Refuge choice specificity increases with predation risk in a rocky reef fish

Refuge choice specificity increases with predation risk in a rocky reef fish

Journal of Experimental Marine Biology and Ecology 520 (2019) 151207 Contents lists available at ScienceDirect Journal of Experimental Marine Biolog...

681KB Sizes 0 Downloads 17 Views

Journal of Experimental Marine Biology and Ecology 520 (2019) 151207

Contents lists available at ScienceDirect

Journal of Experimental Marine Biology and Ecology journal homepage: www.elsevier.com/locate/jembe

Refuge choice specificity increases with predation risk in a rocky reef fish a,b,⁎

c,d

a,e

T

f

José Anchieta C.C. Nunes , Antoine Leduc , Ricardo J. Miranda , Pedro H. Cipresso , João P. Alvesa, Eduardo Mariano-Netoc, Cláudio L.S. Sampaiog, Francisco Barrosa a

Laboratório de Ecologia Bentônica, CIENAM, Instituto de Biologia, Universidade Federal da Bahia, Campus Ondina, Salvador, BA CEP 40170-115, Brazil Laboratório de Ecologia e Conservação Marinha, Centro de Formação de Ciências Ambientais, Universidade Federal do Sul da Bahia, Brazil Instituto de Biologia, Universidade Federal da Bahia, Campus Ondina, Salvador, BA CEP 40170-115, Brazil d Marine Ecology Laboratory, Department of Oceanography and Limnology, Universidade Federal do Rio Grande do Norte, Av. Via Costeira S/N, Natal, RN CEP 59014002, Brazil e National Marine Science Centre, Southern Cross University, Coffs Harbour, NSW 2450, Australia f Projeto Conservação Recifal, Recife, Brazil g Laboratório de Ictiologia e Conservação, U. E. Penedo, Campus Arapiraca, Programa de Pós-graduação em Diversidade Biológica e Conservação nos Trópicos, ICBS, Universidade Federal de Alagoas (UFAL), Maceió, Alagoas, Brazil b c

A R T I C LE I N FO

A B S T R A C T

Keywords: Flight initiation distance Reef fish behaviour Escape microhabitat, ecology of fear Antipredation strategies

Reef ecosystems are structurally complex and characterized by an array of abiotic (e.g., rocks and crevices) and biotic (sessile benthic organisms) physical features, many of which having the potential to act as refuge for prey organisms. Small cryptic reef fish species, including the redlip blenny, Ophioblennius trinitatis, rely on refuges to survive against predators, suggesting that adequate refuge choice is an important part of this species' antipredator strategy. Here we investigated blennies' selectivity in refuge choice along its ontogeny simulating predation risk in the field and laboratory. Our results revealed that when exposures to a predator model in the field, blennies chiefly fled to only two refuge types, namely crevices and sea urchins, and these choices reflected blennies' ontogeny. Furthermore, blennies' densities and flight-initiation distances were positively and negatively correlated with sea urchin densities, respectively, underscoring the refuge role played by this benthic organism. On high risk conditions, the fish translated into faster retreat to shelter and to higher refuge selectivity, when compared to low risk. The choices observed in situ were partly similar among initial and terminal phases. These findings suggest that small reef fishes may be selective in their refuge choice even using habitats with high structural complexity. This choice selectivity may be part of an antipredator strategy that considers the relative level of protection offered by the physical features of habitats, along with their conspicuity within rocky reef ecosystems.

1. Introduction Ecologists have long recognized the importance of habitat selection as a mechanism underpinning both species occurrence in a given habitat and their propensity to coexist in communities (MacArthur and Pianka, 1966; Rosenzweig, 1981; Pereira et al., 2015). In predator-prey interactions, refuge selection may be thought as a form of habitat choice, which allow preys to persist in the same habitat as their predators (Stankwich and Blumstein et al., 2005). Given refuges may drastically improve a prey's survival, important pressures on prey should exist to choose adequately which physical structures are used as refuge (Hixon and Beets, 1993). Considering many mobile preys living in structurally complex

habitats (e.g., rainforests, coral and rocky reefs), multiple types of refuges may potentially exist. Examples in terrestrial systems include tree holes, branches and bushes, and in the oceans, burrows, rocks, crevices and benthic organisms. While this is common knowledge, we have scant information on the mechanisms by which preys select among multiples available refuges, and for preys that do not build their own refuge, whether these are selective in the type of refuge they use under risk. Recently, Cooper and Samia (2018) proposed a refuge selection model for typical mobile preys. This theoretical model provides a mechanistic understanding on how prey may select against multiple available refuges (i.e. considering relative distance of predator and refuges, level of risk imposed on prey to attain a given refuge), it does not address, per se, whether qualitatively different types (i.e., different

⁎ Corresponding author at: Laboratório de Ecologia Bentônica, CIENAM, Instituto de Biologia, Universidade Federal da Bahia, Campus Ondina, Salvador, BA CEP 40170-115, Brazil. E-mail address: [email protected] (J.A.C.C. Nunes).

https://doi.org/10.1016/j.jembe.2019.151207 Received 2 January 2019; Received in revised form 11 July 2019; Accepted 10 August 2019 0022-0981/ © 2019 Elsevier B.V. All rights reserved.

