5 Geometric Morphometric Analysis of Channa striata (Striped Snakehead) Populations from Laguna de Bay, Philippines Reveals Shape Differences in Relation to Water Quality Shenna Kate M. Torres* Institute of Biology, College of Science University of the Philippines Diliman Natural Sciences Research Institute, College of Science University of the Philippines Brian S. Santos Francis S. Magbanua Institute of Biology, College of Science University of the Philippines Diliman ABSTRACT Channa striata, locally known as dalag, constitute a major aquaculture resource in Laguna de Bay. Owing to its popularity as a food source, threats such as overfishing may potentially place this species at risk. However, studies regarding its status within the lake is lacking. One way to address this gap is through population studies using geometric morphometrics. In this study, a total of 82 specimens were collected across three areas of the lake, namely, Binangonan, Calamba, and Tanay. These areas were assessed using secondary data for physicochemical parameters, which revealed significantly higher ammonium-nitrogen levels in Binangonan compared to the other areas. Geometric morphometrics was then used to determine whether shape variation existed among C. striata populations. Results showed that shape variation was greatest in the cranial region, with fish from Binangonan and Tanay having the greatest variation in shape. On the other hand, specimens from Calamba had the highest morphometric values. Lastly, these findings were then correlated with water quality data using Canonical Correlation Analysis. Results indicated that shape variation in the cranial region was correlated with differences in dissolved oxygen and pH content of the lake. The weight and length of fish were inversely correlated to the levels of ammonium-nitrogen and total dissolved solids, with specimens from Binangonan displaying a high sensitivity to ammonium-nitrogen. Keywords: dalag, freshwater fish, shape variation, Laguna Lake, physicochemical parameters * Corresponding Author Science Diliman (July-December 2020) 32:2, 5-24 Geometric Morphometric Analysis of Channa striata (Striped Snakehead) Populations 6 INTRODUCTION Laguna de Bay, located at the eastern part of Metro Manila, Philippines, is the largest lake in the country, with a surface area of 900 km2 and an average depth of 2.5 m (LLDA 2016). The lake is one of the top producers of freshwater fish, providing a source of food and income and contributing heavily to the economic growth of the country (Cuvin-Aralar 1990; Aquino et al. 2011). Among the fishes present in the lake are introduced cultured species (milkfish Chanos chanos, bighead carp Aristichthys nobilis, Tra catfish Pangasianodon hypophthalmus and Nile tilapia Oreochromis niloticus), native and other species (silvery theraponid Leiopotherapon plumbeus, Manila sea catfish Arius manillensis, gobies Glossogobius giuris, Giuris margaritacea, and striped snakehead Channa striata) (Cuvin-Aralar 2016). However, despite the management plans and policies implemented for the lake, a decrease in fish catch (Tamayo-Zarafalla et al. 2002) as well as recent fish kills (Angeles 2019; Cinco 2020) were noted to have been occurring at the lake. Various factors can be attributed to these events, such as overfishing, intensive aquaculture (Santos-Borja and Nepomuceno 2006), illegal use of destructive fishing gear (i.e., spear and drag seine) (Palma et al. 2002), rapid land use, and water pollution (e.g., organic waste, solid waste) (Kosmehl et al. 2008). These problems pose harm to the biodiversity in the area while also negatively affecting local aquaculture enterprises. Among the economically important fish species with an observed decline in catch data is the dalag (Channa striata). C. striata, commonly known as striped snakehead, is regarded to be of high economic value in the country as it contributed a total value production cost of 1,043,474.86 Philippine pesos in 2015, placing this species at the third spot among species contributing to inland water capture production (PSA 2018). It is a popular fish due to its high-quality meat, low-fat content, few intramuscular spines, tasty flavor, and relatively cheaper price compared to other fishes (Song et al. 2013). Perhaps due to the increasing demand for the species, a decline in C. striata catch data was noted by the Bureau of Agricultural Statistics (2012; 2015; 2018; 2019) from 10,469.58 metric tons in 2010 to 9,512.3 metric tons in 2017, or an average decline of 7% in nine years. Given the high value of this fish, especially as a top target fish that is prone to overfishing and population decline, it is imperative to study their population. One way to manage