown sponges (see Experiment 4 below). Settler size (SL) was measured prior to use, and spanned the full size range from 8 to 18 mm SL (SLmean ± SD = 11.75 mm ± 2.34; Table 1). A circular arena with a 6-m diameter was established in a shallow (<2 m deep) sand patch on the leeward side of Curlew Caye to test alternative hypotheses concerning the habitat characteristics and social cues that E. lori settlers might use to choose sponge habitat. Dye tests were conducted each day to measure current speed, direction, and to observe mixing. For each experiment, 2 habitat types were placed in alternating positions at 60-degree intervals along the arena’s perimeter (e.g. 3 yellow sponges × 3 PVC pipes; Figure 4a) and the position of habitat types was rotated 180° midway through each experiment. Following a 2 min acclimation period, individual settlers were released from a glass jar onto the sand in the center of the arena and allowed to choose from among the habitat types being tested. Settlers that did not move from the center of the arena within 5 min of release were excluded from the experiment (n = 82 of 344 fish). Preliminary observations showed that settlers remained on their first habitat choice for >24 h. Thus, “preference” was recorded by a snorkeling observer (for similar methodology see: Lecchini et al. 2005b) as the first habitat with which a settler made contact. A test ended once the settler made contact with the outer surface of either a sponge or PVC pipe. Data were analyzed using a chi-square goodness of fit test (P = 0.05). Figure 4 View largeDownload slide Diagram of the in situ arena configured for testing E. lori preferences for sponge vs. PVC pipe habitat, under three sensory conditions: (a) no cues manipulated, (b) chemical cues manipulated—optically clear plastic covers sealed to eliminate water flow, (c) visual cues manipulated—optically opaque mesh covers transparent to water flow. For clarity, we only illustrate water (wavy lines) flowing through the top of mesh covers, but in practice, water flows through both the top and sides. (Note: diagram is not drawn to scale). Figure 4 View largeDownload slide Diagram of the in situ arena configured for testing E. lori preferences for sponge vs. PVC pipe habitat, under three sensory conditions: (a) no cues manipulated, (b) chemical cues manipulated—optically clear plastic covers sealed to eliminate water flow, (c) visual cues manipulated—optically opaque mesh covers transparent to water flow. For clarity, we only illustrate water (wavy lines) flowing through the top of mesh covers, but in practice, water flows through both the top and sides. (Note: diagram is not drawn to scale). Sensory cue manipulations were conducted using the arena to determine which sensory cues E. lori settlers use when choosing settlement habitat. Initial trials were conducted without manipulation of sensory cues (Figure 4a). To manipulate chemical cues, a clear plastic cylinder (PETG, Visipak, USA) was placed over each habitat choice in the arena, sealed with plastic film and secured approximately 5 cm into the sand to eliminate chemical cues carried by water flow emanating from sponges (Figure 4b). During a preliminary trial, dye was released within the cylinder to verify that chemical cues could not escape the cylinder. To manipulate visual cues, a mesh cylinder was placed over each habitat choice in the arena (Figure 4c). A dye test was conducted to verify that water potentially carrying chemical cues could escape through the top and sides of the mesh. Settler habitat preferences were evaluated using a chi-square goodness of fit test (P = 0.05). Five preference experiments, each comparing 2 habitat types, were conducted to explain the observed distribution of E. lori settlers on sponge habitat. Preference experiments included comparisons of yellow sponges versus grey PVC pipes, yellow sponges versus brown sponges, large (>35 cm) versus small (10 cm) yellow sponges, multi-tube versus single-tube yellow sponges, and E. lori resident-occupied versus unoccupied yellow sponges. Experiments were conducted under different sensory cue conditions for each arena in which settlers displayed a significant habitat preference (no manipulation, manipulation of chemical cues, manipulation of visual cues; Figure 4a–c). The yellow sponge versus gray PVC pipe combination was used to demonstrate that E. lori could choose between habitat types. Each pair of habitats, except large versus small yellow sponges, were size-matched by tube length to control for the effect of sponge size on preference behavior. In multi-tube versus single-tube preference tests, 2 single-tube sponges of similar tube length (~35 cm) were placed side-by-side to create repeatable multi-tube sponges, thereby controlling for tube size variation that occurs naturally among multi-tube sponges. Finally, to allow time for resident E. lori to establish themselves and generate chemical cues, residents collected from yellow sponges were relocated to single-tube sponges in the arena 2 h before the start of resident-occupied vs. unoccupied preference tests. Experiment 4: evaluating the effects of prior habitat experience Due to constraints imposed by the study system, settlers used in the experimental components of this study were not naïve with respect to habitat. They were collected on yellow sponges, where they most commonly occur. Therefore, the preceding experiments cannot discriminate between a preference for a particular habitat type and a habitat with which the fish has had prior experience. To specifically address this issue, 1) we tested the habitat preferences of E. lori settlers collected from brown sponges and 2) conducted an observational study to determine the distribution of new settlers arriving from the water column (<10 mm SL with minimal pigmentation; Table 1, Figure 3a). To test the hypothesis that settlers prefer habitat on which they have had prior