Journal of Experimental Marine Biology and Ecology 520 (2019) 151207

J.A.C.C. Nunes, et al.

predominantly covered by coralline algae, sea urchins (Echinometra lucunter), and corals (Favia gravida, Millepora spp. and Siderastraea spp.) (Ferreira et al., 2015).

physical structures) are discriminated to provide protection against predators. In essence, the remaining questions is: do specific physical structures present in a prey's habitat are more likely to be chosen as refuges than others? Adding to this deficiency, studies on refuge choice are rarely conducted in situ, such that actual species-specific preferences for refuge are poorly known. Tropical rocky reefs are structurally complex systems, and thereby may support speciose communities composed of vagile and sessile organisms (Ferreira et al., 2001; Ferreira et al., 2015). In fact, rocky reefs are structurally complex not only because of the presence of abiotic structural components (e.g., rocks and crevices), as well as it harbors an array of benthic species (i.e., invertebrates and algae) that further add rugosity to the basic abiotic structure (Carvalho and Barros, 2017). Given their complex habitat structure (Carvalho and Barros, 2017) and the rich fauna they harbour (Ferreira et al., 2015), rocky reef ecosystems may be considered a good model to test refuge selection for typical mobile prey whose relative proximity to multiple physical structures may require making choices when seeking a refuge. The mechanisms by which prey species may choose between different refuge availability are general not well known. Indeed, habitat structure is an important factor influencing reef fish assemblages at various spatial scales (MacNeil et al., 2009; Vergés et al., 2011) and can help fishes to avoid predation (Almany, 2004). Choosing good study models to investigate choice of shelter from predation and associated with field experimentation and controlled situations (laboratory) using the same models may increase our understanding of shelter choices. A good candidate to enter select escape micro-habitat is the relationship between some reef fishes (e.g. blennies) and sea urchins. Cryptobenthic fishes (including gobies and blennies) use sea urchins as refuge from predation (e.g. Patzner, 1999; Patzner and Santos, 1992; Giglio et al., 2018). However, most studies have look at the correlations between fish and urchins density and no studies investigated fish escape behaviour and flight metrics. The latter approaches can bring important insights on prey refuge ecology and may be crucial to separate the use from simple habitat coincidence. While ontogenetic shifts in microhabitat preference of the temperate reef fish was well studied (McDermott and Shima, 2006; Pérez-Matus and Shima, 2010), nothing is known regarding cryptic reef fish refuge choices in predation context on tropical rocky reefs. Here, we investigated whether a cryptobenthic prey fish, the Brazilian redlip blenny (Ophioblennius trinitatis) is selective or indiscriminate when seeking refuge under the threat of predation. We hypothesized that refuge use is an antipredator strategy of O. trinitatis, actively selecting elements of habitat which may increase survival. We also predicted that higher predation risk should lead to increase in selectivity of choice. Specifically, we tested in situ i) which of rocky reefs' physical structures blennies will flee to when exposed to a model predator (i.e., overall refuge selectivity). Under laboratory conditions, ii) whether blenny specificity in refuge choice varies as a function of the perceived risk level. For this latter goal, we choose two types (i.e., sea urchins and rocks) that blennies were shown to frequently use under natural conditions.

2.1.2. Model species The redlip blenny, O. trinitatis, is an abundant endemic fish of the coast of Brazil occuring mainly in intertidal zones, often associated with sea urchins (Dalben and Floeter, 2012; Mendes, 2007). We considered both initial and terminal phases of O. trinitatis because the role (i.e., diet) of a species in a community may change along its ontogeny and hence each onthogenic phase may have distinct requirements (Jones, 2002). These blennies' ontogenetic phases are characterized by distinct coloration patterns (described in Rangel and Mendes, 2009), behavior and diet (Humann and Deloach, 2002). These fish are herbivorous that form territories and have intra-agonistic behavior and interspecific in defense against intruders (Nursall, 1977). The sea urchin Echinometra lucunter lives in crevices and are commonly found in tropical rock shores and a considerable portion of their diet consists of algae (Ogden and Lobel, 1978; Cordeiro et al., 2014). 2.1.3. Field trials To determine the type of refuges used by O. trinitatis when faced by a predator, we used a 30 cm fiberglass grouper (Epinephelus genus) predator model (SM1). We chose this predator model given at this size, this taxonomic group has the ability to effectively prey on O. trinitatis. This predator model was attached to a negatively buoyant meter-long stainless-steel rod enabling its manipulation by a trained observer (JACCN). Following its detection by the observer, each target prey individuals was approached by swimming in its direction horizontally, near the substrate, until reaching the minimal distance of 1.5 m from the target fish, upon which the observer stretched his arm to present the model to target fish at a rate of approximately 1 m s−1. After this approach, the target fish invariably fled. The location of escape, namely, the type of physical feature used was recorded. For example, refuge type (i.e., nearest structure) may have consisted of sea urchins, crevices, corals, algae, sponges or zoanthids. These aforementioned biotic and abiotic structures were shown to be used a potential refuge by O. trinitatis under a predator-prey context (Mendes, 2006). We wish to determine whether O. trinitatis used these structures as refuge from predator (i.e., not for other aim) following escape. Thus, we estimated the availability of these potential refugia using a 25 × 25 cm quadrat with 20 intercept points. The organism or abiotic structure (crevice in rocks) located on each intercept point was recorded. We also tested the potential ‘refuge’ role of physical features by measuring O. trinitatis' flight initiation distance (FID) in a gradient of sea urchin density. Thus, following each escape by O. trinitatis, the predator model was laid on the bottom and the distance between the model's snout and the last observed location of the target fish (i.e., from where it started to fled) was measured using a scale ruler. This measure was reliably used to infer the relative level of risk a prey may perceive in a given condition. In total, we conducted this procedure with 100 individuals of O. trinitatis. To avoid recording the same individual repeatedly, we moved by at least 3 m at the end of each observation before looking for the next individual fish. In order to estimate population density of O. trinitatis and Echinometria lucunter, we conducted stationary visual censuses of 2 m radius during five minutes (Minte-Vera et al., 2008). These censuses (n = 60) were haphazardly distributed along the studied rocky reef. The length of each fish was also estimated and recorded. Flight initiation distance sampling and stationary visual censuses were done between 10:00 am and 14:00 pm.

