INTRODUCTION
The Tripterygiidae family with 29 genera and around 171 species (Eschmeyer & Fong, Reference Eschmeyer and Fong2014) is distributed in temperate and tropical regions of the Atlantic, Indian and Pacific Oceans. Several species of triplefins are used in the aquarium fish trade. Furthermore, they are strongly site-associated and guard benthic eggs, characteristics that may make populations vulnerable to habitat impacts (Baker, Reference Baker and Baker2009). Their taxonomic identification is difficult due to their similarities in gross morphological and meristic characters (Gon, Reference Gon, Gon and Heemstra1990; Cancino et al., Reference Cancino, Farías, Lampas, González and Cuevas2010). Therefore, the use of otoliths could provide a proper tool for the identification of the species of this family. Moreover, otoliths are often found in the stomach content of various organisms as well as in fossil sediments, thus being a very useful tool for taxonomic, ecological and paleontological studies (Wirtz, Reference Wirtz1976; Schwarzhans, Reference Schwarzhans1980; Schwarzhans & Grenfell, Reference Schwarzhans and Grenfell2002; Reichenbacher et al., Reference Reichenbacher, Sienknecht, Küchenhoff and Fenske2007). Otoliths are complex polycrystalline structures composed of calcium carbonate (approximately 96%) and trace elements immersed in a protein matrix (Campana et al., Reference Campana, Thorrold, Jones, Gunther, Tubrett, Longerich, Jackson, Halden, Kalish, Piccoli, de Pontual, Troadec, Panfili, Secor, Severin, Sie, Thresher, Teesdale and Campbell1997). These structures are located in the inner ear of fishes and have a role in hearing and maintenance of equilibrium (Popper & Zhongmin, Reference Popper and Zhongmin2000). They are enclosed in three end-organs of the inner ear in teleost fishes (Popper et al., Reference Popper, Rogers, Saidel, Sox, Aterma, Fay, Popper and Tavolga1988). The saccular otolith (sagitta) is the largest, at least in most teleost families (Schulz-Mirbach & Reichenbacher, Reference Schulz-Mirbach and Reichenbacher2006).
The morphometry and morphology of the otoliths have been widely used to identify species of other families. For example, Callicó Fortunato et al. (Reference Callicó Fortunato, Benedito Durà and Volpedo2014) have used the morphometry to identify species of mullets (Mugilidae) from the North-eastern Atlantic and Mediterranean Sea, while Tuset et al. (Reference Tuset, Lombarte and Assis2008) performed the characterization of 348 fish species using otolith morphology and morphometry. Reichenbacher et al. (Reference Reichenbacher, Sienknecht, Küchenhoff and Fenske2007) used otolith morphology and morphometry for assessing taxonomy and diversity in fossil and extant killifish (Aphanius). This paper is one of the first studies dedicated to the combined use of otolith morphology and morphometry. Otolith morphometry has also been used for the identification of fish stocks (e.g. Avigliano et al., Reference Avigliano, Martinez and Volpedo2014; Avigliano et al., Reference Avigliano, Comte, Rosso, Mabragaña, Della Rosa, Sanchez, Volpedo, del Rosso and Schenone2015a, Reference Avigliano, Villatarco and Volpedoc), and the simultaneous use of morphometry and morphology has been employed for the study of ecological patterns in fish (Volpedo & Echeverría, Reference Volpedo and Echeverría2003; Volpedo & Fuchs, Reference Volpedo and Fuchs2010; Curcio et al., Reference Curcio, Tombari and Capitanio2014). However, the studies related to the identification of species of Tripterygiidae family using otoliths are few. Chaine (Reference Chaine1956) described the saccular otolith of Tripterygion tripteronotus (Risso, 1810). Wirtz (Reference Wirtz1976) briefly described the otolith morphology of three existing members of the genus Tripterygion in the Mediterranean Sea, while Smale et al. (Reference Smale, Watsony and Hecht1995) described the morphology of the saccular otolith of two triplefin fishes (Cremnochorites capensis (Gilchrist & Thompson, 1908) and Helgogramma obtusirostris (Klunzinger, 1871)). Furthermore, Schwarzhans & Grenfell (Reference Schwarzhans and Grenfell2002) reported on the presence of otoliths of four triplefin fishes from New Zealand. Later, Jawad (Reference Jawad2007) described the morphology of the otoliths of several tripterygiids species in order to contribute to the taxonomy of the species of this family.