experience, settlers collected from brown sponges were provided a choice between yellow versus brown sponges in the arena, as described above in Experiment 3) the experimental test of habitat preferences. To observe the distribution of new settlers arriving from the water column, the 120 tagged sponges were cleared of settlers and then surveyed for new settlers every 24–48 h throughout 2 lunar cycles (28 May–25 July 2015). We constructed a generalized linear mixed-effects model (GLMM; distribution = binomial; link = logit) using the “lme4” package in R (Bates et al. 2015) to evaluate how habitat and social variables influence the distribution of new settlers on sponge habitat. The arrival of multiple new settlers on an individual sponge was rare. Therefore, we investigated the relationship between the presence or absence of an E. lori settler (0 or 1, respectively) and all habitat and social variables. Sponge ID was included as a random effect to control for repeated observations of the same 120 tagged sponges. Model selection We constructed statistical models to determine the effects of habitat and social variables on the distribution and persistence of E. lori settlers using a forward stepwise approach in an information theoretic framework. In this framework, a model with an additional variable was retained if the corrected Akaike information criteria (AICc) score was lower than other candidate models by ≥ 2 ∆AICc units. When multiple candidate models were within 2 ∆AICc either 1) the most parsimonious model was selected, or 2) in the case that candidate models contained the same number of variables, the model with the lowest AICc score was chosen. Candidate models within 2 ∆AICc were compared to the reduced model from the previous step using a likelihood-ratio test. If the reduced model and candidate models were not significantly different, then the reduced model was selected as the best-fit model based on parsimony. RESULTS Observational survey of natural settler distribution Logistic regression analyses revealed that sponge species and maximum tube length predicted the presence or absence of settlers on sponge habitat. In the best-fit model, the probability of a settler occurring on a sponge increases with sponge size and is higher on yellow than on brown sponges (Table 2; Figure 5). The results of this analysis indicate that E. lori settlers are more likely to be found on large yellow sponges rather than on small yellow sponges or brown sponges. Table 2 Summary of the best-fit logistic regression model evaluating the association between multiple habitat and social variables and the presence of E. lori settlers on sponge habitat Predictor  Estimate (OR)  SE  z value  P value  Intercept  −3.39 (0.03)  0.711  −4.766  <0.001***  Maximum Tube Length (cm)  0.04 (1.04)  0.016  2.665  0.008**  Sponge Species (yellow)  1.05 (2.88)  0.485  2.177  0.029*  Predictor  Estimate (OR)  SE  z value  P value  Intercept  −3.39 (0.03)  0.711  −4.766  <0.001***  Maximum Tube Length (cm)  0.04 (1.04)  0.016  2.665  0.008**  Sponge Species (yellow)  1.05 (2.88)  0.485  2.177  0.029*  For sponge species, yellow sponges are compared to the reference group brown sponges; (OR), odds ratios. *P < 0.05, **P < 0.001, ***P < 0.001. View Large Figure 5 View largeDownload slide Predictors of the natural distribution of Elacatinus lori settlers on sponge habitat. Lines represent the probability of settler occurrence as a function of maximum tube length and sponge species estimated from the best-fit logistic regression (dashed, yellow sponges; solid, brown sponges). Shaded regions represent the 95% confidence intervals. Figure 5 View largeDownload slide Predictors of the natural distribution of Elacatinus lori settlers on sponge habitat. Lines represent the probability of settler occurrence as a function of maximum tube length and sponge species estimated from the best-fit logistic regression (dashed, yellow sponges; solid, brown sponges). Shaded regions represent the 95% confidence intervals. Experimental test of settler persistence Of 240 settlers that were seeded onto sponges, 44 persisted throughout the entire two-week experimental period. A mixed-effects Cox proportional hazards regression showed that multiple factors predict settler persistence on sponge habitat. The model including sponge species, maximum tube length, resident density, and the starting size of the seeded settlers was the best predictor of settler persistence on sponge habitat ( χ52 = 65.45, P < 0.001). The hazard of disappearance was lower on yellow than on brown sponges and decreased with increasing sponge size (Table 3; Figure 6). In contrast, the hazard of disappearance increased with increasing resident density and was higher for larger (presumably older) settlers (Table 3). Thus, settlers persist longer on large yellow sponges than they do on small yellow sponges or on brown sponges. Further, settlers persist longer on sponges with low densities of residents than they do on sponges with high densities of residents. Table 3 Summary of the best-fit mixed-effects Cox proportional hazards regression evaluating the association between multiple habitat and social variables and E. lori settler persistence on sponge habitat Predictor  Estimate (exp.)  S.E.  χ2  df  P value  Sponge Species (yellow)  −0.815 (0.44)  0.185  38.624  1  <0.001***  Maximum Tube Length (cm)  −0.030 (0.97)  0.007  9.303  1  0.002**  Resident Density  0.809 (2.25)  0.193  14.080  1  <0.001***  Settler Size (mm)  0.125 (1.13)  0.049  6.299  1  0.012*  Predictor  Estimate (exp.)  S.E.  