Space partitioning among damselfishes in the Caribbean coast of Panama: the role of habitat preferences

Este estudo teve como objetivo avaliar a abundância e uso do habitat por juvenis e adultos de peixes-donzela, assim como as comunidades bentonicas em distintos recifes de Isla Colon, Bocas del Toro, Panama. Os recifes foram escolhidos considerando diferentes niveis de exposicao a ondas e profundidade. As comunidades de peixes e bentos estudadas se mostraram distintas. A especie mais abundante em recifes com maior profundidade e abrigados foi Stegastes planifrons, seguida por S. leucostictus, que tambem foram observadas em recifes rasos abrigados. As baixas abundâncias de S. partitus e S. variabilis tambem foram observadas no primeiro recife citado e estas especies foram aparentemente restritas a profundidade maiores. Adicionalmente, estes mesmos recifes apresentaram uma vasta cobertura de corais massivos, vivos ou mortos. Stegastes adustus apresentou sua maior abundância em baixas profundidades, indicando uma preferencia por habitats rasos, aumentando em duas vezes com a exposicao a ondas. Os recifes rasos tambem tiveram maior abundância de Microspathodon chrysurus. Alem disso, o recife com maior exposicao apresentou extenso percentual de cobertura por algas e o coral-de-fogo Millepora alcicornis, reconhecido habitat preferencial de M. chrysurus e outras especies de peixes. Neste trabalho, diferentes padroes de uso do habitat foram encontrados, sendo regulados pela profundidade e/ou exposicao e ainda a disponibilidade de determinada cobertura bentonica. Para especies fortemente associadas ao substrato, a especializacao, a competicao e padroes nao aleatorios do recrutamento sao conhecidos determinantes dos padroes distribucionais. Este estudo apresentou resultados similares a outros conduzidos ha 30 anos no Panama e outros recifes do Caribe, indicando que as especies mais abundantes mantem forte afinidade com o habitat. Entretanto, severas reducoes na disponibilidade de habitat, seguidas de eventos de mortalidade em massa, podem ter possivelmente afetado seus padroes de distribuicao.


Introduction
Coral reefs are among the most productive systems in the marine environment (Spalding et al., 2001). These systems are changing worldwide in the last decades, due to climate changes, hurricanes, diseases, overfishing, and eutrophication (Hughes et al., 2003;Pandolfi et al., 2005). Such disturbances have often resulted in ecological phase shifts, from coral-to macroalgal-dominated reefs (McCook, 1999;Scheffer et al., 2001), which may be observed on many Caribbean reefs (Hughes, 1994;McClanahan et al., 2002) and other regions worldwide (Hughes et al., 2007).
Coral cover declined by nearly 80% in the Caribbean in the late 1970s and 1980s (Gardner et al., 2003), with decreases as high as 50-70% recorded in several areas along the coast of Panama (Guzman, 2003). Massive coral loss, in turn, has caused declines in structural complexity, leading to negative consequences to the entire reef community (Alvarez-Filip et al., 2009). Especially when species strongly dependent upon the cover, diversity and complexity of their respective habitat are usually more jeopardized by these natural and/or anthropogenic disturbances (Jones & Syms, 1998;Jones et al., 2004;Graham et al., 2006).
This group has been studied throughout the Caribbean, with data available for habitat use, recruitment, behavior, diet and, age and growth of some species (Robertson, 1984(Robertson, , 1996Gutierrez, 1998;Holbrook et al., 2000;Wilson & Meekan, 2002). However, most of these studies to date focused separately on a single or few species, thus not providing more comprehensive data on damselfish community and their association with environmental descriptors. This study aimed to determine abundance and space use of damselfishes across a coastal island of Panama, as well as their relationship with environmental factors such as depth, wave exposure and benthic cover using a more quantitative approach.

