Social flexibility to balance habitat fragmentation? Insights from the Mediterranean cave-dwelling cardinalfish Apogon imberbis

1 INTRODUCTION

Environmental changes have always existed, long before the dominating influence of humans (Wong & Candolin, 2015). However, the extensive and rapid nature of anthropogenic changes confronts biodiversity to new major threats. Habitat fragmentation is one of them (Fahrig, 2003). Mitigating the negative impact of habitat fragmentation on biodiversity requires conservation programs to study species' responses to this threat in order to be effective.

The underwater caves of the Mediterranean Sea constitute a privileged model to study the impact of habitat fragmentation on species distribution and dynamics (Chevaldonné, Rastorgueff, Arslan, & Lejeusne, 2015; Lejeusne & Chevaldonné, 2006; Muths, Rastorgueff, Selva, & Chevaldonné, 2015; Rastorgueff & Bianchimani, 2016; Rastorgueff, Chevaldonné, Arslan, Verna, & Lejeusne, 2014). They mostly originate from the marine flooding of limestone karstic networks during the last transgression (Bianchi, Cattaneo-Vietti, Cinelli, Morri, & Pansini, 1996). Hence, they constitute a naturally fragmented habitat along the rocky coast made of small crevices to large cavities. They are characterized by severe environmental conditions compared to the outside photic zone: absence of light, reduced hydrodynamism, and strong oligotrophy (Fichez, 1990). These prevailing harsh environmental conditions make the underwater marine caves of the Mediterranean Sea a very fragile, sensitive, and poorly resilient ecosystem (Harmelin, Vacelet, & Vasseur, 1985; Montefalcone et al., 2018). However, they also constitute a favorable habitat to a variety of specialized and endemic fauna (Cicogna, Bianchi, Ferrari, & Forti, 2003; Harmelin et al., 1985). They can even be considered as a reservoir of biodiversity (Gerovasileiou & Voultsiadou, 2012). This diversity of biological models allows us to study the different aspects of species' response to habitat fragmentation (Chevaldonné et al., 2015; Lejeusne & Chevaldonné, 2006; Muths et al., 2015; Rastorgueff & Bianchimani, 2016; Rastorgueff et al., 2014).

The Mediterranean cardinalfish Apogon imberbis is an interesting model when studying the impact of habitat fragmentation on species distribution and dynamics because of its surprisingly high population connectivity for a brooding cave dweller. It shows enough gene flow to homogenize the genetic pattern at a regional scale (Muths et al., 2015). Several factors can explain such absence of genetic structure. First, it goes out of caves at night (Bussotti et al., 2018; Bussotti, Guidetti, & Belmonte, 2003). Second, besides brooding, it displays a pelagic larval phase of ca. 18–24 days (Azzurro, Pais, Consoli, & Andaloro, 2007; Raventós & Macpherson, 2001). Third, it can be found from the surface to 200 m depth which makes dispersal in depth conceivable (Poortvliet et al., 2013; Tortonese, 1986). And fourth, it is able to use an exceptionally wide range of shadowed shelters: caves, cracks, cavities, under overhanging rocks, boulders, and seagrass matte (Bussotti & Guidetti, 2009). A. imberbis is not restricted to one form of social aggregation. It can be solitary in small crevice (Figure 1b) as well as in big swarms in large caves (Figure 1a) (Bussotti et al., 2018; Bussotti et al., 2003; Mazzoldi, Randieri, Mollica, & Rasotto, 2008). This ability to use a wide range of sizes of habitat patches, due to the high phenotypic plasticity in its sociality, is believed to contribute to balance the effect of habitat fragmentation on connectivity disruption of its populations (Muths et al., 2015). However, the mechanism behind such phenotypic plasticity remains unknown.

Building up on the study of the genetic structure of its populations (Muths et al., 2015), the present paper aims at proposing preliminary insights to a functional hypothesis of the exceptional social plasticity of A. imberbis, from a behavioral perspective. It will also explore the resulting ecological and evolutionary implications regarding habitat fragmentation and environmental changes.