2. Material and methods 2.1. Experiment 1: field assessment of refuge selectivity 2.1.1. Study area This study was conducted between October and December of 2015 at Barra rocky reefs (13′ 23o S, 38′ 55o W). These reefs are located at the Eastern entrance of Todos os Santos Bay, adjacent to the Salvador city, State of Bahia, Brazil. The study area has a maximum depth of 6 m, enabling in situ observations while snorkeling. The reef benthic community is composed mainly by filamentous algae, macroalgae, ascidians, zoanthids (Palythoa caribaeorum and Zoanthus spp). However, in shallow areas where O. trinitatis abounds, large areas (i.e., > 5 m) were

2.2. Experiment 2: laboratory refuge choice test 2.2.1. Experimental setting We investigated the specificity of refuge choice upon exposure to a 2

Journal of Experimental Marine Biology and Ecology 520 (2019) 151207

J.A.C.C. Nunes, et al.

3. Statistical analysis

predation threat. This was done using six 40-L aquariums, but that were filled with only 20 L of salt water, (made from a commercially-purchased Red Sea ™ aquarium salt) (SM2). We used this reduced water volume given the benthic nature of test fish permitted maintaining an adequate level of realism (i.e., water depth was sufficient to allow swimming in three dimensions), while minimizing the volume of water needed be replaced between trials. Tests were conducted once daily between 10:00 am and 14:00 pm. Using a removable glass barrier, each aquarium was divided transversally into two compartments. On one side of each aquarium, a single O. trinitatis was placed. The other aquarium side was divided longitudinally by a glass divider, which contained a sea urchin E. lucunter and a rock (proportional similar size), separated from each other by a glass barrier. The side (left or right) containing the urchin or the rock was randomized for each trial by flipping a coin. O. trinitatis and E. lucunter were acclimated for 24 h before conducting a trial. Water quality was controlled such that the mean value ( ± standard error, SE) for salinity (mean specific gravity) was 1.26 g NaCL−1 ( ± 1.1), pH was 8.3 ( ± 0.1), ammonium, nitrite and nitrate were 0.01 ppm ( ± 0.01), 0.2 ppm ( ± 0.2) and 0.03 ppm ( ± 0.02), respectively. Laboratory temperature and photoperiod were set a 25 °C and 12 L: 12D (respectively), and did not change during all period of the experiment.

3.1. Experiment 1: field assessment of refuge choice and FID An electivity index (Ivlev's Electivity Index) was used to identify if O. trinitatis preferred a certain physical feature of the reef substrate when fleeing from the predator model. This selectivity was calculated using the formula: Ei = (ri - ni)/(ri + ni), where Ei is the value of electivity for the type of substrate i; ri is the percentage of flight to the substrate i and ni is the percentage of substrate i in the studied location. Ivlev's Electivity Index varies from −1 (i.e., low preference or rejection) to 1 (i.e., high preference) for a particular substrate (Krebs, 1989). To generate a 95% confidence interval around the observed Ei, we used non-parametric bootstrapping procedures (10,000 randomizations). Generalized Linear Models (GLM) with Gaussian distribution were used to investigate if E. lucunter density and number of rock crevices (continuous variables) and O. trinitatis life phase (two levels: initial and terminal phase) influenced FID in O. trinitatis (dependent variable). Posterior deviance analysis was performed to evaluate the best minimal model and significance of each term. Residual analysis was performed to evaluate the quality of final model (Crawley, 2007). All analyses were done in R (R Development Core Team, R, 2016).

