Variations in fish body and scale shape among Aphanius dispar (Cyprinodontidae) populations: insights from a geometric morphometric analysis

[Research]

Variations in fish body and scale shape among Aphanius dispar (Cyprinodontidae) populations: insights from a geometric morphometric analysis

A. Khosravi, M. Shaker Golmakan, A. Teimori*

Department of Biology, Faculty of Sciences, Shahid Bahonar University of Kerman, Kerman, Iran.

* Corresponding author’s E-mail: a.teimori@uk.ac.ir

(Received: Sep. 06. 2016 Accepted: Feb. 25. 2017)

Phenotypic variations in fish body and scale shape were investigated among the three populations of Aphanius dispar (Rüppell, 1829)in Southern Iran through the use of landmark-based geometric morphometric analyses. This species is widely distributed in the region, and therefore, considerable morphological variations exist among the geographically allopatric populations. Based on the Principle Component Analysis (PCA), variation in body shape of the females is prominently related to the dorsal fin region, while in the males it is related to the dorsal fin and caudal peduncle. Moreover, the shape variations in the scales are obviously linked to the tip of anterior portion of the scales, and the left and right boundaries between anterior and posterior regions of the scales. The lateral sides of the fish scales in site I are concave, while they are laterally convex in sites II and III. The observed variation seen in the fish body shape and scales among the three studied sites are probably caused by the different ecological conditions of their habitats particularly variation in water flow.

INTRODUCTION

Generally, many geographical barriers to gene flow exist for the freshwater and brackish water fishes and, therefore, most species and populations have opportunity to show natural variation (Marcil et al. 2006). As a consequence, fishes in these environments are often assumed to be more genetically heterogeneous than for example marine fishes. For that reason, the freshwater and brackish water fishes are suitable candidates to study environmental-based variation and also genetic divergence in species and populations levels (e.g. Hendry & Stearns 2004).

One of the general characteristics of non-marine environments particularly freshwater resources is their geographic fragmentation in either long distance or short isolation. Fishes that exploit different resources in terms of food, space, and habitat tend to evolve different morphologies, reflecting adaptation to these ecological differences (Alexander 1974; Marcil et al. 2006).

This is probably linked to phenotypic plasticity, which arises when the same genotype produces different phenotypes in different environments (Hutchings et al. 2004). The common tooth-carp, Aphanius dispar (Rüppell 1829) is a cyprinodontid, inhabited brackish and freshwater environments in often desert regions (Wildekamp 1993).

More especially, its principal habitats are coastal lagoons, but can inhabit inland waters and hot sulfuric springs (Wildekamp 1993; Teimori et al. 2012a-b). A. dispar shows a high tolerance to ecological changes, so that some populations occupied the man-made environments such as canals, ponds and streams within the cities in Southern Iran. In addition, owing to its high tolerance to ecological changes, it shows wide distribution ranges, with considerable morphological variations (Wildekampe 1993; Teimori et al. 2012a-b). Therefore, the intraspecific morphological variation in A. dispar is documented to be linked to genetic isolation (in phylogenetic term) owing to geographic isolation (Teimori et al. 2012a-b). Moreover, the members of the genus Aphanius are known for their well-known color pattern, which made them beingalways potential candidate to be trans-located between and within different drainage systems. If this would happen in the case of A. dispar populations, then it can be another reason rather than genetic divergence for the observed morphological variation among allopatric populations. In our recent investigation on this taxon, we collected three populations in Southern Iran with considerable morphological variability. The collected specimens belong to a single population inhabiting a natural river (site I), and two populations inhabiting man-made environments (sites II - III). In this study, these three newly - collected populations were investigated through the use of landmark-based geometric morphometric analyses to infer the pattern of phenotypic variations in fish body and scale shapes. This would be important, since body shape gives an idea to the different adaptations developed by the fish in a particular environment, and also for stock assessment of fish populations.

MATERIALS AND METHODS

Taxon sampling

A total of 104 adult individuals (SL > 25mm) belonging to three populations of A. dispar were collected from three natural and man-made environments in Southern Iran including a man-made canal within Bandarabbas City, capital of Hormuzgan Province (site I in Fig. 1), Shur brackish water River, a natural environment (site II in Fig. 1), and a man-made pool within Kahnuj City in Kerman Province (site III in Fig. 1).

Sample sizes, names and geographic coordinates of the sampling sites along with some ecological parameters of the habitats are listed in Table 1.

Table 1. Sampling sites, geographic coordinates and ecological parameters of the habitats.