2 FLEXIBLE IN SOCIAL BEHAVIOR AND SOCIAL FLEXIBILITY

The plasticity in species aggregative behavior seems to mainly originate from two mechanisms that should not be confused: flexibility in the social behavior and social flexibility (Schradin, 2013). Flexibility in the social behavior (which operates at the individual scale) is the mechanism where individuals' social behavior shown toward other group members is flexible, but the social organization of the group is rigid. As an example, individual primates display flexible tactics (e.g., change of dominance of submission) toward arising conflicts and environmental and social changes to maintain their species-specific social organization and the associated fitness benefit of living in groups. In addition, helping their relatives may provide indirect evolutionary success and being with familiars reduces aggressions. The social flexibility (which operates at the population scale) corresponds to the mechanism where the social organization of a group of conspecifics changes as individuals reversibly change their social tactics in response to short-term environmental changes. The social system and the social composition of the group change because individuals modify their interactions with each other: from individuals to groups and back, as an example (Farine, Montiglio, & Spiegel, 2015).

Interestingly, the two mechanisms seem to be exclusive (Schradin, 2013). A socially flexible species will not show flexibility in the social behavior of its individuals, and vice versa. Social flexibility arises in species where no notable dominance hierarchy exists, and then, groups seem to be highly egalitarian. When the individuals lack of flexibility in their social behavior, resolving pacifically an arising conflict implies switching to solitary as the only adaptive alternative to group living. In addition, social flexibility seems to be a feature of species living in an unpredictably fluctuating environment (Randall, Rogovin, Parker, & Eimes, 2005). Such kind of environment renders hard to maintain a stable group structure, making social flexibility the best chance to survive when rapid adaptive responses are needed (Schradin, 2013).

3 THE UNPREDICTABILITY OF MEDITERRANEAN UNDERWATER MARINE CAVES

The underwater caves of the Mediterranean Sea are holes in the hard bottom of the coast, submerged after the last glaciation, and showing steady environmental conditions which allow slow growing species to survive (Boury-Esnault, Harmelin, & Vacelet, 1993; Harmelin, 1997; Harmelin et al., 1985). Consequently, they are clearly not a highly unpredictable rapid changing environment, but more of an environment stable in space and time. However, the unpredictability of underwater marine caves resides in the quasi-infinite variability of the geomorphology of each cave, making each of them unique (Bussotti, Terlizzi, Fraschetti, Belmonte, & Boero, 2006; Fichez, 1990; Gerovasileiou, Dimitriadis, Arvanitidis, & Voultsiadou, 2017; Gerovasileiou, Trygonis, Sini, Koutsoubas, & Voultsiadou, 2013; Harmelin, 1997; Parravicini et al., 2010; Rosso, Sanfilippo, Ruggiero, & Di Martino, 2013). Then, for a nycthemeral migrating species, the habitat changes unpredictably from cave to cave. This can have no notable effect on solitary species, but it can be problematic for those who needs big caves (Chevaldonné et al., 2015; Rastorgueff & Bianchimani, 2016; Rastorgueff et al., 2014). The level of population connectivity shown by A. imberbis suggests that it will use many daytime resting sites during the course of its life time, up to five years (Muths et al., 2015; Raventos, 2007). In addition, site fidelity is not mandatory in cardinalfishes, including cave dwellers (Gardiner & Jones, 2016; Rueger, Gardiner, & Jones, 2016). Consequently, caves will appear as a highly unpredictable environment to A. imberbis because of its capacity to inhabit a great variety of shadowed shelters, without prior knowledge of the local geomorphology or the previously occupying conspecifics.

4 SOCIAL FLEXIBILITY OF APOGON IMBERBIS

Apogon imberbis is the only native Apogonidae present in the Mediterranean Sea, exception made of the lessepsian migrant Apogon queketti (Gilchrist, 1903) (Eryilmaz & Dalyan, 2006). Most of the other species lives in tropical or sub-tropical waters inhabiting coral reefs and lagoons (Paxton & Eschmeyer, 1998).

Like other apogonids, it is a nocturnal fish feeding outside at night and spending the day in dark shelters in rocky cliffs, such as caves or crevices, where it can be found forming small groups or large swarms (Figure 1) (Bussotti et al., 2018; Bussotti et al., 2017; Bussotti et al., 2003; Paxton & Eschmeyer, 1998; Raventos, 2007).