2.2.2. Experimental trials In this experiment, fish were of wild origin, but obtained from a certified commercial supplier of marine ornamental organisms for the aquarium trade. In total, 40 replicates were performed (i.e., 20 for each of the initial and terminal phase individuals). As in the field experiment, we aimed to investigate the refuge selectivity following exposure to predation risk. To simulate predation risk in the lab, (i.e., a predatory cue), we used conspecific skin extract, generated from five donor individuals. These chemical alarm cues are known to elicit observable innate alarm response in fish (Brown, 2003). To generate these alarm cues, each donor was euthanized by cervical dislocation (in accordance with animal ethic committee of the Federal University of Bahia, Project 40/2017), following which a skin fillet was extracted from each side of their body. We extracted approximately 10 cm2 of skin which we diluted in saltwater made from distilled water and commercially-purchased aquarium salt at a ratio of 1 cm2 of skin per 80 mL of water (see Leduc et al., 2010). Once prepared, chemical cues were packaged in 5 mL aliquots and frozen until needed. Upon conducting a test, an aliquot was thawed. Frozen aliquots were kept a maximum of two weeks (i.e., duration of the experiment). This duration is consistent with that of other experiments in which frozen alarm cues were stored prior to investigation (e.g., Leduc et al., 2010). As control for the injection procedure, we used saltwater. Each trial was initiated following the injection of 5 mL one of the two stimuli (randomly chosen) through airline tubing, using a 10 mL syringe. Following the injection procedure, we slowly removed the glass barrier separating the fish from the two physical features present in the aquarium (i.e., sea urchin or rock). Each trial lasted ten minutes and was video-recorded using a portable camera (GOPRO model Hero ™ Black Hero 3+) attached to a tripod. After the completion of all trials, fish were measured and all aquariums were emptied and refilled with unused sea water. Approximately 90% of all fish were returned to their natural rocky reef environment, however 10% died before their release.

3.2. Experiment 2: laboratory refuge choice test Using a Chi-Square test, we evaluated the relationship between the level of threat (i.e., alarm cue versus water) and the type of micro-habitat (i.e., sea urchins vs. rocks) first chosen as potential refuge. The time spent to arrive at the escape micro-habitat was investigated among treatments (i.e., alarm cues or control) using three-way Analysis of Variance (ANOVA), where arrival time was the dependent variable while the independent variables included the life phase (two levels: initial and terminal phase), microhabitat type (three levels: sea urchin, rocks and neither, if no choice was made) and the stimulus type (two levels: alarm cues or control). The normality and the homogeneity of the data were assessed with quantile–quantile (Q–Q) plots and the Levene's test, respectively.

4. Results 4.1. Experiment 1: field assessment of refuge selectivity Initial fish mostly fled to crevices (53% of the flights) and to E. lucunter (42%) while sponges and macroalgae were rarely chosen (2–3% of escapes). Terminal phase mostly fled to crevices (64%), followed by E. lucunter (25%), corals (9%), 1% to macroalgae and sponges (1%). E. lucunter occupied from 0 to 10% of the total benthic cover where O. trinitatis were found, suggesting an association between these organisms. Results of Ivlev's electivity index showed that both initial and terminal phase O. trinitatis preferentially fled to crevices and E. lucunter, and that they avoided macroalgae and zoanthids (Fig. 1). However, initial phase selected coral and sponges and terminal phase avoided these organisms. GLM analysis showed significant differences between the FID for O. trinitatis initial (18.78 ± 9.16, mean ± SD) and terminal phase (58.3 ± 24.26; F = 140.7, P < .001). Furthermore, FID was negatively influenced by E. lucunter density (F = 31.6, P < .001; Fig. 2a) and rock crevices (F = 15.6, P < .001, Fig. 2c), suggesting a refuge role provided by both. GLM analysis showed significant differences between densities for O. trinitatis initial (7.1 ± 6.6) and terminal phase (2.38 ± 1.6; F = 110.3; GLM, P < .001). E. lucunter densities influenced negatively fish densities (F = 80.5; GLM, P < .001, Fig. 2b) and rock crevices also influenced negatively fish densities (F = 60.9; GLM, P < .007, Fig. 2d).

2.2.3. Video analysis and data extraction All videos were analyzed by a single observer with no prior knowledge of the experimental treatments (i.e., chemical alarm cue or control). No order was established between fish samples (i.e., initial and terminal phase) or among experimental treatments. Precisely, the time spent until each fish reached one refuge option (by a body length or less) was recorded after the removal of the barrier, as well as the identity of the first refuge choice made. 3

Journal of Experimental Marine Biology and Ecology 520 (2019) 151207

J.A.C.C. Nunes, et al.

Fig. 1. Ivlev's index showing selection or avoidance (positive or negative value, respectively) between six physical features present on the rocky reefs. Gray and black bars represent initial and terminal phase of the redlip blenny Ophioblennius trinitatis, respectively.