In the present work, we describe for the first time nine morphometric otolith indices of 41 species of the Tripterygiidae family collected mainly from New Zealand, Australia, Chile, South Africa, Mediterranean Sea and North America. This work is considered a potentially important tool for the identification of species of triplefins using saccular otoliths. The results of this study may also be important for the fossil record, taxonomic and diet studies (stomach contents of predators).
MATERIALS AND METHODS
Ichthyological material
Fish specimens are from New Zealand (several localities) (number of species = 24); Australia (Tasmania, Lizard Island, Avalon, Port Phillip Bay) (N = 9); South Africa (False Bay, Sodwana Bay) (N = 5); Chile (Quintero) (N = 1); USA (California) (N = 1); Spain (Ibiza) (N = 1). Examined material (Table 1) comes from the Museum of New Zealand (Wellington, New Zealand), Australian Museum (Sydney, Australia) and School of Biological Sciences (University of Auckland, Auckland, New Zealand). Those specimens belonging to the School of Biological Sciences were made available by Kendall Clements (University of Auckland) by means of scuba diving using slurp guns. Specimens without registration number are non-museum specimens and they are being kept in the School of Biological Sciences (University of Auckland).
Examined materials of studied triplefin. AM, Australian Museum, Sydney; NMNZ, Museum of New Zealand.
Specimens of all species were identified using the traditional taxonomic identification methods and no genetic study has been applied. The museum specimens used in this study were already identified when they have been borrowed. The non-museum specimens were identified using the following references: for Australian and New Zealand specimens, Fricke (Reference Fricke1994); for Chile, Fricke (Reference Fricke1997), for South Africa, Holleman (Reference Holleman, Smith and Heemstra1986) and Fricke (Reference Fricke1997), for Spain, Carreras-Carbonell et al. (Reference Carreras-Carbonell, Pascual and Macpherson2007) and for USA, Allen & Robertson (Reference Allen and Robertson1994).
The animals were measured (SL; most anterior point to the posterior tip of the vertebral column) using a digital caliper (model IP54, 150 mm moisture-proof electronic digital caliper, Shenzhen Pride Instruments, Inc., China) to the nearest 1 mm. The saccular otoliths were removed by turning the ventral side of the fish upward to allow removal of the lower jaw, the gills and the hypobranchial apparatus, and to expose the base of the skull. Later, the otoliths were cleaned with 70% ethanol and stored dry in a small plastic tube.
Otolith morphometry
Scanning electron microscopy (SEM) was used to investigate right saccular otolith ultrastructure. Otoliths examined by SEM were air dried and mounted on aluminium stubs using double-sided sticky tape. When dry, the otoliths and stubs were sputter coated with gold to a thickness of 28–30 nm in a vacuum of about 40 × 10−3 Torr. Otoliths were viewed using the secondary electron image of Philips XL45 FEG at an accelerating voltage of 5.0 KV.
According to the terminology used by Avigliano et al. (Reference Avigliano, Martinez and Volpedo2014) the following morphometric variables have been determined based on the images (Figure 1): otolith length (OL, mm), otolith width (OW, mm), otolith perimeter (PO, mm), otolith surface (OS, mm2), sulcus perimeter (SP, mm), sulcus surface (SS, mm2), sulcus length (SuL, mm), cauda length (CL, mm), ostium length (OSL, mm), rostrum width (RW, mm) and rostrum length (RL, mm). These parameters were measured in all the otoliths using image processing systems (Image-Pro Plus 4.5®). Subsequently, otolith shape indices were calculated: circularity (PO2/OS), rectangularity (OS/(OL × OH)), aspect ratio (OW/OL, %), percentage of the otolith surface occupied by the sulcus (SS/OS, %), percentage of the sulcus length occupied by the cauda length (CL/SuL, %), percentage of the sulcus length occupied by the ostium length (OSL/SuL, %), rostrum aspect ratio (RW/RL, %) and percentage of the rostrum length occupied by the otolith length (RL/OL, %). CL/SuL and OSL/SuL indices were used for the first time in this work.