χ2  df  P value  Sponge Species (yellow)  −0.815 (0.44)  0.185  38.624  1  <0.001***  Maximum Tube Length (cm)  −0.030 (0.97)  0.007  9.303  1  0.002**  Resident Density  0.809 (2.25)  0.193  14.080  1  <0.001***  Settler Size (mm)  0.125 (1.13)  0.049  6.299  1  0.012*  For sponge species, yellow sponges are compared to the reference group brown sponges; (exp.), exponentiated coefficient, *P < 0.05, **P < 0.001, ***P < 0.001. View Large Figure 6 View largeDownload slide Kaplan–Meier survival curves for the proportion of settlers remaining on a yellow or brown sponge as a function of time, stratified by sponge size and species, and controlling for resident density and the size of settlers. Lines (Kaplan–Meier survival curves) represent the relationship between proportion of settlers remaining and the independent variables included in the model (see Table 3). Figure 6 View largeDownload slide Kaplan–Meier survival curves for the proportion of settlers remaining on a yellow or brown sponge as a function of time, stratified by sponge size and species, and controlling for resident density and the size of settlers. Lines (Kaplan–Meier survival curves) represent the relationship between proportion of settlers remaining and the independent variables included in the model (see Table 3). Experimental tests of habitat preference Yellow sponges versus PVC pipe We tested the hypothesis that settlers prefer yellow sponges over gray PVC pipe in each sensory cue treatment. Settlers displayed a preference for yellow sponges over gray PVC pipe habitat when sensory cues were unmanipulated (χ2 = 19, df = 1, P < 0.001, n = 19) and when chemical cues were manipulated using clear plastic covers (χ2 = 13, df = 1, P < 0.001, n = 13), but not when visual cues were manipulated using opaque mesh covers (χ2 = 0.53, df = 1, P = 0.47, n = 17). Yellow sponges versus brown sponges We then tested the hypothesis that settlers collected from yellow sponges prefer yellow sponges over brown sponges. Settlers preferred yellow over brown sponges when sensory cues were unmanipulated (χ2 = 13.33, df = 1, P < 0.001, n = 30; Figure 7) and when chemical cues were manipulated using clear plastic covers (χ2 = 4.5, df = 1, P = 0.034, n = 32; Figure 7), but not when visual cues were manipulated using opaque mesh covers (χ2 = 0.2, df = 1, P = 0.65, n = 20; Figure 7). Figure 7 View largeDownload slide Proportion of E. lori settlers collected from yellow tube sponges that chose yellow versus brown tube sponges during trials with no cues manipulated, chemical cues manipulated, or visual cues manipulated. Significant habitat preferences indicated by * (chi-square test; P < 0.05). Figure 7 View largeDownload slide Proportion of E. lori settlers collected from yellow tube sponges that chose yellow versus brown tube sponges during trials with no cues manipulated, chemical cues manipulated, or visual cues manipulated. Significant habitat preferences indicated by * (chi-square test; P < 0.05). Large versus small yellow sponges We tested the hypothesis that settlers would prefer large (>35 cm tube length) over small (10 cm tube length) yellow sponges. Settlers displayed a preference for large over small yellow sponges when sensory cues were unmanipulated (χ2 = 16.2, df= 1, P < 0.001, n = 20; Figure 8) and when chemical cues were manipulated using clear plastic covers (χ2 = 16.2, df= 1, P < 0.001, n = 20; Figure 8), but not when visual cues were manipulated using opaque mesh covers (χ2 = 1.64, df = 1, P = 0.20, n = 22; Figure 8). Figure 8 View largeDownload slide Proportion of settlers collected from yellow tube sponges that chose large vs. small yellow sponges during trials with no cues manipulated, chemical cues manipulated, or visual cues manipulated. Significant habitat preferences indicated by * (chi-square test; P < 0.05). Figure 8 View largeDownload slide Proportion of settlers collected from yellow tube sponges that chose large vs. small yellow sponges during trials with no cues manipulated, chemical cues manipulated, or visual cues manipulated. Significant habitat preferences indicated by * (chi-square test; P < 0.05). Multi versus single-tube yellow sponges We tested the hypothesis that settlers would prefer multi-tube over single-tube yellow sponges in each sensory cue treatment. Settlers displayed no preference for multi vs. single-tube sponges when sensory cues were unmanipulated (χ2 = 0.53, df = 1, P = 0.53, n = 23). Thus, we did not conduct additional experiments manipulating chemical or visual sensory cues. Resident-occupied versus unoccupied yellow sponges Finally, we tested the hypothesis that settlers would prefer yellow sponges already occupied by E. lori residents over unoccupied sponges. Settlers had no preference for resident-occupied vs. unoccupied yellow sponges when cues were unmanipulated (χ2 = 0, df = 1, P = 1, n = 16). Thus, we did not conduct additional experiments manipulating chemical or visual sensory cues. Evaluating the effects of prior habitat experience Despite having been collected from brown sponges, settlers preferred yellow over brown sponges in the arena (χ2 = 18.24, df = 1, P < 0.0001, n = 29; Figure 9a). Considering the distribution of new settlers arriving to sponges, maximum tube length, sponge species, and sponge depth predicted the presence or absence of a new settler (n = 142 new settlers). In the best-fit model, the probability of a new settler occurring on a sponge increases with increasing sponge size (Table 4; Figure 9b), is higher on yellow than on brown sponges (Table 4; Figure 9b), and increases with the depth at which a sponge was located (Table 4). Table 4 Summary of the best-fit mixed-effects logistic regression model evaluating the association between multiple habitat and social variables and the arrival of E. lori settlers on sponge habitat Predictor  Estimate (OR)  SE  z value  P value  (Intercept)  −14.119 (0)  1.386  −10.188  <0.001***  Maximum Tube Length (cm)  0.045 (1.05)  0.007  6.102  <0.001***  Sponge Species (yellow)  1.723 (5.60)  0.307  5.608  <0.001***  Depth  0.477 (1.61)  0.084  5.706  <0.001***  Predictor  Estimate (OR)  SE  z value  P value  (Intercept)  −14.119 (0)  1.386  −10.188  <0.001***  Maximum Tube Length (cm)  0.045 (1.05)  0.007  6.102  <0.001***  Sponge Species (yellow)  