4.2. Experiment 2: laboratory refuge choice test For both treatments (alarm cues and control), O. trinitatis initial phase mostly escaped toward E. lucunter (13 times out of 20 trials; X2 = 6; P < .05), while choosing rocks and the absence of choice occurred on 6 and 3 occasions, respectively). Terminal phase similarly choose between E. lucunter and rocks (n = 9 each; X2 = 1.5; P < .5), while the absence of choice occurred only twice. Although only marginally significant (P = .054), refuge choice was not affected by the interaction between ontogeny and treatment (Table SM3). The type of stimulus influenced time before reaching a potential refuge (ANOVA: F = 21.91; P < .001). For instance, arrival time to reach a potential refuge was different along with ontogeny and stimulus type (ANOVA: P < .05). Under control conditions, initial phase reached a refuge in slightly more than five minutes (5.18 ± 1.58,

Fig. 3. Average time before taking refuge in each of the micro-habitats for each of the initial (gray) and terminal (black) phases. In both figures (a and b), two left bars are sea urchins and other two (right side) are rock crevices. Treatments using water (control) and chemical alarm cues (experimental) are represented by the panels A) and B), respectively. The upper limits of lines indicate standard deviation.

Fig. 2. Relationships between (a) sea urchin Echinometra lucunter density (m2) and flight initiation distance (FID) of redlip blenny Ophioblennius trinitatis exposed to a predator model, (b) E. lucunter and O. trinitatis densities, (c) density of rock crevices and FID, and (d) rock crevices and fish densities. Black and gray dots represent data from terminal and initial phases, respectively. The continuous black and gray lines represent the best fit model for terminal and initial phases, respectively.

4

Journal of Experimental Marine Biology and Ecology 520 (2019) 151207

J.A.C.C. Nunes, et al.

reach only few centimeters in height, reducing their effectiveness to provide safe space against predators. Thus, the negative Ivlev's index for corals in terminal phase fish may be related to their small size. Perhaps, in pristine areas, where these corals reach greater size, prey fish may learn to associate them as effective refuge area. We found a negative relationship between sea urchin density and FID, suggesting that the presence of shelter (sea urchins) translates into greater tolerance toward predator approach. This strategy likely results in energy reduction allocated to escape (Cooper and Frederick, 2007). FID of initial phase fish was more strongly related with sea urchin densities than for terminal phase. According to our observations, juvenile O. trinitatis can swim between spines of the sea urchins whereas adults may only swim nearby to avoid predation. Almeida et al. (2010) found a positive correlation between sea urchins and small goby Elacatinus figaro densisties, suggesting the use of sea urchins for refuge. Body size can affect the antipredator strategy employed, and consequently how species will escape from particular threats (Blumstein, 2006). Large-bodied individuals might be more vulnerable, since they are generally less agile (Witter et al., 1994), but have higher detection abilities, which may select for great flight initiation distance (Blumstein et al., 2005). On the other hand, larger individual may be able to flee at greater speed, which may dampen this effect. Demersal and benthic fish's behaviour and densities are expected to be influenced by habitat complexity (Dalben and Floeter, 2012; Nunes et al., 2013). Habitat complexity (e.g., sea urchin density) was expected to influence FID of O. trinitatis and we found a positive association between O. trinitatis density and sea urchins. Beukers and Jones (1998) posited that an interplay exists between habitat complexity and predation risk, whereby upon abundant habitat features, predation risks are proportionally lower. This may be the results of predators' increased difficulty in detecting or physically reaching prey and/or by decreasing the frequency of encounters. Dalben and Floeter (2012) reported that cryptobenthic fish density was significantly correlated with black sea urchins (Echinometra lucunter), and mentioned that Starksia brasiliensis was only observer behind or near sea urchins. At Santa Catalina Island (California), manipulative experiments found a relationship between local abundance of a small temperate cryptobenthic fish, Lythrypnus dalli and presence of sea urchin Centrostephanus coronatus (Hartney and Grorud, 2002). Patzner (1999) also recorded the effect of sea urchins as a hiding place for juvenile benthic fishes in the Mediterranean Sea. In terms of refuge selectivity, the laboratory experiments corroborated the field findings for initial phase fish. Indeed, we found under both field and laboratory conditions that sea urchins were the most selected structural feature when escaping from predator. Furthermore, blenny's selectivity and its readiness to escape toward these refuges increased under higher perceived predation risk (alarm cues). In fact, high risk conditions translated into lower time before fleeing to refuge, which was lower for sea urchins than for rock when considering initial phase blennies. On the other hand, terminal phase strongly selected for rock crevices in field experiments, (sea urchin in second place) to seek refuge. Under laboratory conditions, this choice was confirmed under higher perceived predation risk, whereby the latency to escape to rock was shorter than for sea urchins. These ontogenic differences in refuge choice are most likely attributed to body size. We cannot conclusively state that the observed response patterns are due to the detection of conspecific alarm cues, but the response patterns are consistent with the well-studied chemosensory risk assessment system (Ferrari et al. 2010). Sea urchin populations are controlled mainly by fish predation (Sala, 1997; Sala and Zabala, 1996). Fishing activities have increased the population of sea urchins in many locations around the world and such high populations can lead to destructive bioerosion, loss of reef framework, and a decrease in coral species diversity (Bak, 1994; Eakin, 1996; McClanahan and Muthiga, 2007). The increases of urchin densities on tropical rocky reefs does not necessarily have a positive impact on small, cryptobenthic fish survivorship, experiments should be

Table 1 ANOVA three-way for aquarium experiments testing time to arriving in first microhabitat choice (sea urchin Echinometra lucunter or rock crevice) for two ontogenetic phases (initial and terminal) in the redlip blenny Ophioblennius trinitatis, under two stimulus (with and without predation risk). Bold values indicate significant results.