Generalized scheme of the inner surface of saccular otoliths of triplefins illustrating the most relevant features. OW, otolith width; RW, rostrum width; RL, rostrum length.
Data analysis
Analysis of covariance (ANCOVA) with fish size (standard length) as a covariate was carried out for each index to test the effect of size on indices (Campana et al., Reference Campana, Chouinard, Hanson, Frechet and Brattey2000; Kerr & Campana, Reference Kerr, Campana, Cadrin, Kerr and Mariani2014). ANCOVA is robust to violations of the assumption of homogeneity of variance (Olejnik & Algina, Reference Olejnik and Algina1984). All indices varied significantly with fish size (ANCOVA, P < 0.05) and they were corrected using the common within-group slope (b) for each variable on fish standard length (e.g. Longmore et al., Reference Longmore, Fogarty, Neat, Brophy, Trueman, Milton and Mariani2010; Kerr & Campana, Reference Kerr, Campana, Cadrin, Kerr and Mariani2013; Avigliano et al., Reference Avigliano, Comte, Rosso, Mabragaña, Della Rosa, Sanchez, Volpedo, del Rosso and Schenone2015a, Reference Avigliano, Velasco and Volpedob, Reference Avigliano, Villatarco and Volpedoc).
A Principal Component Analysis (PCA) was applied to identify the variables that explain the highest proportion of variability and to investigate the morphometric patterns shown between species. The selection of axes for interpretation was performed using a screen plot (Hubert et al., Reference Hubert, Rousseeuw and Verdonck2009).
A cluster analysis was performed using the unweighted pair group method with arithmetic average (UPGMA) on an Euclidean distance matrix to assess morphological dissimilarity among species. In order to estimate the good fit between similarity matrix and the dendrogram, the coefficient of cophenetic correlation was calculated. A high cophenetic correlation suggests a good fit among the similarity matrix and dendrogram. Prior to Euclidean distance calculation the data were standardized to have a mean of zero and a variance of one.
Finally a Canonical Discriminant Analysis (CDA) was performed to test the accuracy of using those indices for the identification of fish species. To determine the discriminatory importance of each index (i.e. the value of each index that contributed most to the separation of the species) across all discriminant functions, the mean discriminant coefficient was calculated using the following equation (Backhaus et al., Reference Backhaus, Erichson, Plinke and Weiber2006): Mean discriminant coefficient bj = Σ|bjk|*EAk (k = 1, k = . . . .) where bjk is the standardized discriminant function coefficient for the variable j with respect to the discriminant function k, and EAk is the proportion of the eigenvalue of the discriminant function k in relation to the sum of all eigenvalues.
Data processing was performed using SPSS 19 and INFOSTAT statistical programs.
RESULTS
The mean and range of the indices calculated are shown in Table 2 and the otolith images of all species studied are shown in Figure 2.
Left saccular otoliths of different species of triplefins studied.
Mean and standard deviation and range (minimum–maximum) of the morphological indices of 41 species of Tripterygiidae. SL, Fish standard length; OL, otolith length; OW, otolith width; PO, otolith perimeter; OS, otolith surface; SP, sulcus perimeter; SS, sulcus surface, SuL, sulcus length; CL, cauda length; OSL, ostium length; RW, rostrum width; RL, rostrum length; CI, circularity and RE, rectangularity.
Four principal components were extracted from the PCA which accounted for 77% of the total variance of the original nine morphological variables (Figure 3A, B). The first axis (PC1) explained 30% of the total variability (Figure 3A). The morphometric variables that contributed most to the formation of the spatial gradient of the PC1 scores were CL/SuL (eigenvector = 0.50), OW/OL (eigenvector = 0.46) and RL/OL (eigenvector = 0.48) (Figure 3A). The second axis (PC2) explained 19% of the total variability with the most important variables being RW/RL (eigenvector = 0.59), CI (eigenvector = −0.43) and OSL/SuL (eigenvector = −0.40) (Figure 3A). In the plot (Figure 3A), an association between CL/SuL and RL/OL indices and the species F. flavonigrum, R. whero and R. decemdigitatus were observed. The OW/OL index was associated with B. medius, F. malcolmi and N. yaldwyni (Figure 3A). In addition, RW/RL was associated with F. lapillum, F. varium, H. cunninghami and K, while CI with F. nigripenne, among other species (Figure 3A).