Like all other of his kind, it is a small fish, reaching a maximum size of 15 cm at adulthood (Raventos, 2007). However, it seems to have the longest life span of the apogonids: up to five years for both sexes (Raventos, 2007). Both sexes reach maturity at the age of 1 year, and the spawning season takes place between June and September (Raventos, 2007). The nuptial court is made of several steps (Garnaud, 1950, 1963). The male turns around the female following a spiral pattern, while the female turns on herself following the male by sight. When the male is the closest to the female, the external fecundation starts until the female has expelled all her eggs. Then, the male keeps them in its mouth until the juvenile stage. This mouth brooding behavior demands fasting from the male and appears to generate an aggressive behavior toward its conspecifics by chasing them.

While many elements can be found on the biology and ecology of A. imberbis, its social structure remains largely unknown. However, several indirect factors suggest that social flexibility could constitute its social structure. Hierarchical dominance appears rare in cardinalfishes (Kolm, Hoffman, Olsson, Berglund, & Jones, 2005). Flexibility in sociality is largely documented in cardinalfishes (Gardiner & Jones, 2005, 2016). Numerous cardinalfish species range from solitary living to large schools (Randall et al., 2005). Apogon imberbis can be solitary, especially during brooding time, most likely to avoid costly conflicting situations when energy saving is mandatory for brooding, especially due to the associated fasting. Cardinalfishes able to make more new home sites are less constrained by habitat and social preferences (Gardiner & Jones, 2016). This appears to be the case for A. imberbis since its genetic connectivity suggests poor site fidelity and its homing behavior does not result in the formation of kin groups (Muths et al., 2015). This is corroborated by unpublished data of a site fidelity study: After 8 days, around 60% of fish marked were not resighted in the cave where they were captured and released, and around 50% of fish marked were not resighted in the cave where they were initially captured when released away outside (Webster, Swann, & Richtik-Rinaudo, 2010). These preliminary results suggest that although there is a tendency to site fidelity, it only concerns half of the fish studied. The other half would necessarily use other shadowed shelters.

All these elements, and particularly the genetic pattern at regional and population scales, indicate that habitat and social preferences do not entirely drive the homing behavior of A. imberbis at dusk. Protecting a nest or looking for the benefits of kin-related conspecifics does not mainly determine the dispersing movement of A. imberbis either. Consequently, the only apparent driver of its nycthemeral migration from caves to the open sea and back is the sciaphilic constraint. The only requirement A. imberbis should have to compile at dusk is to find a shadowed shelter, independently of the size of the habitat patch or the kinship of already-in conspecifics. Finally, besides being sciaphilic, A. imberbis appears as a true generalist cave dweller.

The ecological and evolutionary perspectives also suggest that a flexible social structure presents several advantages for A. imberbis. It allows maintaining a high level of connectivity in a fragmented habitat by making profit of all the sizes of cavities. It allows concentrating dispersers in large caves, which eventually enhance population connectivity. It can also enhance survival by increasing the ability to find a shelter when displaced and by reducing the predation pressure (Marnane, 2000; Webster et al., 2010; Wilson et al., 2008). Being a generalist also increases the chance of survival and recovery when facing environmental disturbances. However, concentrating in the same caves for an extended period of time may represent localized and predictable resources for predators, especially for colorful species such as cardinalfishes (Marnane, 2000). Then, the cardinalfish individuals would move from a shadowed shelter to another in order to avoid encountering a predator (Marnane, 2000). This burden can also be reduced by the dilution effect. Being highly similar in morphology and color pattern can reduce the predation pressure on individuals by transferring it to the group.