Intercept Life phase Stimuli Microhabitat Life phase*Stimuli Life phase*Microhabitat Stimuli*Microhabitat Life phase*Stimuli*Microhabitat Error

DF

MS

F

P

1 1 1 2 1 2 2 2 28

274.35 13.27 47.33 0.32 4.28 11.67 0.47 6.97 2.16

126.98 6.14 21.91 0.14 1.98 5.40 0.21 3.22

< 0.01 < 0.01 < 0.01 0.86 0.17 < 0.05 0.80 0.05

min ± SD) while adults spent slightly more than three minutes (3.18 ± 2.43) to reach a refuge. However, initial phase (3.83 ± 1.12) individuals spent less time than terminal phase (5.3 ± 1.6) to arrive in rocks under water stimulus. Under high risk conditions (alarm cues), initial phase individuals fled significantly faster to E. lucunter than terminal phase (Fig. 3; Table 1). However, initial phase was slower (1.69 ± 1.06) than terminal phase (1.21 ± 1.51) to reach rocks under predation stimulus. There was no difference on time spent to arrive regarding type of escape micro-habitat (ANOVA: F = 0.14; P < .9). 5. Discussion Our results suggest that the Brazilian red lip blenny Ophioblennius trinitatis is selective in the type of structural feature used as refuge when fleeing from a predator. Indeed, we showed that blenny fled from predators to sea urchins and crevices, and avoided other common habitat features (e.g., coral and sponges). To our knowledge, this is the first study that use a predation risk metric to demonstrate the habitat selection of sea urchin as refuge. Furthermore, the laboratory experiment results underscore that under high perceived predation risk (i.e., chemical alarm cues), this prey fish increases its selectivity in the type of habitat feature sought. Overall, these results suggest specificity in habitat features selection when escaping from predation threats. Sea urchins and crevices may serve well the purpose of escaping predator as the space they provided (i.e., between sea urchins' spines or within the crevices' cavity) is an effective barrier to stop predator advance (Giglio et al., 2018). Added to this ‘safe space’, the relative abundance of these features (i.e., both urchins and crevices are ubiquitous on these rocky reefs) may results in positive associations between these and the relative safety they provide. Indeed, many fish have the capacity to make positive associations learning with previous experiences (Brown, 2003), which presumable may be reinforced at each predatory encounter that prey survives. These two characteristics, the security provided and their availability, may be the most essential variables allowing prey to associate and directly select when seeking refuge. Contrary to urchins, which are abundant and relatively mobile, crevice do not change location hence these fish may escape to those given their keen cognitive abilities (i.e., learning and memory; Brown, 2003). Under high risk conditions, the danger of choosing a novel (and thereby untested) feature of its habitat may reduce the propensity to choose such feature when other “tested” features are present. Several coral species were found on these rocky shores, small massive corals Favia gravida and branching corals Millepora spp. are the most common. Although relatively small corals as Favia gravida do not provide protective refuge for fish, branching corals has been described as effective refuge for fishes (Coni et al., 2012; Leal et al., 2013). However, given the intense human pressures (e.g., fishing and tourism), Millepora spp. have been damaged in this region where their branch 5

Journal of Experimental Marine Biology and Ecology 520 (2019) 151207

J.A.C.C. Nunes, et al.

conducted aiming to verify if sea urchins increased density reduces predation effectively. However, it is likely that other small fish, such as labrisomids and other blenniids, also select sea urchins for refuge from predators, future studies are required to formally test this. Thus, the role and importance of sea urchins in providing safety for fishes will be clarified. Future experiments should manipulate refuge location and quality to test other aspects in refuge choice. As coral complexity and composition diminish due to a series of local and global impacts it is important to understand the ecological role of sea urchins on reef fish populations and also regulating predator-prey interactions. In general our study showed that on high risk conditions, the fishes translated into faster retreat to shelter and to higher refuge selectivity, when compared to low risk. The choices observed in situ were partly similar among initial and terminal phases. Our findings suggest that small reef fishes may be selective in their refuge choice even using habitats with high structural complexity.