Biplot on the first four principal components (PC) based on nine morphological indices of 41 species of Tripterygiidae. (A) PC1 vs PC2; (B) PC3 vs PC4. The species are indicated by numbers (Table 2). RE, rectangularity index; CI, circularity index.
The third axis (PC3) explained 16% of the variability with the most important indices being SS/OS (eigenvector = 0.59) and OSL/SuL (eigenvector = −0.49) (Figure 3B). Finally, the fourth component explained 12% of the variability with the most significant variables OL/SuL (eigenvector = 0.57) and CI (eigenvector = 0.47) (Figure 3B). According to the third and fifth component, an association between SS/OS and the species B. medius, H. obtusirostris and L. haplodactylus were observed (Figure 3B). The OSL/SuL index was associated with C. gracilis and E. rufopileus (Figure 3B). In addition, OL/SuL and CI were associated with B. dorsalis and F. flavonigrum (Figure 3B).
The cophenetic correlation coefficient (UPGMA dendrogram) was 0.85, suggesting a good fit between the similarity matrix and the matrix derived from the dendrogram. The similarity analysis showed the existence of at least five main different groups (Figure 4). Group 1 is grouped by B. medius, N. caerulpunctus and the genera Cryptichthys, Gilloblennius and Ceratobregma. This subgroup is associated with higher values of the index RL/OL (>22).
UPGMA cluster based on morphological indices in 41 species of Tripterygiidae.
Group 2 is characterized by L. haplodactylus while group 3 by the studied species of the genera Apopterygion and Blennodon.
Group 4 can be divided into several subgroups and contains all species of the genera Enneapterygius, Forsterygion, Notoclinus, Ruanoho, Matanui, Trinorfolkia, Karalepis and Ucla. It shares species of the genera Bellapiscis and Notoclinops with group 1. Subgroup 4a contains E. paucifaciatus, F. nigripenne, N. segmentatus and E. gracilis, while subgroup 4b contains the other species and genera mentioned previously for group 4. Overall, the fish in group 4 (Figure 4) had otoliths with developed rostrum lengths (RL/OL > 7) except in the genera Karalepis and Ucla (RL/OL: 0–4.8). On the other hand the species of subgroup 4a showed low rectangularity (RE < 0.9) and very variable relative size of rostrum. The rostrum is absent in E. paucifaciatus (Table 2). The species of subgroup 4bii showed the lowest values of RW/RL index (<152) and the SS/OS and OSL/SuL indexes were similar for all species of this subgroup (Figure 4).
Finally high similarity was also observed for species of the genera Crocodilichthys, Tripterygion and Acanthanectes and the species H. springeri (group 5). These species showed high rectangularity (RE > 0.1) and low relative size of rostrum (RL/OL < 6.5) (Table 2) (Figure 4).
The species H. obtusirostris and Cremnochorites and Helcogrammoides genera showed low similarity in relation to the mentioned groups (Figure 4). Particularly Cremnochorites has the lowest CL/SuL index.
The Canonical Discriminant Analysis showed a separation between some species (Table 3). The CDA proved to have greater accuracy in classifying the species H. cunninghami, K. stewarti, L. haplodactylus, N. compressus, U. xenogrammus (66–100%) (Table 3). However, the percentage of correctly classified individuals was low (50%) for A. rufus, B. dorsalis, C. jojettae, F. flavonigrum, F. varium, F. gymnotum, M. bathytaton, N. caerulpunctus and bad (<50%) for the other species.
Classification matrix of the CDA. The percentages in the last column represent the classification of each species. The species are indicated by numbers, as listed in Table 2. The current classifications for the individual species are marked in bold.
Based on the mean discriminant coefficients the CL/SL was identified as the most important index followed by the OSL/SuL, OL/SuL and RW/RL indices (bj = −1.01, bj = 0.59, bj = 0.33, −0.32 respectively).
