An approach to unraveling the coexistence of snappers (Lutjanidae) using otolith morphology

The sagittae otolith morphology of marine fishes has been used in many ecomorphological studies to explain certain ecological adaptations of species to habitat. Our study compares the sagittal otolith shapes of ten species of snappers (Family Lutjanidae) inhabiting the Persian Gulf. We used a morphometric analysis of the otolith measurements (length, height, perimeter, area and weight) and of the ratio between the area of the sulcus acusticus and the area of the otolith (S:O). The otolith contour was also analysed using wavelets as a mathematical descriptor. Morphological variations in the otoliths were associated with the morphology and external colouration of snappers as well as ecological traits. An analysis of the interspecific S:O ratio suggested that the highest ratios occurred in snappers inhabiting shallower waters. A categorical multivariate analysis, including morphological, ecological and otolith size factors, showed that the species adapted to dim light conditions had a greater otolith perimeter. An analysis of variance of the otolith contour revealed zones with a higher interspecific variability, although only the antero-dorsal zone showed differing patterns. Although the otolith patterns appear to have a phylogenetic component, they might also be related to diel activity rhythms or to the light conditions in the habitat. The results of the study showed that variation in otolith morphology can be used to explain the coexistence of sympatric species.


INTRODUCTION
Sensory ecology acts as the interface between processes occurring within organisms and those occurring between organisms and their environment (Weissburg 2005).Fishes have a variety of sensory receptors that enable them to glean information from their surroundings (Atema et al. 1988).Among these receptors, the inner ear is associated with balance and sound detection (Popper andFay 1993, Popper andLu 2000).Usually, fishes are classified as hearing generalists if they can detect sound frequencies no greater than 1 to 1.5 kHz; they are classified as hearing specialists if they can detect sound frequencies greater than 1.5 kHz (Popper et al. 2003).Morphologically, the inner ear of teleostean fishes is essentially formed by three semicircular canals and otolithic organs (sacculus, utriculus and lagena), within which are located the otoliths (sagitta, lapillus and asteriscus, respectively) (Assis 2003, 2005, Cermeño et al. 2006).The otoliths are acellular concretions of calcium carbonate and other inorganic salts developing over a protein matrix (Carlström 1963, Blacker 1969, Degens et al. 1969) and in close association with the sensorial macula (Platt and Popper 1981, Lychakov and Rebane 2000, Schulz-Mirbach et al. 2011).The otoliths, especially the sagittae, play an important role in inner ear functions (Platt and Popper 1981, Popper and Fay 1993, Popper and Lu 2000).Previous studies have indicated that the size of the sagittae is an adaptive factor associated with sensitivity to sound (Myrberg 1980, Montgomery and Pankhurst 1997, Paxton 2000, Cruz and Lombarte 2004).Fishes with large otoliths produce sounds and show highly developed intraspecific acoustic communication (Luczkovich et al. 1999, Holt 2002).These characteristics enable them to live in coastal and deep environments where visual and light communications are less important (Deng et al. 2011(Deng et al. , 2013)).Moreover, it has been reported that females can use the auditory sense to detect and locate vocalizing males during the breeding season and can change their hearing sensitivity depending on their reproductive status (e.g.Winn 1967, Sisneros andBass 2003).
The snappers (Lutjanidae) are a group of circumtropical fishes comprising 23 genera and 123 species (Froese and Pauly 2011).Twelve species of snappers have been identified along the Iranian coasts of the Persian Gulf and the Oman Sea (Assadi andDehgani 1997, Valinassab et al. 2010).Ecologically, snappers play an important role in near-shore systems, including mangroves, seagrass beds and freshwater streams, and in open-water habitats, inside or around reefs (Aiken 1993, Appeldoorn and Meyers 1993, Cervigón 1993, Baisre 2000, Claro et al. 2001).These habitats play different roles in development and life history by serving as daytime refuges, feeding nurseries and/or nesting areas for many species, including snappers.They also offer pre-recruits and juveniles abundant food resources, less competition with adults and less predation (Druzhinin 1970, Thayer and Chester 1989, Nagelkerken et al. 2001, Cocheret et al. 2003).Recently, Sadighzadeh et al. (2012) demonstrated that otolith shape descriptors and morphometrics are useful for discriminating among Lutjanus species in the Persian Gulf.In this study, a novel methodology for analysing otoliths based on outline sections is developed.