and assess the population of this fish is through phenotypic variation using morphometric identification. Changes in the growth and development of a fish often creates a difference in body shape within a species and may be influenced by the interplay among the environment, genetics, and selection S.K.M. Torres et al. 7 on the life history of a species (Cadrin 2000). A widely used method to study shape variation is geometric morphometrics (GM) (Santos and Quilang 2012), which is unlike traditional morphometric tools that make use of linear measurements, counts, and ratios (Adams et al. 2004) to differentiate between populations. Landmark points of a species, which are defined as the anatomical points in the species (Richtsmeier et al. 2002), are commonly used in GM. Santos and Quilang (2012) studied populations of catfish (Arius manillensis and Arius dispar) in Laguna de Bay that were observed to be in decline likely due to local overfishing. In their study, the left side body and dorsal head view of the two species were subjected to GM analysis, which revealed that most of the shape variation came from body size rather than the dorsal head of the fish, which suggests influence by various factors such as species diet, movement, and habitat. The objective of the study was to examine the shape variation between populations of C. striata found in Laguna de Bay, specifically in the northwest, south, and central regions of the lake, using GM which can be used to provide information regarding the current condition of C. striata in the lake. The different areas within the lake were chosen according to the differences in land use, specifically: the northwest bay (Binangonan), which is the closest bay to Metro Manila and is surrounded by highly urbanized communities, industrialized sites, and ports for small boats; the south bay (Calamba), which is mostly surrounded by residential areas; and the central bay (Tanay), which is mainly surrounded by agricultural areas and farmland (Johnson and Iizuka 2015). Likewise, these areas were the major landing sites (Rizal and Laguna) for C. striata found in Laguna de Bay (BAS 2018). In addition, water quality conditions in these three areas, using the physicochemical parameters measured by the Laguna Lake Development Authority, were assessed to know whether a significant change in the physicochemical parameters happened in the lake. The correlation between the shape variation found in C. striata and possible changes in physicochemical parameters were also investigated. This information can be used to reveal whether certain selective pressures within the area could favor certain phenotypic structures that could result in different morphotypes of C. striata. MATERIALS AND METHODS Specimen collection Channa striata specimens were obtained from Binangonan (14°27’47.36” N, 121°11’34.98” E), Calamba (14°12’38.93” N, 121°09’53.15” E), and Tanay (14°29’33.79” N, 121°17’17.54” E) areas of Laguna de Bay (Figure 1). A total of 82 fish Geometric Morphometric Analysis of Channa striata (Striped Snakehead) Populations 8 specimens were collected on 20 January 2018 and 17 March 2018 through the help of local fishermen around the lake. Specifically, 30 specimens from Binangonan, 27 specimens from Calamba, and 25 specimens from Tanay were collected. The fishing gear used to collect the specimens were a combination of a fish shelter set at the lake bottom that served as an aggregating site and a manual seine to catch the fish (locally called takibo) (Palma et al. 2002; 2017 personal communication with fishermen in the sampling areas). Across the three areas, the same set of fishing gear (takibo) was used by the fishermen. In addition, the total weight and length measurements (total length, body depth, and pectoral length) were measured using a weighing scale and metric ruler, respectively. Figure 1. Map of Laguna de Bay showing the three sampling areas in Binangonan, Calamba, and Tanay. S.K.M. Torres et al. 9 Water quality data The following water quality data were obtained from the Laguna Lake Development Authority (LLDA): water temperature (in °C) and total dissolved solids (TDS) (in mg/L) (from 2013-2016), dissolved oxygen (DO) (in mg/L) and pH (from 2015-2017), and ammonium-nitrogen (in mg/L) and nitrate-nitrogen (mg/L) (from 2016-2017). Each data on physicochemical parameters was specific among the three sites. The DO, pH, ammonium-nitrogen, and nitrate-nitrogen of the lake were measured monthly, while the temperature and TDS were measured yearly. Analysis on length, weight, condition factor, and water quality data The condition factor of each specimen, which was calculated by dividing the total weight (in grams) by the cube of the total length (in