Ferrari, M.C.O., Wisenden, B.D., Chivers, D.P., 2010. Chemical ecology of predator–prey interactions in aquatic ecosystems: a review and prospectus. Can. J. Zool. 88 (7), 698–724. https://doi.org/10.1139/Z10-029. Ferreira, C.M., Coni, E.O.C., Medeiros, D.V., Sampaio, C.L.S., Reis-Filho, J.A., Barros, F., Loiola, M., Nunes, J. de A.C. da C., 2015. Community structure of shallow rocky shore fish in a tropical bay of the southwestern Atlantic. Braz. J. Oceanogr. 63, 379–396. Giglio, V.J., Ternes, M.L.F., barbosa, M.C., Cordeiro, C.A.M.M., Floeter, S., 2018. Reef fish associations with sea urchins in an Atlantic oceanic island. Mar. Biodivers. 48 (4), 1833–1839. Hartney, K.B., Grorud, K.A., 2002. The effect of sea urchins as biogenic structures on the local abundance of a temperate reef fish. Oecologia 131, 506–513. https://doi.org/ 10.1007/s00442-002-0908-6. Hixon, M.A., Beets, J.P., 1993. Predation, prey refuges and the structure of coral reef fish assemblages. Ecol. Monogr. 63, 77–101. Humann, P., Deloach, N., 2002. Reef Fish Identification, 3rd edn. New World Publication, FL. Jones, K.M.M., 2002. Behavioural overlap in six Caribbean labrid species: intra- and interspecific similarities. Environ. Biol. Fish 65, 71–81. Krebs, C.J., 1989. Ecological Methodology. Harper Row 654. https://doi.org/10.1007/ s007690000247. Leal, I.C.S., Pereira, P.H.C., De Araújo, M.E., 2013. Coral reef fish association and behaviour on the fire coral Millepora spp. in north-East Brazil. J. Mar. Biol. Assoc. United Kingdom 93, 1703–1711. Leduc, A.O.H.C., Roh, E., Brown, G.E., 2010. Effects of acid rainfall on juvenile Atlantic salmon (Salmo salar) antipredator behaviour: loss of chemical alarm function and potential survival consequences during predation. Mar. Fresh Resear. 60 (12), 1223–1230. MacArthur, R.H., Pianka, E.R., 1966. On the optimal use of a patchy environment. Am. Nat. 100, 603–609. MacNeil, M.A., Graham, N.A.J., Polunin, N.V.C., Kulbicki, M., Galzin, R., HarmelinVivien, M., Rushton, S.P., 2009. Hierarchical drivers of reef-fish metacommunity structure. Ecology 90, 252–264. McClanahan, T., Muthiga, N., 2007. Ecology of Echinometra, in: Biology and Ecology. pp. 297–317. McDermott, C.J., Shima, J., 2006. Ontogenetic shifts in microhabitat preference of the temperate reef fish Forsterygion lapillum: implications for population limitation. Mar. Ecol. Prog. Ser. 320, 259–266. Mendes, L.D.F., 2006. História natural dos amborés e peixes-macaco (Actinopterygii, Blennioidei, Gobioidei) do Parque Nacional Marinho do Arquipélago de Fernando de Noronha, sob um enfoque comportamental. Rev. Bras. Zool. 23, 817–823. Mendes, L.D.F., 2007. Ophioblennius trinitatis (Pisces: Blenniidae) from the oceanic archipelagos of São Pedro e São Paulo, Fernando de Noronha and Atol das Rocas. Braz. J. Oceanogr. 55, 63–65. https://doi.org/10.1590/S1679-87592007000100008. Minte-Vera, C.V., De Moura, R.L., Francini-Filho, R.B., 2008. Nested sampling: an improved visual-census technique for studying reef fish assemblages. Mar. Ecol. Prog. Ser. 367, 283–293. Nunes, J. de A.C.C., Sampaio, C.L.S., Barros, F., 2013. How wave exposure, group size and habitat complexity influence foraging and population densities in fishes of the genus Halichoeres (Perciformes: Labridae) on tropical rocky shores. Mar. Biol. 160, 2383–2394. https://doi.org/10.1007/s00227-013-2233-5. Nursall, J.R., 1977. Territoriality in redlip blennies (Ophioblennius atlanticus – Pisces; Blenniidae). J. Zool. (Lond.) 182, 205–223. Ogden, J.C., Lobel, P.S., 1978. The role of herbivorous fishes and urchins in coral reef communities. Environ. Biol. Fish 3, 49–63. Patzner, R.A., 1999. Sea urchins as a hiding-place for juvenile benthic teleosts (Gobiidae, Gobiesocidae) in the Mediterranean Sea. Cybium 23, 93–97. Patzner, R.A., Santos, R.S., 1992. Field observations on the association between the clingfish Diplecogaster bimaculata pectoralis (Briggs 1955) and different species of sea urchins at the Azores. Zeitschrift für Fischkunde. 1, 157–161. Pereira, P.H.C., Munday, P.L., Jones, G.P., 2015. Competitive mechanisms change with ontogeny in coral dwelling gobies. Ecology 96, 3090–3101. https://doi.org/10.1890/ 14-1689.1. Pérez-Matus, A., Shima, J., 2010. Density- and trait-mediated effects of fish predators on amphipod grazers: potential indirect benefits for the giant kelp Macrocystis pyrifera. Mar. Ecol. Prog. Ser. 417, 151–158. R Development Core Team, R, 2016. R: A Language and Environment for Statistical Computing. https://doi.org/10.1007/978-3-540-74686-7. Rangel, C.A., Mendes, L.F., 2009. Review of Blenniid fishes from Fernando de Noronha archipelago, Brazil, with description of a new species of Scartella (Teleostei: Blenniidae). Zootaxa 51–61. Rosenzweig, M.L., 1981. A theory of habitat selection. Ecology 62, 327–335. Sala, E., 1997. Fish predators and scavengers of the sea urchin Paracentrotus lividus in protected areas of the north-West Mediterranean Sea. Mar. Biol. 129, 531–539. Sala, E., Zabala, M., 1996. Fish predation and the structure of the sea urchin Paracentrotus lividus populations in the NW Mediterranean. Mar. Ecol. Prog. Ser. 140, 71–81. Vergés, A., Vanderklift, M.A., Doropoulos, C., Hyndes, G.A., 2011. Spatial patterns in herbivory on a coral reef are ifluenced by structural complexity but not by algal traits. PLoS ONE 6, e17115. https://doi.org/10.1371/journal.pone.0017115.g006. Witter, M.S., Cuthill, I.C., Bonser, R.H.C., 1994. Experimental investigations of massdependent predation risk in the European starling, Sturnus vulgaris. Anim. Behav. 48, 201–222.