DISCUSSION
Environmental factors such as salinity, water temperature and depth have been suggested to be responsible for some inter- and intra-specific differences in, for example, sulcus area and otolith length (e.g. Lombarte, Reference Lombarte1992; Lombarte et al., Reference Lombarte, Palmer, Matallanas, Gómez-Zurita and Morales-Nin2010, Avigliano et al., Reference Avigliano, Martinez and Volpedo2014; Reichenbacher & Reichard, Reference Reichenbacher and Reichard2014). However several variables such as otolith size, rostrum and sulcus morphology are principally under genetic control in the same groups of fishes. Therefore, the taxonomic value of otoliths is well established (e.g. Gierl et al., Reference Gierl, Reichenbacher, Gaudant, Erpenbeck and Pharisat2013, Reichenbacher & Reichard, Reference Reichenbacher and Reichard2014). Because of these characteristics, otolith morphometry has been widely used to identify fish stocks (e.g. Campana & Casselman, Reference Campana and Casselman1993; Burke et al., Reference Burke, Brophy and King2008; Cañas et al., Reference Cañás, Stransky, Schlickeisen, Sampedro and Fariña2012; Avigliano et al., Reference Avigliano, Martinez and Volpedo2014), to differentiate species (e.g. Nolf, Reference Nolf and Schultze1985; Smale et al., Reference Smale, Watsony and Hecht1995; Tuset et al., Reference Tuset, Azzurro and Lombarte2011; Tuset et al., Reference Tuset, Parisi Baradad and Lombarte2013; Zhuang et al., Reference Zhuang, Ye and Zhang2014), and to describe ecomorphological patterns of species (e.g. Platt & Popper, Reference Platt, Popper, Tavolga, Popper and Fay1981; Gauldie, Reference Gauldie1988; Lombarte et al., Reference Lombarte, Olaso and Bozzano2003; Volpedo & Echeverría, Reference Volpedo and Echeverría2003; Volpedo & Fuchs, Reference Volpedo and Fuchs2010; Jaramilo et al., Reference Jaramilo, Tombaris, Dura, Rodrigo and Volpedo2014; Avigliano et al., Reference Avigliano, Villatarco and Volpedo2015c), as an environmental indicator (Nelson et al., Reference Nelson, Hutchinson, Li, Sly and Hedgecock1994; Avigliano et al., Reference Avigliano, Tombari and Volpedo2012, Reference Avigliano, Villatarco and Volpedo2015c) and to determine fossilized specimens (e.g. Wirtz, Reference Wirtz1976: Schwarzhans, Reference Schwarzhans1980; Reichenbacher et al., Reference Reichenbacher, Sienknecht, Küchenhoff and Fenske2007). Among the most commonly used indexes are rectangularity, circularity, aspect ratio, OL/SuL (Burke et al., Reference Burke, Brophy and King2008; Tuset et al., Reference Tuset, Lombarte and Assis2008; Longmore et al., Reference Longmore, Fogarty, Neat, Brophy, Trueman, Milton and Mariani2010; Cañas et al., Reference Cañás, Stransky, Schlickeisen, Sampedro and Fariña2012; Jaramilo et al., Reference Jaramilo, Tombaris, Dura, Rodrigo and Volpedo2014; Avigliano et al., Reference Avigliano, Villatarco and Volpedo2015c, among others), and recently, the RL/OL index has been widely used by various authors (Reichenbacher et al., Reference Reichenbacher, Sienknecht, Küchenhoff and Fenske2007, Reference Reichenbacher, Sienknecht, Küchenhoff and Fenske2009; Teimori et al., Reference Teimori, Jawad, Al-Kkarusi, Al-Mamry and Reichenbacher2012a, Reference Teimori, Schulz-Mirbach, Esmaeili and Reichenbacherb; Annabi et al., Reference Annabi, Said and Reichenbacher2013; Reichenbacher & Reichard, Reference Reichenbacher and Reichard2014, among others). Few studies use relationships based on the sulcus such as SS/OS (Gauldie, Reference Gauldie1988, Lombarte, Reference Lombarte1992; Avigliano et al., Reference Avigliano, Martinez and Volpedo2014, Reference Avigliano, Villatarco and Volpedo2015c; Jaramilo et al., Reference Jaramilo, Tombaris, Dura, Rodrigo and Volpedo2014; Zhuang et al., Reference Zhuang, Ye and Zhang2014)
Commonly used indexes in this paper such as OL/SuL, circularity and rectangularity were not efficient to characterize the studied species (see PCA, Figure 3). However, the variables that explain the greatest proportion of variability were those associated with rostrum morphometry (RL/OL and RW/RL), OW/OL and the sulcus (CL/SuL, OSL/SuL and SS/OS) (Figure 3), with CL/SuL and OSL/SuL being used for the first time in this paper.