centimeters) and multiplying the quotient by 100, was recorded. Next, the weight and length measurements, as well as the condition factor, were subjected to one-way Analysis of Variance (ANOVA) with post-hoc tests (Tukey’s HSD) using the IBM SPSS Statistics version 26 (IBM Corporation2019). Similarly, the water quality data across the three sampling sites were subjected to one-way ANOVA followed by post-hoc tests (Tukey’s HSD) using the IBM SPSS Statistics version 26 (IBM Corporation2019). These tests were used to detect the differences in specimen morphometry and physicochemical parameters across the three sampling sites. Likewise, the condition factor was used to assess the well-being and degree of fatness of the fish (Zelditch et al. 2004). Geometric morphometrics and data analysis The left side body of each specimen was pinned in place on a white background with a standard metric ruler at the bottom in order to provide scale. Each specimen was photographed using a Canon EOS 700D DSLR Camera. The ten landmarks, serving as anatomical points, chosen for this study were based on those used by Song et al. (2013), namely: (1) anterior tip of the snout; (2) posterior aspect of the neurocranium; (3) origin of dorsal fin; (4) insertion of dorsal fin; (5) anterior attachment of dorsal membrane from caudal fin; (6) posterior end of vertebrae column; (7) insertion of anal fin; (8) original of anal fin; (9) origin of pelvic fin; and (10) posterior end of lower jaw (Figure 2). The landmarks were plotted digitally on each image using the tpsDig2 software (Rohlf 2010). The raw landmark coordinates were superimposed as shape variables using the CoordGen8 software through the Generalized Procrustes Analysis (GPA). GPA was used to ensure that differences in shape would be independent of size, position, or orientation of the fish (Slice 2007) and to check for possible outliers (Sotola et al. 2019). The corrected coordinates Geometric Morphometric Analysis of Channa striata (Striped Snakehead) Populations 10 generated from GPA were used for subsequent analysis. The centroid size, which is the square root of the sum of squared distances of the landmarks in a configuration to the average location (Slice 2007), of each specimen was calculated using the CoordGe8 software. Figure 2. Landmarks of C. striata used in the study: (1) anterior tip of the snout; (2) posterior aspect of the neurocranium; (3) origin of dorsal fin; (4) insertion of dorsal fin; (5) anterior attachment of dorsal membrane from caudal fin; (6) posterior end of vertebrae column; (7) insertion of anal fin; (8) origin of anal fin; (9) origin of pelvic fin; and (10) posterior end of lower jaw. Principal Component Analysis (PCA) was performed using the PCAGen8 software to examine overall shape variability among all specimens collected from the three localities. To test whether the observed shape variations were not dependent on the size of each specimen, Multivariate Analysis of Covariance (MANCOVA) was performed using IBM SPSS Statistics version 26 (IBM Corporation 2019). In this test, the CV 1 and CV 2 scores were treated as the dependent variable, the standard length as the covariate, and the study sites as the fixed factor (Zelditch et al. 2004). To further validate the results of MANCOVA, regression of the PC 1 scores on the logarithm of the centroid size was performed to determine the growth pattern of C. striata using Microsoft® Excel®for Microsoft 365. The MANCOVA and regression analysis were both used to identify whether C. striata exhibits isometric or allometric patterns of growth. Multivariate Analysisof Variance (MANOVA) was performed using the CVAGen8 to differentiate the three localities. Results were summarized using Canonical Variate Analysis (CVA). Deformation grids and vector plots were generated to visualize the shape variation among the population. In addition, pairwise comparisons between populations were conducted using CVAGen8 and TwoGroup 8 software. On the other hand, TwoGroup8 software was used to calculate the Goodall’s F-test. MANOVA and Goodall’s F-test were performed to detect the differences or similarities among the three populations. S.K.M. Torres et al. 11 Correlation of morphometric values to water quality data The correlation between the morphometric values of the C. striata populations and the water quality data using the physicochemical parameters measured were examined using Canonical Correlation Analysis (CCA). Separate CCA were used to describe the correlation between fish morphometric values (weight, total length, body depth, and pectoral length) and physicochemical parameters of the lake (total dissolved solids and ammonia content). The other one was done between the 10 