Data collection

Landmark data were obtained from fish body and scales photographs taken with a Dino-Lite digital microscope AM-423X ver.2.0 connected to the stereomicroscope (Olympus SZ40). The dorsal sides of body shape of specimens were photographed. Millimeter paper was included in the images to allow the acquisition of a scaling factor after wards.

Digitizing shape information

The digitized information of the fish body shape and scales for each individual was analyzed using a landmark-based method (Rohlf & Marcus 1993). The coordinates of the landmarks for fish body shape of each individual were acquired from a dorsal (left

side) image of the fish, and for the scales acquired from anterior part using the tpsDig2 software (Rohlf 2005).

Body shape landmarks

According to Park et al. (2013), the numbers of 11 homologous landmarks for the shape bodies of both males and females were selected and defined as follow (Fig. 2A):

Landmark 1 (LM1): anterior tip of upper jaw

Landmark 2 (LM2): left-rear notch of skull immediately lateral to dorsal midline

Landmark 3 (LM3): anterior base of the first ray of dorsal fin

Landmark 4 (LM4): posterior base of dorsal fin

Landmark 5 (LM5): dorsal base of caudal fin membrane

Landmark 6 (LM6): posterior-most tip of caudal peduncle at lateral midline

Landmark 7 (LM7): ventral base of caudal fin membrane

Landmark 8 (LM8): posterior base of anal fin

Landmark 9 (LM9): anterior base of the first ray of anal fin Landmark 10 (LM10): the most ventral opening of operculum

Landmark11 (LM11): posterior edge of angular (lower jaw) bone

Fig. 1. Geographic overview and details of sampling sites in Southern Iran. Site I is a man-made canal in Bandarabbas City, site II is Shur brackish water River, and site III is a man-made pool in Kahnuj City.

Scale landmarks

According to Ibanez et al. (2007) and Requieron et al. (2012), the number of seven homologous landmarks for the scales of both males and females were selected and defined as follow (Fig. 2B):

Landmark 1 (LM1): the left lateral tip of the anterior portion of scale

Landmark 2 (LM2): the center of the anterior edge of scale

Landmark 3 (LM3): the right lateral tip of anterior portion of scale

Landmark 4 (LM4): the left boundary between anterior area with circuli and posterior area

Landmark 5 (LM5): the focus of scale

Landmark 6 (LM6): the right boundary between anterior area with circuli and posterior area

Landmark 7 (LM7): the tip of posterior portion of scale.

Fig. 2. Arrangement of the landmarks on fish body and scale. (A) Lateral image of the fish specimen with 11 digitized landmarks, and (B) scale with seven digitized landmarks.

Geometric morphometric

A Generalized Procrustes Analysis (GPA) is used in order to remove disparities and rotation (Rohlf & Slice 1990) and to overlay the configurations of the landmarks in all fish specimens and scales to a common coordinate system, and to generate a set of shape variables (Gower 1975; Rohlf & Slice 1990). The multivariate analyses as Principal Component Analysis (PCA) and Canonical Variable Analysis (CVA) were used in Morpho J ver. 1.05c (Klingenberg 2011) to show possible separation of the shape variation. The visualization of shape changes, which is explained by the Canonical Variates (CV) was carried out using Morpho J ver. 1.05c (Klingenberg 2011), by generating deformed outline drawings with the average shape as a reference to analyze the data. The pairwise testing for detecting possible sexual

dimorphism and also differences between groups was tested by Discriminant Function Analysis (DFA) using Morpho J ver. 1.05c (Klingenberg 2011). Since DFA showed no sexual dimorphism in the scales shape (p > 0.0001), therefore, the further analyses on scales were applied for both sexes together. However, because of the well-known sexual dimorphism seen in external parts of almost all the Aphanius species, analyses on fish body were applied separately for the males and females. 

To evaluate the overall pattern of morphometric relationships between the three studied populations, a UPGMA cluster analysis was performed on the matrix of shape distances (Euclidean Distances) using PAST (PAlaeontological STatis­tics, version 1.81 (Hammer et al. 2001). Moreover, Mahalanobis distance (p value < 0.0001) was applied to show the significance of means shape between the compared populations.

The effects of size on variation in shapes of the fish specimens and their scales (i.e., between populations allometry) were tested separately by multivariate regression analysis (Monteiro, 1999) of Procrustes coordinates-Pco on their size using Centroid size-Cs.

The multivariate regression analyses were performed by Morpho J ver. 1.05c (Klingenberg 2011).

RESULTS

Discriminant Function Analysis shows difference between three studied populations and between sexes. However, there are some minimal overlap regarding to scale shape between sites I vs. III and II vs. III (Fig. 3A, C).