1.723 (5.60)  0.307  5.608  <0.001***  Depth  0.477 (1.61)  0.084  5.706  <0.001***  For sponge species, yellow sponges are compared to the reference group brown sponges; (OR), odds ratio; *P < 0.05, **P < 0.001, ***P < 0.001. View Large Figure 9 View largeDownload slide Evaluating the effects of prior habitat experience on preference behaviors of E. lori settlers. (a) Proportion of E. lori settlers collected from brown tube sponges that chose yellow vs. brown sponges during arena trials with no cues manipulated. Significant habitat preferences indicated by * (chi-square test; P < 0.05). (b) Probability of settlement to a sponge as a function of maximum tube length and sponge species, with sponge depth held at its species mean. Lines represent the relationship between the probability of settlement to a sponge and the independent variables estimated from the parameters of the logistic model (dashed, yellow sponges; solid, brown sponges); Shaded regions represent the 95% confidence intervals. Figure 9 View largeDownload slide Evaluating the effects of prior habitat experience on preference behaviors of E. lori settlers. (a) Proportion of E. lori settlers collected from brown tube sponges that chose yellow vs. brown sponges during arena trials with no cues manipulated. Significant habitat preferences indicated by * (chi-square test; P < 0.05). (b) Probability of settlement to a sponge as a function of maximum tube length and sponge species, with sponge depth held at its species mean. Lines represent the relationship between the probability of settlement to a sponge and the independent variables estimated from the parameters of the logistic model (dashed, yellow sponges; solid, brown sponges); Shaded regions represent the 95% confidence intervals. DISCUSSION In organisms with dispersive offspring, distribution patterns may be established by variation among habitats in offspring supply, pre-settlement preference behaviors, and post-settlement mortality (Booth and Wellington 1998; Jenkins 2005; Bohn et al. 2013). Here we demonstrate that the distribution, persistence, and preferences of Elacatinus lori settlers are primarily influenced by characteristics of their settlement habitat: settlers occur more often on (Table 2; Figure 5), persist longer on (Table 3; Figure 6) and prefer large yellow sponges (Table 4; Figures 7–9) rather than small yellow sponges or brown sponges. Taken together, these results suggest that variation in post-settlement persistence selects for habitat preferences that ultimately determine the distribution of E. lori settlers on the reef. Distribution of Elacatinus lori Species-specific microhabitat associations are relatively common in marine systems (Thiel et al. 2003; Baeza 2008), especially in coral reef fishes (Munday et al. 1997; Elliott and Mariscal 2001; Bonin 2011). Microhabitat characteristics such as host species, location, size, and morphology are often linked with their quality as settlement habitat (Connell and Jones 1991; Munday 2001; Harrington et al. 2004). Here, we found that the occurrence of E. lori settlers is predicted by: 1) the sponge species and 2) maximum sponge tube length (Table 2; Figure 5), suggesting that the quality of sponge habitat varies by sponge species and size. These results confirm and extend the conclusions of previous surveys conducted within the South Water Caye Marine Reserve (Carrie Bow Cay, D’Aloia, et al. 2011; Curlew Cay, Majoris J.E., unpublished data). Although these surveys were completed in different locations and years, the variables that predict settler distribution were consistent, suggesting that settler distribution patterns were spatially and temporally stable over this time frame. Stable distributions are often attributed to consistency in settlement behavior, habitat availability, and post-settlement mortality (Holbrook et al. 2000; Jenkins 2005; Bohn et al. 2013). In E. lori, the stable spatiotemporal distribution of settlers suggests that the processes regulating their distribution may also be spatially and temporally stable. Variation in settler persistence Determining the evolutionary consequences associated with alternate settlement habitats is necessary to understand the processes that establish population level distribution patterns (Booth and Wellington 1998; Safran et al. 2007). For E. lori, settler persistence was related to sponge species and size, suggesting that there is predictable variation in settler mortality and/or movement among sponge habitats (Table 3; Figure 6). Similar observations have been made in coral-dwelling gobies, where individuals benefit from increased growth and survival when associating with their preferred species of coral, and at a finer scale, from corals with smaller inter-branch spacing (Munday 2000; Munday 2001). In this system, the outer surface of large yellow sponges are typically highly rugose and may shelter E. lori settlers from predation, compared to the smooth surface of brown sponges (Figure 2), which leaves them more exposed. We also found that resident density had a negative effect on settler persistence (Table 3). In other fishes, antagonistic interactions with resident conspecifics often result in a higher risk of eviction and mortality for settlers (Holbrook and Schmitt 2002; Almany 2003). In the clown anemonefish Amphiprion percula, residents evict (Buston 2003) and may cannibalize (Elliott et al. 1995) incoming settlers when anemone saturation is high. Similarly, E. lori residents evict settlers from occupied sponge tubes, forcing them to the outer sponge wall and, on occasion, have been observed cannibalizing settlers. As sponge tubes become saturated with residents, higher rates of eviction are expected to result in increased mortality or movement of settlers. As settlers were too small to tag in this study, it was not possible to determine the relative contributions of mortality and movement to variation in settler