anatomical points of the fish and physicochemical parameters of the lake (pH and dissolved oxygen). The canonical variate (CV) scores, canonical correlation coefficients, and Wilk’s test of significance were generated through the CCA package of R studio version 1.1.442 (Torres 2020). These data were used to examine whether environmental factors were correlated with the shape variation observed among C. striata populations. RESULTS Length, weight, and condition factor The descriptive statistics of the measured weight and length of each specimen from the three sampling sites within Laguna de Bay were summarized in Table 1. The measured parameters were as follows: weight ranged from 76.00 g to 1040.00 g, total fish length ranged from 20.80 cm to 52.10 cm, body depth ranged from 2.90 cm to 7.50 cm, and pectoral fin length ranged from 2.60 cm to 8.20 cm. The calculated condition factor ranged from 0.30 to 1.60, while the calculated centroid size ranged from 22.72 to 53.37. When these measurements were subjected to one-way ANOVA with post hoc tests, it was observed that of the six morphometric variables, only the condition factor (P=0.092) did not differ across the sampling sites (Table 1). Results of the post-hoc tests revealed that the total weight (P=0.003), total length (P=0.001), body depth (P<0.001), and centroid size (P<0.001) of specimens collected from Calamba statistically differ from those obtained in Binangonan and Tanay. Geometric Morphometric Analysis of Channa striata (Striped Snakehead) Populations 12 Table 1. Summary (means ± SE, F-values, and P-values) of the ANOVAs comparing water quality variables across three areas in Laguna de Bay. DO – dissolved oxygen, TDS – total dissolved solids. Ranking for post hoc tests in cases with significant effects are given. P-values <0.05 are in bold print. Parameter Binangonan Calamba Tanay F-value P-value Ranking Water temp (°C) 28.70 ± 0.08 28.50 ± 0.05 28.20 ± 0.14 2.923 0.105 DO (mg/L) 8.27 ± 0.40 8.32 ± 0.19 8.80 ± 0.42 0.752 0.474 pH 8.20 ± 0.08 8.36 ± 0.79 8.40 ± 0.10 1.403 0.251 TDS (mg/L) 385.25 ± 100.68 352.25 ± 102.90 386.50 ± 128.50 0.030 0.970 Nitrate-N (mg/L) 0.39 ± 0.11 0.22 ± 0.08 0.22 ± 0.07 1.265 0.289 Ammonium-N (mg/L) 0.13 ± 0.04 0.04 ± 0.01 0.05 ± 0.01 3.965 0.024 Binangonan>(Calamba=Tanay) Water quality The summary of data for the physicochemical parameters measured by the Laguna Lake Development Authority were summarized in Table 2. Across the sampling sites, no significant difference was found in water temperature, DO, pH, TDS, and nitrate- nitrogen (Table 2). However, ammonium-nitrogen levels across sampling sites (P=0.024) statistically differed, with a higher mean concentration in Binangonan (0.13 mg/L) compared to Calamba and Tanay (Table 2). Table 2. Summary (means ± SE, F-values, and P-values) of the ANOVAs comparing morphometric variables of C. striata across the three areas in Laguna de Bay. Ranking for post hoc tests in cases with significant effects are given. P-values <0.05 are in bold print. Morphometric Variable Binangonan (N=30) Calamba (N=27) Tanay (N=25) F-value P-value Ranking Total weight 285.63 ± 25.99 448.63 ± 53.97 294.56 ± 17.25 6.444 0.003 Calamba > (Binangonan =Tanay) Total length 30.58 ± 0.94 36.37 ± 1.54 32.43 ± 0.68 7.093 0.001 Calamba > (Binangonan =Tanay) Condition factor 0.92 ± 0.01 0.85 ± 0.04 0.85 ± 0.02 2.463 0.092 Centroid size 31.53 ± 1.00 37.04 ± 1.82 31.43 ± 0.67 6.367 0.003 Calamba > (Binangonan =Tanay) Body depth 4.13 ± 0.12 5.10 ± 0.23 4.22 ± 0.06 11.808 <0.001 Calamba > (Binangonan =Tanay) Pectoral fin length 4.43 ± 0.14 5.31 ± 0.23 3.93 ± 0.06 15.980 <0.001 Calamba < Binangonan is a University Research Associate at the Natural Sciences Research Institute, University of the Philippines Diliman working on population studies of economically important fish. She is also a member of the Molecular Population Genetics Laboratory, Institute of Biology, UP Diliman. She finished her Bachelor of Secondary Education major in Biology and Master of Science in Biology in the same university. Brian S. Santos is an Assistant Professor and a Principal Investigator from the Molecular and Population Genetics Laboratory, Institute of Biology, University of the Philippines Diliman. He received his Ph.D. in Biology from the University of the Philippines Diliman. He specializes in population genetics of commercially important fishery species. Francis S. Magbanua is an Assistant Professor and head of the Aquatic Biology Research Laboratory, Institute of Biology, University of the Philippines Diliman. He received his Ph.D. in Zoology from the University of Otago, Dunedin, New Zealand. He specializes in freshwater ecology and biomonitoring using fish and benthic macroinvertebrates.