Body shape in females

The PCA performed on lateral body shape of the females provides two principal components (PCs), which accounted for over 50% of the total variance. Morphological variation explained by PC1 is approximately 35% of the total variance, which is prominently related to the anterior tip of upper jaw (LM1), left-rear notch of the skull immediately lateral to the dorsal midline (LM2), anterior base of the first ray of the dorsal fin (LM3), posterior base of the dorsal fin (LM4) and the posterior edge of angular bone in lower jaw (LM11).

The CVA reveals that the first two CVs include 63.73% and 36.27% of the total shape variation respectively (Fig. 4A).

Shape changes associated with CV1 and CV2 are mainly due to the change in length of the body (in relation to original shape), anterior base of the first ray of dorsal fin (LM3), posterior base of dorsal fin (LM4), dorsal and ventral bases of caudal fin membrane (LMs5, LM7) and posterior and anterior bases of anal fin (LMs 8, 9) (Fig. 4A).

Regressing the independent contrast of shape onto the independent contrast of centroid size in body shape of the females for 3.01% of the variation in shape, and there is a statistically insignificant trend (P = 0.151) (Fig. 5A).

Fig. 3. Discriminant Function Analysis (DFA) on shape variables of the studied populations based on (A) lateral body shape in females, (B) lateral body shape in males and (C) scale shape.

Body shape in males

The PCA performed on lateral body shape of the males provides two principal components (PCs), which accounted for more than 40% of the total variance. Morphological variation explained by PC1 is approximately 31% of the total variance, which is prominently related to the anterior tip of upper jaw (LM1), anterior

base of the first ray in dorsal fin (LM3), posterior base of dorsal fin (LM4), posterior-most tip of caudal peduncle at the lateral midline (LM6), ventral base of caudal fin (LM7), anterior base of the first ray in anal fin (LM9), and posterior edge of angular bone in lower jaw (LM11).

Fig. 4. Canonical Variable Analysis (CVA) scatter plot (axes 1 and 2) on shape variables of the studied populations based on (A) lateral body shape in females, (B) lateral body shape in males and (C) scale shape.

The CVA of body shape reveals that the first two CVs include 76.41% and 23.59% of the total shape variation respectively (Fig. 4B). Shape changes associated with CV1 are mainly due to the change in relative length of body, ventral base of caudal fin (LM 7), posterior base of anal fin (LM 8), and anterior base of the first ray in anal fin (LM 9).

The shape changes associated with CV2 are mainly due to the change in anterior base of first fin ray of dorsal fin (LM 9) and posterior-most tip of caudal peduncle at the lateral midline (LM 6) (Fig. 4B).

Regressing the independent contrast of shape onto the independent contrast of centroid size in body shape of the males for 5.65% of the variation in shape, and there is a statistically insignificant trend (P=0.006) (Fig. 5B).

Scale shape variation

The PCA performed on shape of the scales provides two principal components (PCs), which accounted for more than 60% of total variance (plot not shown).

Morphological variation explained by PC1 is 48.8% of variance, which is mainly related to the left and right lateral tips of anterior portion of scale (LMs 1,3), the left and right boundary between the anterior area with circuli and posterior region (LMs 4,6).

The CVA analysis reveals that the first two CVs include 96.04%, and 3.96% of the total shape variation respectively (Fig. 4.C).Shape changes associated with CV1 are mainly related to change in the left and right lateral tips of anterior portion of scale (LMs 1,3), and focus of scale (LM5) while, the shape changes associated with CV2 are mainly related to all the landmarks with the exception of LM 1 and LM 5 (Fig. 4C). Regressing the independent contrast of shape onto the independent contrast of centroid size in scales for 5.12% of the variation in shape, and there is a statistically insignificant trend (P = 0.045) (Fig. 5C). UPGMA dendrogram with 10000 replicates based on the data set of shape information from the lateral body shape in females (Fig. 6A), lateral body shape in males (Fig. 6B) and the scale shape (Fig. 6C) show the same pattern on phenotypic relationships.

In three dendrograms, populations from sites II and III are sister, and these together are sister to site I.

Fig. 5. Bivariate graphs derived from the multivariate regressions performed from the Procrustes coordinates (Pco) against the Log-transformed centroid size (Log-Cs). (A) Lateral body shape in females, (B) lateral body shape in males, and (C) scale shape.

Fig. 6. Dendrogram obtained from the UPGMA data set shape information from of the (A) lateral body shape in females, (B) lateral body shape in males, and (C) scale shape. The dendrogram used Euclidean distances between group means (branch bootstrap support shown as percentage at the nodes, 10000 replicates).