The members of Enneapterygius are represented with six species in this work: E. abeli, E. atrogulare, E. rufopileus, E. ventermaculatus, E. paucifasciatus and E. gracilis. The first four are grouped in the same subclade (4bi) (see Figure 4). Enneapterygius abeli is a cryptic and benthic species that can be found on rocky or coral tropical reefs amongst shallow photic waters (Longnecker & Langston, Reference Longnecker and Langston2005). This species feeds mainly on benthic invertebrates (Longnecker & Langston, Reference Longnecker and Langston2005). Enneapterygius atrogulare is found on intertidal and subtidal areas, specifically on reef surfaces usually in weedy areas, on algal-covered rocks or on rubble (Kuiter, Reference Kuiter1993). It prefers silty habitats of upper regions usually on pylons, estuaries and harbours (Kuiter, Reference Kuiter1993) and it feeds mainly on tiny invertebrates and algae (Randall et al., Reference Randall, Allen and Steene1990). Enneapterygius rufopileus is a species that prefers cooler water and lives on large green or brown brain corals in shallow water and tidal pools (Fricke, Reference Fricke2002). It is common to find this species in beaches or rockpools with overhangs and the algae Zonaria sp. and H. banksii. Enneapterygius ventermaculus is a demersal fish of which little is known and usually inhabits depths below 1 m (Randall, Reference Randall1995). Considering that E. abeli, E. atrogulare, E. rufopileus and E. ventermaculatus presented high similarity within the subgroup 4bi (Figure 4) and that they inhabit different microenvironments, the otolith morphometry is not reflecting differential use of environments.
Enneapterygius gracilis and E. paucifasciatus were the only studied species of the genus Enneapterygius that are not in subgroup 4bi (Figure 4) however, all studied indexes except those related to rostrum morphology were similar between these species. This similarity is reflected in the dendrogram (subgroup 4a) (Figure 4). Very little is known about the habitat of these two species. Enneapterygius paucifasciatus inhabits coral reefs in depths of 2–4 m (Fricke, Reference Fricke2002), while E. gracilis occurs in shallow tidal pools (depth range 0–15 m) and seems to be associated with coralline rocks and seagrass (Fricke, Reference Fricke1994). Once again, the otolith morphometry is not reflecting differential use of environments as happens in other species (Volpedo & Echeverría, Reference Volpedo and Echeverría2003; Volpedo & Fuchs, Reference Volpedo and Fuchs2010; Curcio et al., Reference Curcio, Tombari and Capitanio2014). On the other hand, another cause could be related to a great genetic influence. The previously mentioned group was characterized in this work specially by the low size of the rostrum (RL/OL) and there is solid evidence of genetic influence in relation to the size of the rostrum. For example, Reichenbacher et al. (Reference Reichenbacher, Kamrani, Esmaeili and Teimori2009), Teimori et al. (Reference Teimori, Jawad, Al-Kkarusi, Al-Mamry and Reichenbacher2012a, Reference Teimori, Schulz-Mirbach, Esmaeili and Reichenbacherb) and Reichenbacher & Reichard (Reference Reichenbacher and Reichard2014) have observed a strong correlation between their RL index and genetic factors in species of killifishes. Moreover, the study of Vignon & Morat (Reference Vignon and Morat2010) performed with Lutjanus kasmira (Lutjanidae) also supports this hypothesis. The results on the species of Ruanoho also confirm that genetics appears to be more prominent than adaptation to environments in the otoliths of the studied family. According to genetic studies, R. whero and R. decemdigitatus are sister species (Wellenreuther et al., Reference Wellenreuther, Barrett and Clements2007) however they make a differential use of the habitat and have otoliths with similar morphometric characteristics (Figure 4).