Study area
The present research was conducted in Bocas del Toro archipelago, Isla Colón, Panama. The whole archipelago comprises an area of ca. 3,500 km², formed by six large forested islands and hundreds of small mangrove cays (Guzmán, 2003). Surrounding all these islands and along a large part of the mainland coast, there are fringing coral reefs up to 20 m deep and dozens of shallow reef patches. The whole area represents the second most important contributor on reef formations in the Caribbean coast of Panama with approximately 87 km² of fringing reefs (Guszmán, 2003). The reefs of the archipelago have been evaluated in the past decade and live coral cover up to 50% has been reported, depending on the depth, reaching up to 90% in shallow areas (Guzmán & Guevara, 1998a, 1998b, 1999, 2001. Nevertheless, declines on coral cover throughout this area have also been observed in the last decade (Guzmán, 2003).
Three distinct reefs were chosen for this study (Fig.1). Pete's Reef (PT) is a shallow lagoon with patch reefs near mangrove mud flats (~1m). Water motion is low and benthic communities are dominated by numerous species of sponges and the fingercoral Porites porites (Pallas, 1766). Conch Point (CO) is a wave-protected near-shore fringing reef lined with mangroves (Rhizophora mangle L.). Depths range from 0.5 to 1 m at the mangrove edge and increase to approximately seven m at a 20 m distance from shore. The substratum immediately adjacent to the mangrove edge is covered with turtle grass Thalassia testudinum Koenig and green algae, while deeper areas are characterized by the presence of patch reefs and sponges. Shallower (~0.50 m) and relatively more exposed, Lime Point (LM) has a much longer fetch reaching the open waters of the Caribbean; lacking mangroves and characterized by patches of T. testudinum and Syringodium filiforme Kuetzing (manatee grass), as well as sand alongside its extension. Within this reef, sites closer to the reef edge, and deeper (~1 m) ones showed higher coverage of live coral. On PT and LM observations were done through snorkeling, whereas SCUBA was used in CO.

Surveys
The abundance and distribution of territorial damselfishes and benthic community were assessed in December 2010 and January 2011. Abundance of juveniles and adult damselfishes, as well as habitat use, were assessed using visual census for fish density and percent benthic cover over transects of 10 x 2 m (PT=20; CO=20, and LM=16; N=56). The studied damselfish species are easily distinguishable underwater based on their coloration, both as juveniles and adults. To minimize edge effect, records included damselfishes with at least 50% of its territory inside the transect boundaries.

Statistical analysis
One-way Analysis of Variance (ANOVA) was used to test differences between species abundance at both life stages (juvenile and adult) and benthic percent cover among reefs. When necessary, data was log10 (x + 1) transformed, and the Student-Newman-Keuls a posteriori test was used. The structure of damselfish assemblages (i.e., relative abundance of the different species and life stages) was analyzed using non-Metric Multidimensional Scaling Analysis (MDS), which was later tested for significance of groups formed for exposure and depth factors by the Analysis of similarity (oneway ANOSIM). Separate One-way ANOSIM for these factors was also used to test differences between benthic communities (Clarke & Gorley, 2001). Species' life stages and benthic community relationship was obtained, using the canonical correspondence analysis (CCA) ordination. Significant environmental parameters were selected with the Monte Carlo permutation test. Data were log-transformed and the CANOCO for Windows 4.5 software was used (ter Braak, 1995).