Sampling
Juvenile (close to the size of first maturity, according to the literature) and adult fishes were collected with bottom traps from January 2010 to December 2011 in the Persian Gulf commercial fishery (Fig. 1).A total of ten species of snappers Lutjanus spp.were collected and measured (total length, TL in cm).The sagittal otoliths were removed, washed, dried and stored in labeled plastic vials.Otoliths from the left side of the fish were oriented with the inner side (sulcus acusticus) up and digitized using a microscope attached to an image analyser.Large otoliths were directly digitized using a digital camera (Canon 450D with 24-105 mm lens).All images included an embedded millimeter scale (Fig. 2).

Otolith morphometry
The area (OA in mm 2 ), height (OH in mm), length (OL in mm), perimeter (OP in mm) and sulcus acusticus area (related to sensory macula area) (SA in mm 2 ) were measured using Image-Pro Plus version 4.1.0software (Media Cybernetics, Inc.).The otolith weight (OW in mg) was also obtained and included in the analysis (Table 1).Kolmogorov-Smirnov and Levene tests were used to check normality of the data distributions and variance homogeneity, respectively.The relationships between the fish length (X) and otolith variables (Y) were estimated using the power equation Y=aX b , which was log transformed to estimate a and b with a simple linear regression.A one-way analysis of variance (ANOVA) was applied to compare the slopes (b) among species using a post hoc Tukey test.A one-way ANOVA was used to compare the ratio between the sulcus acusticus area and otolith area (S:O) among species (Gauldie 1988, Lombarte 1992).In all cases, variances were unequal at the 95% confidence level.Because the assumption of equal variances was rejected, Tamhane's T2 was used as a post hoc test.The statistical analyses were performed with the SPSS statistical package (SPSS Inc. 2010).

Interaction between otolith size and environment
To test the relevance of otolith size to the ecological role of snappers in the ecosystem, a multivariate analysis was performed with a categorical principal component analysis (CatPCA) (SPSS Inc. 2010).This procedure simultaneously quantified categorical variables and reduced the dimensionality of the data.A two-dimensional plot was then created to represent the morphological similarity of the categorical variables among snappers.The similarity between the variables was assessed on a nominal and numerical scale using the categories created at data collection (Meulman and Heiser 2005)

(Table 2).
A principal component analysis (PCA) was conducted with the morphometric measurements (OA, OH, OL, OP and SA) of the otoliths from all specimens to avoid multicollinearity.First, the effect of fish size on the otolith variables was removed according to Lombarte and Lleonart (1993).The mean value of the variables for each species was then used in the PCA.Thus, the factors obtained were rescaled by dividing each observed value by the minimum value observed for that feature, yielding categorical values between 1 and 10.In addition, the following variables were also included in the CatPCA: visual field (adapted to light or dim light; species with nocturnal activity and species inhabiting turbid or deep habitats are considered species adapted to dim light conditions), environment (marine or euryhaline), depth distribution (coastal, deep or both), life history pattern (groups or primarily solitary) and visually contrasting markings (with spots on the body or lacking spots).The depth distribution was split into three categories; the remaining variables were each split into two categories.The ecological characteristics of each species are given in Table 2.