Forsterygion is represented in our study by eight species of which seven are grouped in the same clade (Figure 4, subgroup 4b) and have common ecological features such as dwelling on the top and sides of rocks at low to medium depth (Feary & Clements, Reference Feary and Clements2006; Wellenreuther et al., Reference Wellenreuther, Barrett and Clements2007). The morphometric differences found for F. nigripenne, a member of the other subgroup (Figure 4, subgroup 4a), would seem not to respond to genetic factors because studies made with three mitochondrial genes (12S, 16S and region control) and the nuclear gene (ETS2) show high similarity within the genus (Wellenreuther et al., Reference Wellenreuther, Brock, Montgomery and Clements2010).
Forsterygion nigripenne inhabits shallow estuarine habitats. Besides, it has peculiarities on its lateral line, implying that ecologically divergent species can be caused by a process of functional adaptation with the main selective pressure being the level of background hydrodynamic activity (Feary & Clements, Reference Feary and Clements2006; Berger & Mayr, Reference Berger and Mayr1992; Wellenreuther et al., Reference Wellenreuther, Brock, Montgomery and Clements2010). These features suggest that in the case of Forsterygion nigripenne otoliths may reflect bioecological and not genetic differences.
It is interesting to consider that for example Notoclinops represented in this work with the species N. caerulpunctus, N. segmentatus and N. yaldwyni is fragmented into different groups and subgroups (Figure 4, group 1 and subgroup 4a,b). Notoclinops caerulpunctus lives at 10 m depth (Feary & Clements, Reference Feary and Clements2006) while N. segmentatus and N. yaldwyni although sister species make a differential use of the habitat (Wellenreuther et al., Reference Wellenreuther, Barrett and Clements2007). As was observed for Forsterygion, otolith morphometry seems to reflect the diversity of microenvironments used by the members of the genus Notoaclinops.
Furthermore the genera Apopterygion, Cremnochorites, Crocodilichthys and Lepidoblennius are isolated in the dendrogram (Figure 4) and could not be associated with any pattern in relation to morphometric indexes, these being very variable between them (Table 2).
This work makes it evident that the relationship between otolith morphometry and environmental or genetic factors is extremely complex and can vary between different genera and species.
The CDA allowed identification of only six of the 41 studied species. However, the power of discrimination of used indices may be underestimated due to the relatively small sizes of the collected sample.
The high number of species in the family Tripterygiidae and their wide geographic distribution make it difficult to obtain a larger number of samples. It is expected that a greater number of samples would allow more effective discrimination. However, the results presented are of great value because they make it possible to associate different morphometric indices with several species. Furthermore, the new CL/SuL and OSL/SuL indices were among the most important to discriminate species and could be evaluated for use in other groups of fishes.
In summary, this study shows for the first time a series of nine morphometric indices and high quality images. These data together with the previous explanation related to the morphological descriptions about some of the species studied in this paper (Jawad, Reference Jawad2007) result in an interesting tool for identifying some species of triplefin. This is of great value especially for palaeontological and taxonomic studies. Two new morphometric indices were also described and tested. It is highly useful to discriminate six species of triplefin. In addition the paper is a baseline for further research that needs to intensify studies aimed at separate groups of triplefin. For example, related methodologies such as the analysis of otolith edges (Parisi-Baradad et al., Reference Parisi-Baradad, Lombarte, García-Ladona, Cabestany, Piera and Chic2005) could provide tools to separate different groups of triplefin that could not be individualized in this work.
ACKNOWLEDGEMENTS
We thank Dra. Bettina Reichenbacher and the anonymous reviewer for helpful comments on the manuscript.
FINANCIAL SUPPORT
The authors thank CONICET, MINCYT, Universidad de Buenos Aires (UBACYT CC05 and UBACYT 20620110100007) and ANPCyT (PICT 2010-1372) for financial and logistic support.