Damselfish distribution among reefs
Fifty-six visual censuses yielded sightings of 1,178 damselfishes belonging to six common Caribbean species: the dusky damselfish Stegastes adustus (Troschel 1865), the threespot damselfish S. planifrons (Cuvier, 1830), the beaugregory S. leucostictus (Müller & Troschel, 1848), the bicolor damselfish S. partitus (Poey, 1868), the cocoa damselfish S. variabilis (Castelnau, 1855), and the yellowtail damselfish Microspathodon chrysurus (Cuvier, 1830) ( Table  1). The former two species represented the majority of overall abundance, including adults and juveniles (~90%). Stegastes adustus was especially abundant in shallow reefs (PT and LM) (Fig. 2). The abundance of both juveniles and adults from this species were very similar in Pete's Reef, but adult abundance was significantly different between this site and Lime Point (ANOVA; p<0.0001) -on the latter it was observed a two-fold increase with wave (Fig. 2). Stegastes planifrons presented closer affinity to sheltered reefs (PT and CO), with a significant three-fold increase (ANOVA; p<0.0001) in shallow waters (PT). On the exposed reef (LM) juveniles presented very low abundances and adults were absent. The beaugregory S. leucostictus was not very abundant when compared with other species, but were observed only in sheltered reefs (PT and CO). Adults from this species were restricted to deeper patch reefs (CO), whereas juveniles did not differed in abundance between depths. Juveniles and adults of S. variabilis and S. partitus presented very low abundances and were restricted to deeper patches (CO). The yellowtail damselfish M. chrysurus was recorded both in the shallow and exposed (LM) and the deeper and sheltered reef (CO), showing no constraint concerning depth or exposure. Adults and juveniles, however, significantly increased 3x and 8x, respectively at LM (ANOVA; p=0.0037 and p=0.0002, respectively) together with the abundance of adult S. adustus (Table 1; Fig. 2), indicating preference for exposed reefs.
Non-metric multidimensional scaling analysis presented in Fig. 3 (a-b) showed distinct groups for damselfish communities, which were separated between wave exposure and depth. The low stress value indicated that the observed sample distribution in the two dimensional space provided a reliable interpretation of the differences among damselfish community structure at the study sites. Separated one-way ANOSIM corroborated the formation of these groups, with low degree of overlap between exposure (global R= 0.654, p=0.001) and depth (global R=0.589, p=0.001).

Benthic communities
Benthic communities also contributed for differences found among reefs. However, one-way ANOSIM tests detected significant differences among reefs regarding exposure (R=0.655; p=0.001), but not for benthic cover (R=0.092; p=0.055).
Pete's Reef (PT) and Conch Point (CO), both being sheltered reefs, presented higher cover of live or dead corals, including massive corals (mainly brain corals) and the branching finger coral Porites spp. (Table 1). Coral cover was always higher on the shallower reef (PT). Particularly, the deeper reef (CO) presented higher cover of sponges and sand, the latter being a consequence of more dispersed patch reefs. The exposed reef (LM) was an algae-dominated reef, comprising crustose, articulated and macroalgae. Also, colonies of the firecoral Millepora alcicornis Linnaeus, 1758, zoanthids and rubble were commonly observed. Table 1. Species abundance by life stage and percent composition of substrata (Mean ± SE), plus ANOVA results and Student-Newman-Keuls (SNK) comparison between reefs. ns = non-significant; *p<0.05; **p<0.01; ***p<0.001.

Fish-habitat relationships
Habitat relationships for damselfish and benthic descriptors using the canonical correspondence analysis are presented on Fig. 4a-b. The first two axes were significant (p=0.001) and accounted for 61.4% and 21.6%, respectively, of the relationship between species and explanatory variables. The first axis, responsible for most of explained variation, was positively related to habitats that presented high cover of branching corals, macroalgae and articulated calcareous algae (Fig. 4a). Such descriptors represented samples of Lime Point (shallow and exposed), which S. adustus and M. chrysurus were more associated with (Fig. 4b). A discrete segregation between juvenile and adult conspecifics from these species was observed, where S. adustus juveniles were strongly associated to the complex structure of fire-corals. This axis was, in the other hand, negatively represented by deeper patch reefs, contributing to high records of sandy bottoms, together with a high cover of live and dead massive corals. Such descriptors represented mostly Pete's reef and Conch Point samples (Fig. 4b). Axis I drove most of the juveniles of S. leucostictus, S. partitus, and S. variabilis abundance, as they were found associated with depth. Stegastes planifrons (juveniles and adults) showed a close relationship with massive and dead coral cover, together with adult S. partitus and S. leucostictus, showing preference for sparser patch reefs with higher cover of massive corals interspersed with sandy bottoms. Adult S. variabilis presented a distinct segregation from juveniles, but also associating with deeper patch reefs, explained by axis II.