persistence among sponge habitats. However, previous studies have shown that microhabitat specialist reef fishes display strong site fidelity and are unlikely to move among habitats spaced greater than 1–2 m apart (Buston 2003; Feary 2007). In addition, we have observed that E. lori settlers suffer high predation rates when they attempt to leave their sponge (Majoris, personal observation). This observation suggests that mortality plays an important role in determining patterns of settler persistence, while successful movement among sponges may be low. Additional work is necessary to determine the relative influence of mortality and movement on persistence in this species. Habitat preferences As individuals select settlement habitat, they are expected to rely on habitat characteristics or social cues that are positively correlated with habitat quality (Booth and Wellington 1998; Safran et al. 2007). For example, Harrington et al. (2004) found that coral larvae evolved behavioral preferences for the species and growth form of crustose coralline algae that enhance their post-settlement survival. In this study, we found that E. lori settlers prefer (Figures 6 and 7) and persist longer on (Figure 6) large yellow sponges over small yellow sponges or brown sponges. This result suggests that E. lori settlers have evolved a preference for sponge characteristics that are positively correlated with post-settlement persistence. These habitat preferences, in combination with differential post-settlement persistence, can explain their abundance on large yellow sponges. The use of social cues from conspecifics for assessing habitat quality is common across taxa (Sweatman 1983; Stamps 1991; Muller 1998; Fletcher 2007). Despite this, E. lori settlers did not discriminate between resident occupied and unoccupied sponges, suggesting that they did not use social cues which could have improved their post-settlement persistence (Table 3). There are several potential explanations for this mismatch: 1) settlers may not be able to assess resident density—residents are, after all, hidden within sponge tubes, 2) movement of residents between sponges could weaken selection for avoidance behaviors, and 3) the benefits of settling on large yellow sponges may outweigh the costs of settling on occupied sponges. Our results suggest that E. lori use habitat characteristics, rather than social cues, to choose settlement habitat. The role of sensory cues in habitat preference Though individuals may use visual, chemical, and/or auditory cues at settlement, this study showed that E. lori settlers rely on visual cues when selecting sponge habitat over small spatial scales (Figures 7 and 8). This was an unexpected result, as sponges are known to emit a range of chemicals that were expected to provide settlers with cues for habitat selection (Pawlik 2011), and previous research has demonstrated the importance of chemosensory cues during habitat selection by some marine species (Krug and Manzi 1999; Forward et al. 2001; Lecchini and Nakamura 2013). However, previous studies have shown that other reef fishes also rely on visual, rather than chemical, cues for habitat recognition (Lecchini et al. 2005a; Igulu et al. 2011; Lecchini et al. 2014). The large size and bright coloration of yellow sponges make them conspicuous on the reef. These characteristics may allow settling E. lori to quickly locate sponge habitat using visual cues. Effects of prior habitat experience A limitation of this study was the inability to collect naïve larvae for use in habitat preference experiments, or to collect sufficient quantities of settlers from brown sponges to implement a fully cross-factored experimental design. Given this limitation, all settlers were collected from and had prior experience on yellow sponges. Sale (1971) found that, when given a choice between two coral species as settlement habitat, recently-settled Dascyllus aruanus preferred the species of coral from which they were collected. In contrast, Danilowicz (1996) demonstrated that both naïve Dascyllus albisella settlers and wild-caught settlers collected from several species of coral preferred a single coral species. Given these mixed results for other species, we considered the potential influence of prior experience on the results of this study. We found that settlers collected from brown sponges preferred yellow sponges over the habitat on which they had prior experience (Figure 9a), indicating that prior experience did not influence their preference behaviors. Further, by observing the distribution of new settlers arriving on both sponge species, we found that new E. lori settlers were more likely to occur on large yellow sponges than small yellow sponges or brown sponges (Table 4; Figure 9b). Since persistence was similar on yellow and brown sponges within the first 24–48 h (Figure 6), these results suggest that E. lori larvae chose large yellow sponges at settlement. Settlers also prefer (Figures 7 and 8) and ultimately persist longer on (Table 4; Figure 6) large yellow sponges under experimental conditions. Taken together, these results support the hypothesis that E. lori have evolved behavioral preferences for sponge habitats that will maximize their post-settlement persistence, and that decisions at settlement shape the population level pattern of settler distribution on coral reefs. Evolution of habitat preferences In light of these results, it is interesting to consider under what conditions habitat preferences evolve. For preference behaviors to evolve, 1) sensory capabilities and habitat cues must allow individuals to reliably discriminate among habitat types, 2) the habitat cues must be strongly correlated with post-settlement mortality, and 3) individuals must be able to sample and choose freely among habitat types (Fretwell and Lucas 1969). In this study, E. lori settlers prefer habitat characteristics that are strongly