Acknowledgments We thank Lize Souza, Yuri Costa, Amanda Martins and Tiago Albuquerque for helping during field work. Miguel Loiola and Alice Reis for help in revision. Camilo Ferreira (University of Adelaide, AUS) and Daniel Blumstein (UCLA, USA) exchanged ideas with the first author. We also thank CAPES and CNPq for the financial support to J.A.C.C.N. A.O.H.C.L. acknowledge receiving a PNPD post-doctoral grant from CAPES. FB was supported by CNPq fellowships (PQ 306332/ 2014-0; 304907/2017-0). Appendix A. Supplementary data Supplementary data to this article can be found online at https:// doi.org/10.1016/j.jembe.2019.151207. References Almany, G.R., 2004. Does increased habitat complexity reduce predation and competition in coral reef fish assemblages? Oikos 106, 275–284. Almeida, D.F., Solé-Cava, A.M., Calderon, E.N., 2010. The sea urchin Echinometra lucunter 9Echinodermata, Echinoidea as a refuge for the barber goby Elacatinus figaro (Perciformes, Gobiidae). Arq. Mus. Nac. 68, 17–23. Bak, R.P.M., 1994. Sea urchin bioerosion on coral reefs: place in the carbonate budget and relevant variables. Coral Reefs 13, 99–103. Beukers, J.S., Jones, G.P., 1998. Habitat complexity modifies the impact of piscivores on a coral reef fish population. Oecologia 114, 50–59. Blumstein, D.T., 2006. Developing an evolutionary ecology of fear: how life history and natural history traits affect disturbance tolerance in birds. Anim. Behav. 71, 389–399. Blumstein, D.T., Fernández-Juricic, E., Zollner, P.A., Garity, S.C., 2005. Inter-specific variation in avian responses to human disturbance. J. Appl. Ecol. 42, 943–953. Brown, G.E., 2003. Learning about danger: chemical alarm cues and local risk assessment in prey fishes. GE Brown. Fish Fisheries 4 (3), 227–234. Carvalho, L.R.S., Barros, F., 2017. Physical habitat structure in marine ecosystems: the meaning of complexity and heterogeneity. Hydrobiologia 1–9. Coni, E.O.C., Ferreira, C.M., de Moura, R.L., Meirelles, P.M., Kaufman, L., Francini-Filho, R.B., 2012. An evaluation of the use of branching fire-corals (Millepora spp.) as refuge by reef fish in the Abrolhos Bank, eastern Brazil. Environ. Biol. Fish 96, 45–55. Cooper, W.E., Frederick, W.G., 2007. Optimal flight initiation distance. J. Theor. Biol. 244, 59–67. https://doi.org/10.1016/j.jtbi.2006.07.011. Cooper, W.E., Samia, D., 2018. Choosing among alternative refuges: distances and directions. Ethology 124, 209–217. Cordeiro, C.A.M.M., Harborne, A.R., Ferreira, C.E.L., 2014. Patterns of distribution and composition of sea urchin assemblages on Brazilian subtropical rocky reefs. Mar. Biol. 161, 2221–2232. Crawley, M.J., 2007. The R Book, The R Book. John Wiley and Sons. Dalben, A., Floeter, S.R., 2012. Cryptobenthic reef fishes: depth distribution and correlations with habitat complexity and sea urchins. J. Fish Biol. 80, 852–865. https:// doi.org/10.1111/j.1095-8649.2012.03231.x. Eakin, C.M., 1996. Where have all the carbonates gone? A model comparison of calcium carbonate budgets before and after the 1982-1983 El Niño at Uva Island in the eastern Pacific. Coral Reefs 15, 109–119. Ferreira, C.E.L., Goncçalves, J.E.A., Coutinho, R., 2001. Community structure of fishes and habitat complexity on a tropical rocky shore. Environ. Biol. Fish 61, 353–369.

6