Habitat use
Although this work used a more quantitative approach than previous studies, damselfish distribution seemed to follow the prevailing patterns of habitat use found for this group of fishes throughout the Caribbean Sea in the last decades (Itzkowitz, 1977;Waldner & Robertson, 1980;Robertson, 1981Robertson, , 1984Robertson, , 1996. However, differences were found in respect to benthic cover they associated to. In addition, it was possible to detect the habitat partitioning and segregation among different species.
At a broader scale, differences regarding reefs were found not only for damselfish species composition, but also for benthic communities. These differences may be due to the influence of wave energy. It has been demonstrated that wavestressed habitats are dominated by organisms with life histories adapted to a disturbance regime, presenting high resistance, such as coralline algal communities (Witman, 1992). In this study, reefs with more pronounced wave surge allowed a diverse algal community, including crutose types, as much as the encrusting and heavily calcified firecoral Millepora alcicornis, with a major Considering damselfish abundance, the dusky damselfish S. adustus presented higher abundances on Pete's reef and Lime Point, indicating a preference of this species for shallow habitats, but it exhibited a two-fold increase in abundance at higher wave surge. This species is not particularly abundant in other Panamanian reefs (Robertson, 1996), but has been already recognized to prefer shallow and high wave energy habitats (Waldner & Robertson, 1980;Gutierrez, 1998;Dominici-Arosemena & Wolff, 2005). It also refuges on structural complex colonies of the elkhorn coral Acropora cervicornis (Waldner & Robertson, 1980). However, in this study, this species associated with colonies of the firecoral Millepora alcicornis, very resistant to water motion. As this hydrocoral presents a ramified hard structure, it apparently sustained the high abundance of S. adustus, as in many other reef fish species Coni et al., 2012). Moreover, wave exposure and any branching coral seem to be habitat requirements to the dusky damselfish, which should increase their vulnerability to habitat loss considering the massive mortality of Acroporid species throughout the Caribbean (Precht et al., 2002;2010). Colonies of M. alcicornis seem to be also a preferred substrate for M. chrysurus, such as on Jamaica, Florida, Bahamas and Panama (see Waldner & Robertson, 1980;Robertson, 1984), and also on Brazilian tropical reefs (Ferreira et al., 2005). This study found that highest abundance of juvenile M. chrysurus followed this association and fish may use Millepora spp. not only for shelter, but perhaps also for feeding on the soft tissues, mucus or small epibionts (Ciardelli, 1967;. In contrast, Stegastes planifrons was highly abundant in sheltered reefs, as previously noticed in other studies (Waldner & Robertson, 1980;McGehee, 1994;Dominici-Arosemena & Wolff, 2005). It was correlated especially with cover of live massive and dead corals, where its abundance increased. As past information on benthic cover is lacking one could not confirm pervasive habitat loss, however it is strong evidence that a habitat shift have occurred already for S. planifrons, as recently described by Precht et al., 2010. This author provided valuable information on the lethal consequences of habitat shifts of this species, from branching forms (Acropora spp.) to massive corals (Montastraea complex) and Porites porites, which are the main corals found in sheltered reefs of Isla Colón. Stegastes planifrons predation on living tissue and consequent extensive coral mortality, benefits the proliferation of algal gardens (Precht et al., 2010), and its association also with dead corals could be an indication of this activity by S. planifrons. Such relationship should be specifically addressed at this point.