associated with their post-settlement persistence, irrespective of their prior experience. While we cannot quantify the relative contributions of post-settlement mortality and movement in establishing persistence, we predict that variation in mortality among sponges has selected for behavioral preferences for the habitat on which settlers are most likely to persist (i.e. large yellow sponges). Given these preferences, individuals that move post-settlement will likely migrate from less preferred sponges toward large yellow sponges when possible, reinforcing the distribution pattern established by habitat preferences at settlement. However, there are many species in which variation in mortality across settlement habitats is not associated with pre-settlement preference behaviors (Amphiprion percula: Buston 2003, 2004; Stegastes leucostictus: Almany 2003). In these species, the presence of reliable sensory cues for discriminating between habitat types and assumption of free choice between settlement habitats may be violated (Levin et al. 2000). Therefore, individuals would benefit from settling on the first habitat that they encounter rather than continuing to search for an optimal habitat (Buston 2004; Stamps et al. 2005). CONCLUSION Habitat preferences and post-settlement processes are often invoked to explain the distribution and abundance of organisms throughout their range. Here, we demonstrate that settling E. lori select sponge habitat using characteristics that are associated with their potential for post-settlement persistence, and that decisions at settlement shape the population level pattern of distribution on coral reefs. Although this study focuses on a coral reef fish, pre- and post-settlement processes can play an important role in establishing the distribution patterns of diverse taxa. FUNDING This work was funded in part by a startup award from the Trustees of Boston University and National Science Foundation grants (grant numbers OCE-1260424, OCE-1459546) to PMB. Additional funding was provided by a Lerner-Gray Grant awarded by the American Museum of Natural History, a Warren McLeod Summer Research Scholarship awarded by Boston University and a Doctoral Dissertation Improvement Grant (grant number IOS-1501651) awarded by the National Science Foundation to JEM. Data accessibility: Analyses reported in this article can be reproduced using the data provided by Majoris et al. (2017). We would like to thank the Belizean government and Fisheries Department for permission to conduct this research. Thank you to the staff at Wee Wee Caye Marine Station and the International Zoological Expeditions field station for the use of their facilities. Special thanks to Kevin David, Earl David, Udel Forman, Romain Chaput, Emma Shlatter, James Garner, and Alissa Rickborn for assistance in the field, and Kate Langwig for assistance with statistical analyses. Jelle Atema, Karen Warkentin, Jacqueline Webb, John Crawford, Kristina Cohen and two anonymous reviewers provided helpful comments on this manuscript. This study represents a portion of the dissertation work of JEM. All work was approved by the Belize Fisheries Department and the Boston University IACUC (protocol # 10–036). REFERENCES Adam TC. 2011. High-quality habitat and facilitation ameliorate competitive effects of prior residents on new settlers. Oecologia . 166: 121– 130. Google Scholar CrossRef Search ADS PubMed  Almany GR. 2003. Priority effects in coral reef fish communities. Ecology . 84: 1920– 1935. Google Scholar CrossRef Search ADS   Almany GR. 2004a. Differential effects of habitat complexity, predators and competitors on abundance of juvenile and adult coral reef fishes. Oecologia . 141: 105– 113. Google Scholar CrossRef Search ADS   Almany GR. 2004b. Does increased habitat complexity reduce predation and competition in coral reef fish assemblages? Oikos . 106: 275– 284. Google Scholar CrossRef Search ADS   Atema J, Kingsford MJ, Gerlach G. 2002. Larval reef fish could use odour for detection, retention and orientation to reefs. Mar Ecol Prog Ser . 241: 151– 160. Google Scholar CrossRef Search ADS   Baeza JA. 2008. Social monogamy in the shrimp Pontonia margarita, a symbiont of Pinctada mazatlanica, off the Pacific coast of Panama. Mar Biol . 153: 387– 395. Google Scholar CrossRef Search ADS   Bates D, Maechler M, Bolker B, Walker S. 2015. Fitting Linear Mixed-Effects Models Using lme4. J Stat Softw . 67: 1– 48. Google Scholar CrossRef Search ADS   Ben-Tzvi O, Kiflawi M, Polak O, Abelson A. 2009. The effect of adult aggression on habitat selection by settlers of two coral-dwelling damselfishes. PLoS One . 4: e5511. Google Scholar CrossRef Search ADS PubMed  Bohn K, Richardson CA, Jenkins SR. 2013. The importance of larval supply, larval habitat selection and post-settlement mortality in determining intertidal adult abundance of the invasive gastropod Crepidula fornicata. J Exp Mar Biol Ecol . 440: 132– 140. Google Scholar CrossRef Search ADS   Bonin MC. 2011. Specializing on vulnerable habitat: Acropora selectivity among damselfish recruits and the risk of bleaching-induced habitat loss. Coral Reefs . 31: 287– 297. Google Scholar CrossRef Search ADS   Booth DJ, Hixon MA. 1999. Food ration and condition affect early survival of the coral reef damselfish, Stegastes partitus. Oecologia . 121: 364– 368. Google Scholar CrossRef Search ADS PubMed  Booth DJ, Wellington G. 1998. Settlement preferences in coral-reef fishes: Effects on patterns of adult and juvenile distributions, individual fitness and population structure. Aust J Ecol . 23: 274– 279. Google Scholar CrossRef Search ADS   Buston PM. 2003. Forcible eviction and prevention of recruitment in the clown anemonefish. Behav Ecol . 14: 576– 582. Google Scholar CrossRef Search ADS   Buston PM. 2004. Territory inheritance in clownfish. Proc Biol Sci . 