Otolith contour
The analysis of otolith shape was based on a mathematical descriptor, a wavelet (WT), related to the one-dimensional decomposition of the contour (Fig. 3).This procedure is based on expanding the contour into a family of functions obtained as the dilations and translations of a unique function known as a mother wavelet (Mallat 1991): where Ψ s is a function with a support occupying a limited range of the abscissa; choosing its shape adequately and setting a scaling parameter (s) allows the wavelet transform to detect singularities of different sizes in the function analysed.These functions describe the most prominent features of the curve (sharp transitions) in both space and wave number (Fig. 3) (Parisi- Baradad et al. 2005Baradad et al. , 2010)).To obtain the wavelets, a total of 512 Cartesian coordinates on each of the orthogonal projections of the otolith were extracted using Age & Shape software (Infaimon SL, Spain).Wavelet functions from 1 to 3 gave details of small variations of the otolith contour, whereas wavelet functions between 7 and 9 showed few contour features.Wavelet number 5 was selected as an intermediate function (Fig. 3).It was also used in a previous study to discriminate Lutjanus species (Sadighzadeh et al. 2012).
A graphical feature, the wavelet variance, was used for all species to find zones with higher variability that  could indicate different patterns in the shape of the otolith.To determine whether this variability could group the species, a cluster analysis was performed based on quadratic Euclidean distance using Ward's method.To detect significant differences between the mean functions of groups, an ANOVA test was applied based on the analysis of randomly chosen one-dimensional projections (Cuesta-Albertos and Febrero-Bande 2010).This test is implemented in the function anova.RPm in the R library fda.usc (Febrero-Bande and Oviedo de la Fuente 2011).The p-values were obtained using 1000 bootstrap

Otolith morphometric analysis
All morphometric variables of the sagittal otoliths showed a good relationship with fish length for each species, with more than 75% of the variance explained, independently of sample size.Otolith area was the variable with the strongest relationship to fish length (r 2 >0.870), whereas the variation in otolith height was more diverse among species (Table 3).The comparison of slopes showed no specific differences among species for any variables except in the case of L. rivulatus (Table 4).However, the comparisons based on the S:O ratio (Tamhane's T2 test, p>0.05) clustered the species into six groups in decreasing order of relative size (major to minor): 1) L. lutjanus, 2) L. ehrenbergii and L. fulviflamma, 3) L. fulviflamma and L. russellii, 4) L. malabaricus, L. lemniscatus and L. johnii, 5) L. erythropterus and L. rivulatus, and 6) L. argentimaculatus (Fig. 4).
The PCA reduced the otolith dimensions to two sets, OTO1 and OTO2, which were related to the otolith perimeter.The two-dimensional plot of the CatPCA analysis indicated that the first dimension was primarily influenced by environment, visually contrasting markings, the depth distribution and the otolith perimeter.The second dimension was influenced by the otolith morphometry (OTO1) and the visual field (Fig. 5).The total variance explained by the model was 65.8%, including 45.9% along the first dimension and 19.9% along the second.The increase in the depth distribution of the species was positively related to the absence of a spot (visually contrasting markings) on the body of Table 2. − Summary of ecological, functional, morphological and feeding characteristics of snappers in the Persian Gulf according to Allen (1985), Kuiter and Tonozuka (2001).the fish.The species adapted to dim light conditions and deeper distribution had a greater otolith perimeter.