The role of interspecific interactions
Size and aggressiveness differences are among the most important factors driving the outcomes of competition for space in damselfish (Robertson & Lassig, 1980;Robertson, 1984;Danilowicz, 1997). Damselfish of the genus Stegastes are relatively small territorial fish and some are considered extremely aggressive (Itzkowitz, 1977;Robertson, 1984;Menegatti et al., 2003;Osório et al., 2006;Medeiros et al., 2010). An invasion of territory usually results in aggressive responses towards both conspecifics and heterospecifics, including individuals two or three size bigger (Pereira & Ferreira, 2012).
Adults of S. adustus and S. planifrons are more strongly dependent on benthic algae for food than S. partitus, which commonly feeds on both planktonic and benthic material, or S. leucostictus and S. variabilis, which are more omnivorous benthic feeders (Emery, 1973;Robertson, 1984), indicating that coinciding food resource use would only be compensated with less territory overlapping. Direct causal relationships between fish abundances are hard to detect and not always clearly evident, as they are highly dependable on the fluctuations in recruitment and adult populations. However, the effects of interspecific competition on demographics and distribution were already detected in damselfish (Robertson, 1996). Stegastes planifrons, by its aggressiveness and population numbers, is known to have a negative influence on S. partitus and S. variabilis abundance, assuming an asymmetric competitive relationship (Robertson, 1996). Stegastes adustus is also abundant and aggressive (Robertson, 1996), but both species are known to show little overlap in their distribution (Waldner & Robertson, 1980;Robertson, 1984;Gutiérrez, 1998). Herein they clearly followed the same trend in relation to adults of one another. The absence of S. planifrons on LM reefs and presence in lower abundance of S. adustus in PT indicate that the latter, at some level is a more habitat generalist in shallow depths, whereas S. planifrons seems to be more specialized, occurring generally only over live massive coral colonies, as described previously. These distributions indicate that interspecific competition could have strong effects on abundances of species in shared habitats of S. adustus and S. planifrons. However, habitat preferences would be the primary factor responsible for segregation, and that is observed since very early stages of life (Gutiérrez, 1998). Furthermore, habitat partitioning by territorial herbivores  brings also light to other interactions than competition. Stegastes adustus and M. chrysurus reached higher abundances when sharing habitats at Lime Point reefs. In fact, this association was reported over three decades ago by other authors (Clarke, 1977;Itzkowitz, 1977;Waldner & Robertson, 1980;Robertson, 1984), indicating an interesting pattern of behavior. The exclusion of M. chrysurus by adult S. planifrons is usually expected, due to the high aggression of the latter (Robertson, 1984). It has been argued that M. chrysurus may benefit from the smaller-sized S. adustus for its higher ability of territory defense, which may allow then to take long distance movements for reproduction ( Sikkel & Kramer, 2006) or feeding activities.
Adults of M. chrysurus, although also associated with Millepora colonies on this study, are less dependent of shelter, attaining larger shared territories with S. adustus, and taking advantages of co-defense of their boundaries. Regardless of the fitness costs for the "host" Stegastes, this relationship indicates high stability of damselfish behavioral traits.

Further considerations
In spite of small-scale variability, similar trends were found for damselfish habitat use compared to previous studies. Although pre-recruitment processes are likely to play an important role, the high longevity of this group known to be over a decade long (Kohda, 1996;Schwamborn & Ferreira, 2002), probably confers some population stability and contributes for the maintenance of their distributional patterns over a large time span. This stability, however, may be disrupted by major environmental changes such as the mass mortality of important structuring corals during disturbances (e. g. El-Niño events, hurricanes, diseases).
At a local spatial scale the presence of preferred habitats, such as live corals, is of major importance and a massive coral cover loss (almost 80%) throughout the Caribbean in the last decades (Aronson & Precht, 2001;Gardner et al., 2003;Guzman, 2003;Edmunds & Elahi, 2007;Alvarez-Filip et al., 2009) may have resulted in lower recruitment and associated reduction in damselfish population size. For instance, S. planifrons faced decreased adult abundance after declines of Acropora spp. throughout Caribbean reefs, which represented negative consequences for secondary preferred habitats, such algae overgrowing massive corals (Precht et al., 2010).
Although the reefs studied are not yet considered algae dominated systems and past data are lacking, they have probably suffered modifications in the last decades due to natural and land based impacts. Bocas del Toro has been subjected to several anthropogenic disturbances due to the increasing tourism activity. This came together with other human impacts such as artisanal fisheries, sport fishing and the environmental stresses resulting from diving activities (Dominici-Arosemena & Wolff, 2005). Yet, we have no knowledge on how damselfishes responded in a pervasive manner to these changes.
Damselfish patterns of local distribution may be good indicators of health on coral reef systems. As well adapted species to present conditions and in a scenario of community homogenization due to overfishing, hurricanes and other stressors along the Caribbean, the abundance patterns of S. planifrons and S. adustus in Isla Colón may be an important reference point. Moreover, understanding how these fish are facing changes and how good indicators they are would be a useful tool to become aware of such environment modifications.