271( suppl 4): S252– S254. Google Scholar CrossRef Search ADS PubMed  Colin PL. 2002. A new species of sponge-dwelling Elacatinus (Pisces: Gobiidae) from the western Caribbean. Zootaxa . 106: 1– 7. Google Scholar CrossRef Search ADS   Connell SD, Jones GP. 1991. The influence of habitat complexity on postrecruitment processes in a temperate reef fish population. J Exp Mar Biol Ecol . 151: 271– 294. Google Scholar CrossRef Search ADS   D’Aloia CC, Bogdanowicz SM, Francis RK, Majoris JE, Harrison RG, Buston PM. 2015. Patterns, causes, and consequences of marine larval dispersal. Proc Natl Acad Sci U S A . 112: 13940– 13945. Google Scholar CrossRef Search ADS PubMed  D’Aloia CC, Majoris JE, Buston PM. 2011. Predictors of the distribution and abundance of a tube sponge and its resident goby. Coral Reefs . 30: 777– 786. Google Scholar CrossRef Search ADS   Danilowicz BS. 1996. Choice of Coral Species by Naive and Field-Caught Damselfish. Copeia . 1996: 735– 739. Google Scholar CrossRef Search ADS   Dixson D, Pratchett M, Munday P. 2012. Reef fishes innately distinguish predators based on olfactory cues associated with recent prey items rather than individual species. Anim Behav . 84: 45– 51. Google Scholar CrossRef Search ADS   Dixson DL, Jones GP, Munday PL, Planes S, Pratchett MS, Srinivasan M, Syms C, Thorrold SR. 2008. Coral reef fish smell leaves to find island homes. Proc Biol Sci . 275: 2831– 2839. Google Scholar CrossRef Search ADS PubMed  Elliott JK, Elliott JM, Mariscal RN. 1995. Host selection, location, and association behaviors of anemonefishes in field settlement experiments. Mar Biol . 122: 377– 389. Google Scholar CrossRef Search ADS   Elliott JK, Mariscal RN. 2001. Coexistence of nine anemonefish species: differential host and habitat utilization, size and recruitment. Mar Biol . 138: 23– 36. Google Scholar CrossRef Search ADS   Feary DA. 2007. The influence of resource specialization on the response of reef fish to coral disturbance. Mar Biol . 153: 153– 161. Google Scholar CrossRef Search ADS   Feary DA, Almany GR, McCormick MI, Jones GP. 2007. Habitat choice, recruitment and the response of coral reef fishes to coral degradation. Oecologia . 153: 727– 737. Google Scholar CrossRef Search ADS PubMed  Fletcher RJJr. 2007. Species interactions and population density mediate the use of social cues for habitat selection. J Anim Ecol . 76: 598– 606. Google Scholar CrossRef Search ADS PubMed  Forrester GE. 1995. Strong density-dependent survival and recruitment regulate the abundance of a coral reef fish. Oecologia . 103: 275– 282. Google Scholar CrossRef Search ADS PubMed  Forward RB, Tankersley RA, Rittschof D. 2001. Cues for metamorphosis of Brachyuran crabs: an overview. Am Zool . 41: 1108– 1122. Fretwell SD, Lucas H. 1969. On territorial behavior and other factors influencing habitat distribution in birds. Acta Biotheor . 19: 16– 36. Google Scholar CrossRef Search ADS   Gerlach G, Atema J, Kingsford MJ, Black KP, Miller-Sims V. 2007. Smelling home can prevent dispersal of reef fish larvae. Proc Natl Acad Sci U S A . 104: 858– 863. Google Scholar CrossRef Search ADS PubMed  Gutiérrez L. 1998. Habitat selection by recruits establishes local patterns of adult distribution in two species of damselfishes: Stegastes dorsopunicans and S. planifrons. Oecologia . 115: 268– 277. Google Scholar CrossRef Search ADS PubMed  Harrington L, Fabricius K, De’ath G, Negri A. 2004. Recognition and Selection of Settlement Substrata Determine Post-Settlement Survival in Corals. Ecology . 85: 3428– 3437. Google Scholar CrossRef Search ADS   Holbrook SJ, Forrester GE, Schmitt RJ. 2000. Spatial patterns in abundance of a damselfish reflect availability of suitable habitat. Oecologia . 122: 109– 120. Google Scholar CrossRef Search ADS PubMed  Holbrook SJ, Schmitt RJ. 2002. Competition for Shelter Space Causes Density-Dependent Predation Mortality in Damselfishes. Ecology . 83: 2855. Google Scholar CrossRef Search ADS   Holbrook SJ, Schmitt RJ. 2003. Spatial and temporal variation in mortality of newly settled damselfish: patterns, causes and co-variation with settlement. Oecologia . 135: 532– 541. Google Scholar CrossRef Search ADS PubMed  Huijbers CM, Nagelkerken I, Lössbroek PA, Schulten IE, Siegenthaler A, Holderied MW, Simpson SD. 2012. A test of the senses: fish select novel habitats by responding to multiple cues. Ecology . 93: 46– 55. Google Scholar CrossRef Search ADS PubMed  Igulu MM, Nagelkerken I, Fraaije R, van Hintum R, Ligtenberg H, Mgaya YD. 2011. The potential role of visual cues for microhabitat selection during the early life phase of a coral reef fish (Lutjanus fulviflamma). J Exp Mar Biol Ecol . 401: 118– 125. Google Scholar CrossRef Search ADS   Jenkins SR. 2005. Larval habitat selection, not larval supply, determines settlement patterns and adult distribution in two chthamalid barnacles. J Anim Ecol . 74: 893– 904. Google Scholar CrossRef Search ADS   Jones GP. 1988. Some interactions between residents and recruits in two coral reef fishes. J Exp Mar Biol Ecol . 114: 169– 182. Google Scholar CrossRef Search ADS   Jones GP. 1997. Relationships between recruitment and postrecruitment processes in lagoonal populations of two coral reef fishes. J Exp Mar Biol Ecol . 213: 231– 246. Google Scholar CrossRef Search ADS   Krug PJ, Manzi AE. 1999. Waterborne and Surface-Associated Carbohydrates as Settlement Cues for Larvae of the Specialist Marine Herbivore Alderia modesta. Biol Bull . 197: 94– 103. Google Scholar CrossRef Search ADS PubMed  Lecchini D, Carassou L, Frederich B, Nakamura Y, Mills SC, Galzin R. 2012. Effects of alternate reef states on coral reef fish habitat associations. Environ Biol Fishes . 94: 421– 429. Google Scholar CrossRef Search ADS   Lecchini D, Nakamura Y. 2013. Use of chemical cues by coral reef animal larvae for habitat selection. Aquat Biol . 19: 231– 238. Google Scholar CrossRef Search ADS   Lecchini D, Peyrusse K, Lanyon RG, Lecellier G. 2014. Importance of visual cues of conspecifics and predators during the habitat selection of coral reef fish larvae. C R Biol . 