DISCUSSION
The S:O ratio and otolith size are related to the hearing capabilities of marine fishes (Gauldie 1988, Montgomery andPankhurst 1997) and ecological factors such as depth distribution, fish mobility and differences in food and spatial niches (Lombarte 1992, Aguirre and Lombarte 1999, Tuset et al. 2010).Our results stressed the relevance of the sagittal otolith characteristics to the ecomorphological characteristics, showing otolith shape patterns associated with functional and ecological factors.
Several species groups of snappers are recognized on the basis of morphology and external colouration, e.g.'blue-lined', 'black spot' complex, 'yellow-lined' or 'red-lined'.These groups are congruent with phylogenetic evolution (Miller and Cribb 2007).The fishes living in shallower water have acquired a tendency to be yellowish with stripes and form aggregations to avoid large predators.They also have larger eyes and bright colour patterns favouring visual communication.The otoliths are small, most likely to avoid the background noise produced by rough seas (Paxton 2000, Volpedo and Echeverria 2003, Cruz and Lombarte 2004).In contrast, species inhabiting deeper or dimly illuminated waters have a darker colouration.Many are solitary, exhibit territorial behaviour, and possess larger otoliths (Volpedo and Echeverria 2003, Cruz and Lombarte 2004, Lombarte et al. 2010).This ecological pattern was clearly noted in the species studied, illustrating the relationship of morphology and external colouration vs. otolith size.Thus, the snappers of the 'black spot' complex and the 'yellow-lined' group (L.ehrenbergii, L. fulviflamma, L. lutjanus and L. russellii), which inhabit shallow waters (Druzhinin 1970, Kuiter andTonozuka 2001), showed the highest S:O ratio and the smallest otolith size.The clade containing the 'red-lined' and 'blue-lined' snappers (L.argentimaculatus, L. erythropterus, L. malabaricus, L. lemniscatus, and L. rivulatus), which live in deeper or dimly illuminated waters and have a dark colouration (Allen 1985), showed the lowest S:O ratio and highest otolith size.L. johnii has characteristics common to both groups.Although it should have been closer to the 'black spot' species complex, it is genetically closer to L. erythropterus (Miller and Cribb 2007).
Species inhabiting environments with a limited visual field can increase their hearing capabilities (Lombarte Thus, the diel activity rhythm facilitates coexistence between competitors extending beyond the effects of adaptation to different behavioral strategies and feeding habitats (Colmenero et al. 2010, Fox and Bellwood 2011, Azzurro et al. 2013).
The results presented here demonstrate that wavelet analysis is a very useful mathematical procedure for ecomorphological studies in addition to its use in species discrimination (Parisi-Baradad et al. 2005, 2010, Sadighzadeh et al. 2012).The identification of otolith zones with high morphological variability implies that information on shape of the whole otolith may not be necessary for the identification of stocks or species or for ontogenetic or ecomorphological studies.These findings constitute a novel approach to species discrimination.Finally, discrimination of the activity of fishes will be essential for a better understanding of ecosystem functioning and the ecological roles played by fish species (Pulcini et al. 2008, Colmenero et al. 2010, Meakin and Qin 2011, Aguzzi et al. 2013).

Fig. 1 .
Fig. 1. -Map of the Persian Gulf (NE Indian Ocean) showing the study area where snappers were collected.

Fig. 4 .
Fig. 4. -Box plots (maximum, minimum, upper and lower quartiles) for the sulcus acusticus area: otolith area ratio (S:O) for snappers from the Persian Gulf.Numbers indicate the corresponding group.

Fig. 5 .
Fig. 5. -Scatterplot of the CatPCA analysis of ecological, functional and morphological factors influencing the ecomorphological distribution of snappers from the Persian Gulf.

Fig. 6 .
Fig. 6. -Signals of wavelet 5 from the otoliths of snappers from the Persian Gulf.Colours show the similarities between signals.Fig. 7. -Graphics indicating zones with higher variability in wavelet 5. (A) variance for all species, (B) otolith contour, (C) wavelet 5 for each species.

Table 3 .
− Power relationships between fish length and otolith variables for snappers from the Persian Gulf.OA, otolith area; OH, otolith height; OL, otolith length; OP, otolith perimeter; OW, otolith weight; TL, total length.

Table 4 .
− Otolith variables presenting significant differences (Tukey's test) in the slope of relationships between fish length and otolith variables among snappers from the Persian Gulf.ns, not significant; OA, otolith area; OH, otolith height; OL, otolith length; OP, otolith perimeter; OW, otolith weight.Differences are significant (p<0.05) when otolith variables appear.