5 LARVAL CONNECTIVITY AND ADULT SITE FIDELITY

The unpublished results of the site fidelity study of Webster et al. (2010) tend to temper the importance of the social flexibility of A. imberbis in the explanation of its surprising absence of genetic structure. If a notable proportion A. imberbis shows a certain degree of site fidelity, some genetic structure should be expected, not as high as the highly cavernicolous brooding species Hemimysis margalefi, but more intermediate as the brooding cave dweller Harmelinella mariannae (Chevaldonné et al., 2015; Rastorgueff et al., 2014). Consequently, the results of the genetic and site fidelity studies of A. imberbis appear contradictory. However, the genetic pattern represents historical reproductive connectivity, while site fidelity represents contemporary demographic connectivity. Moreover, genetic data alone provide little information about demographic connectivity, marine genetics is hard to interpret, and there can be many confounding factors in the detection of species response to habitat fragmentation (Ewers & Didham, 2006; Lowe & Allendorf, 2010; Palumbi, 2003). Consequently, these two types of connectivity cannot be interpreted at the same level and may not be congruent, but can be used in combination to better understand patterns of dispersal and population structure (Fedy, Martin, Ritland, & Young, 2008). In the case of A. imberbis, we can imagine that the factors promoting connectivity (larval phase, social flexibility, habitat generalist, avoiding concentration point of high predation pressure, depth dispersal, night dispersal) are of more importance in its connectivity dynamic than the factors reducing connectivity (site fidelity, habitat fragmentation, obligate to cave at night, brooding behavior).

6 POSSIBLE EXPERIMENTS

In order to unravel the mystery of the population connectivity and dispersal pattern of A. imberbis, several possible experiments can be realized. They will help identifying the social structure of this cardinalfish, the factors (e.g., which life stage disperse? which not?) or the combination of factors that can explain the surprising absence of population genetic structure for a cave-dwelling brooding species, and at which scale they operate.

Electronic tagging and tracking fish using archival tags would be the best to explore the contemporaneous population connectivity of A. imberbis (Calò et al., 2013). Archival tags store oceanographic data continuously, allowing the visualization of trajectories on an intermediate spatial scale during a long time. The inconvenience is that they have to be retrieved to get the data. This can be illusional according to the genetic connectivity of A. imberbis or unusable for our purpose if site fidelity is strong. However, taking into account the complexity of the model, the facility of access to underwater Mediterranean caves, and the need for more information, it is worth trying.

Otolith microchemistry can be preferred as a first attempt to characterize the contemporaneous connectivity of A. imberbis. It constitutes a fundamental source of information for investigating larval dispersal and connectivity patterns of fishes (Calò et al., 2013). This technic has been successfully used for estimating the age of A. imberbis individuals (Raventos, 2007). In addition, some underwater caves of the Mediterranean Sea can display specific stable isotope signature due to their vicinity of sewage plants (Rastorgueff, Harmelin-Vivien, Richard, & Chevaldonné, 2011) which could facilitate the identification of the origin of dispersers.

Homing experiment by relocating individuals away from their original shelter could be used to better explore the site fidelity and dispersal abilities of A. imberbis adults. This experiment will be the continuation of the preliminary study of Webster et al. (2010) following Rueger et al. (2016) and Gardiner and Jones (2016). This experiment could be easy to settle since cardinalfishes appear easy to displace, and especially in the Marseille (France) area due to the high number of caves (Gardiner & Jones, 2016; Marnane, 2000; Rastorgueff et al., 2014; Rueger et al., 2016; Webster et al., 2010).

Behavioral plasticity and hierarchy can be assessed to explore the reaction of the group to introduced new individuals: The group welcomes them or shows an aggressive behavior toward new conspecifics (Suriyampola et al., 2016). The plasticity of the group structure can also be explored in aquariums by manipulating its member composition or allowing an individual to choose between two groups (Engeszer, Ryan, & Parichy, 2004; Peichel, 2004). This should allow us to see whether A. imberbis tends more toward hierarchical grouping or random association. This experiment would be easy to settle since A. imberbis can be easily maintained in water tanks (Garnaud, 1950).

7 CONCLUSION

The social flexibility of A. imberbis appears as a very seductive hypothesis to explain its exceptional phenotypic plasticity and genetic connectivity and the challenges it faces to survive. However, elements of its biology and ecology tend to temper this enthusiasm and remind us that this hypothesis is still speculative. Nonetheless, few experiments can help us solve this question. They should not be too hard to conduct since A. imberbis is a relatively accessible and easy to manipulate model. In addition, the present paper has revealed the importance of taking into account for the evolutionary and ecological time scales, the different life stages, and their ecology when exploring the population dynamics of this fish. Finally, this speculative article stressed that A. imberbis appears as a much more complex model than previously thought but much more interesting, so far.