337: 345– 351. Google Scholar CrossRef Search ADS PubMed  Lecchini D, Planes S, Galzin R. 2005a. Experimental assessment of sensory modalities of coral-reef fish larvae in the recognition of their settlement habitat. Behav Ecol Sociobiol . 58: 18– 26. Google Scholar CrossRef Search ADS   Lecchini D, Shima J, Banaigs B, Galzin R. 2005b. Larval sensory abilities and mechanisms of habitat selection of a coral reef fish during settlement. Oecologia . 143: 326– 334. Google Scholar CrossRef Search ADS   Leis JM, Siebeck U, Dixson DL. 2011. How Nemo finds home: the neuroecology of dispersal and of population connectivity in larvae of marine fishes. Integr Comp Biol . 51: 826– 843. Google Scholar CrossRef Search ADS PubMed  Levin PS, Tolimieri N, Nicklin M, Sale PF. 2000. Integrating individual behavior and population ecology: the potential for habitat-dependent population regulation in a reef fish. Behav Ecol . 11: 565– 571. Google Scholar CrossRef Search ADS   Majoris JE, D’Aloia CC, Francis RK, Buston PM. 2017. Data from: differential persistence favors habitat preferences that determine the distribution of a reef fish. Dryad Digital Repository . http://dx.doi.org/10.5061/dryad.vn712. Mills M. 2011. Introducing Survival and Event History Analysis . London: SAGE Publications. Google Scholar CrossRef Search ADS   Muller KL. 1998. The role of conspecifics in habitat settlement in a territorial grasshopper. Anim Behav . 56: 479– 485. Google Scholar CrossRef Search ADS PubMed  Munday PL. 2001. Fitness consequences of habitat use and competition among coral-dwelling fishes. Oecologia . 128: 585– 593. Google Scholar CrossRef Search ADS PubMed  Munday PL. 2000. Interactions Between Habitat Use and Patterns of Abundance in Coral-dwelling Fishes of the Genus Gobiodon. Environ Biol Fishes . 58: 355– 369. Google Scholar CrossRef Search ADS   Munday PL, Jones GP, Caley MJ. 1997. Habitat specialisation and the distribution and abundance of coral-dwelling gobies. Mar Ecol Prog Ser . 152: 227– 239. Google Scholar CrossRef Search ADS   Pawlik JR. 2011. The Chemical Ecology of Sponges on Caribbean Reefs: Natural Products Shape Natural Systems. BioScience . 61: 888– 898. Google Scholar CrossRef Search ADS   R Core Team. 2015. R: A language and environment for statistical computing . Vienna (Austria): R Foundation for Statistical Computing. Rickborn AJ, Buston PM. 2015. Life-history transitions of the coral-reef fish Elacatinus lori. J Fish Biol . 86: 637– 650. Google Scholar CrossRef Search ADS   Rützler K, Macintyre IG. 1982. The habitat distribution and community structure of the barrier reef complex at Carrie Bow Cay, Belize. In: The Atlantic barrier reef ecosystem at Carrie Bow Cay, Belize, I. structure and communities . Washington, DC: Smithsonian Institution Press. (Smithsonian contributions to the marine sciences). p. 9– 45. Safran RJ, Doerr VA, Sherman PW, Doerr ED, Flaxman SM, Winkler DW. 2007. Group breeding in vertebrates: linking individual-and population-level approaches. Evol Ecol Res . 9: 1163. Sale PF. 1971. Apparent effect of prior experience on a habitat preference exhibited by the reef fish, Dascyllus aruanus (Pisces: Pomacentridae). Anim. Behav . 19: 251– 256. Google Scholar CrossRef Search ADS   Schmitt R, Holbrook S. 1996. Local-scale Patterns of Larval Settlement in a Planktivorous Damselfish—Do They Predict Recruitment? Mar. Freshw Res . 47: 449– 463. Google Scholar CrossRef Search ADS   Schmitt RJ, Holbrook SJ. 1999. Settlement and recruitment of three damselfish species: larval delivery and competition for shelter space. Oecologia . 118: 76– 86. Google Scholar CrossRef Search ADS PubMed  Shima JS. 2001. Recruitment of a coral reef fish: roles of settlement, habitat, and postsettlement losses. Ecology . 82: 2190– 2199. Google Scholar CrossRef Search ADS   Simpson SD, Meekan MG, McCauley RD, Jeffs A. 2004. Attraction of settlement-stage coral reef fishes to reef noise. Mar Ecol Prog Ser . 276: 263– 268. Google Scholar CrossRef Search ADS   Srinivasan M. 2003. Depth distributions of coral reef fishes: the influence of microhabitat structure, settlement, and post-settlement processes. Oecologia . 137: 76– 84. Google Scholar CrossRef Search ADS PubMed  Stamps JA. 1988. Conspecific Attraction and Aggregation in Territorial Species. Am Nat . 131: 329– 347. Google Scholar CrossRef Search ADS   Stamps JA. 1991. The effect of conspecifics on habitat selection in territorial species. Behav. Ecol Sociobiol . 28: 29– 36. Google Scholar CrossRef Search ADS   Stamps JA, Krishnan VV, Reid ML. 2005. Search costs and habitat selection by dispersers. Ecology . 86: 510– 518. Google Scholar CrossRef Search ADS   Sweatman HP. 1985. The influence of adults of some coral reef fishes on larval recruitment. Ecol Monogr . 55: 469– 485. Google Scholar CrossRef Search ADS   Sweatman HPA. 1983. Influence of conspecifics on choice of settlement sites by larvae of two pomacentrid fishes (Dascyllus aruanus and D. reticulatus) on coral reefs. Mar Biol . 75: 225– 229. Google Scholar CrossRef Search ADS   Therneau T. 2015. coxme: Mixed Effects Cox Models . (R package version 2.2–5). https://cran.r-project.org/package=coxme Thiel M, Zander A, Valdivia N, Baeza JA, Rueffler C. 2003. Host fidelity of a symbiotic porcellanid crab: the importance of host characteristics. J Zool . 261: 353– 362. Google Scholar CrossRef Search ADS   Wellington GM. 1992. Habitat selection and juvenile persistence control the distribution of two closely related Caribbean damselfishes. Oecologia . 90: 500– 508. Google Scholar CrossRef Search ADS PubMed  © The Author(s) 2018. Published by Oxford University Press on behalf of the International Society for Behavioral Ecology. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com

Journal

Behavioral EcologyOxford University Press

Published: Mar 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off