Feeding habits of four-finger threadfin fish, Eleutheronema tetradactylum, and its diet interaction with co-existing fish species in the coastal waters of Thailand

Pattani Bay and fish sampling

The major sampling in this study was conducted in Pattani Bay, Thailand. The bay covers approximately 74 km², located in the lower part of the Gulf of Thailand at Latitude 06°52′5.3″N and Longitude 101°15′0.3″E (Fig. 1). Based on the rainfall, the bay has three seasons; dry season from January to May, moderate rainfall season (southwest monsoon) from May to September, and the heavy rainy season (monsoon) from September to December. Three sampling sites (A, B and C) along the bay mouth area, locally known as threadfish fish fishing fleets, were selected.

Monthly fish sampling was conducted for 13 months from January 2021 to January 2022 at three study sites with 2 km apart in Pattani Bay. Three sets of traditional monofilament gill nets as replicates (mesh size of 4.5 cm stretched, 5 m deep, and 540-meter total length) were simultaneously employed at each site by three traditional fishing boats and left floating for 60 min before being hauled onboard. To cover all three sites, the sampling was conducted between 0500-0900. Samples of E. tetradactylum were removed from the nets, all died after attached to the nets, and preserved with ice. Ten other fish species caught by the nets together with E. tetradactylum, including Scomberomorus commerson (narrow barred Spanish mackerel), Eubleekeria jonesi (Jones’ pony fish), Hilsa kelee (kelee shad), Johnius belangerii (belanger’s croaker), Lutjanus russellii (Russel’s snapper), Otolithes ruber (tiger tooth croaker), Parastomateus niger (black pomfret), Scatophagus argus (spotted scat), Scomberoides lysan (doublespotted queenfish) and Epinephelus coioides (orange-spotted grouper), were also removed and prepared by the same methods. They were brought back to the Faculty of Science and Technology, Prince of Songkla University, sorted and identified immediately. Up to 30 individual fish per species were randomly collected from each of the three sampling sites, and preserved with 10% formalin for four days before being transferred to 70% ethanol for further analyses. This study was approved by institutional animal care and use committee, Prince of Songkla University, Thailand, and the submitted protocol met the criteria of the Exempt Determination Research (MHESI 68014/605 Ref. Ex.001/2020).

Fish sampling from other areas

Additional threadfin fish samples were directly collected from fisherman in the provinces of Satun, Nakorn Si Thammarat and Samut Prakan provinces to cover all coastal regions of Thailand. A total of up to 25 fish samples, where possible, from each province were preserved with ice and brought back to the laboratory, where they were immediately preserved with 10% formalin for four days before being transferred to 70% ethanol for further analyses (Hajisamae, 2009).

Fish/prey dimension and diet analysis

A total of 541 samples of E. tetradactylum and 251 specimens of other fish species collected from the Pattani Bay were used. Additional 140 samples of E. tetradactylum from other areas including the provinces of Satun, Nakorn Si Thammarat and Samut Prakan were included in the analysis. They were initially measured for standard length (cm) and separated into 4 size classes: juvenile fish (<13.0 cm), small-sized fish (13.1–24.0 cm), medium-sized fish (24.1–35.0 cm) and large-sized fish (>35.1 cm). Sexes were separated based on the appearances of testis for male, ovarian for female and transitional gonad for transitional sex. Average standard lengths for male, female and transitional sex fishes used for this study were 19.04 ± 8.13 cm, 29.75 ± 5.04 and 29.21 ± 2.28 cm, respectively. Additionally, vertical mouth opening (M_V) and horizontal mouth opening (M_H) of E. tetradactylum were measured. Mouth opening area (MA) of fish was calculated based on Erzini et al. (1997) and Kyritsi & Moutopoulos (2018). ${\displaystyle \mathrm{MA}=0.25\pi \left(M_V\times M_H\right).}$

Where MA = mouth opening area, M_V = vertical mouth opening, M_H = horizontal mouth opening and π is 3.14.

The surgical scissor was applied to remove the entire gut. The gut fullness index value (FL) was assessed on a scale of 0 (empty stomach) to 6 (fully distended with food) that was expressed as empty, trace, $\frac{1}{4}$ full, $\frac{1}{2}$ full, $\frac{3}{4}$ full, full, and gorged stomachs (Pillay, 1952; Hajisamae, 2009; Soe et al., 2021). As most of the fish samples were freshly caught from nature before preservation, the stomach contents were still fresh with low level (<30%) of physical digestion. They were later identified to the lowest possible taxon. To quantify the diet compositions of E. tetradactylum, a modified point method and a numerical method were applied. For the modified point method, each food component was given points (%) proportionated to its estimation to volume in the scaled petri discs or plates (Hyslop, 1980; Hajisamae, 2009; Soe et al., 2021). Larger diets were visually assessed based on the relative contribution on the scaled petri dish or plate. Small items and zooplankton were assessed under a stereo microscope. Complete food components were also simultaneously identified and counted for numerical analysis. A relative volumetric contribution (%V), relative numerical contribution of each food type (%N) and relative frequency of occurrence (%FO) were calculated. For all fish species collected from Pattani Bay, only a modified point method was applied and later converted to volumetric contribution (%V) for further analysis. Very small items such as phytoplankton component and detritus were clarified and assessed under a compound light microscope.

Three selected major diets including fish, shrimp and squid found in the stomachs of E. tetradactylum were measured for length (cm) and body height (cm). Only completely intact or remaining prey were included in measurement to assess the predator–prey relationship (Kwak, Klumpp & Park, 2015). Each specific prey had different methods of calculating the dimension. For fish, the length was based on standard length, and the height was the greatest body depth between dorsal and ventral side. Shrimp was measured by straightening out and the length calculated from the rostrum to the edge uropod part (tail fan), and the height was the depth between carapace. The length of squid was measured from the edge of the head part to posterior end of the mantle, and the height was between the largest depth of mantle. The prey dimensions were later calculated by multiplying length and the height of prey (Karpouzi & Stergiou, 2003; Paul et al., 2017).

Data analysis

Several diet attributes including average gut fullness index (FL), average number of food type (AF), vacuity index (VI), and diet breadth (Bi) were calculated to assess the contribution of each food type in the diets of different fish species (Hyslop, 1980; Hajisamae, 2009). The diet attributes were also calculated for each factor including fish size class, sex, study site and season for E. tetradactylum. Average gut fullness index (FL) is an average of the relative gut fullness of all guts examined. Dominant food type is the type of food found in the greatest proportion, based on %V data, in the guts. Vacuity index (VI) is the number of empty guts as a percentage of the total number of guts. The Index of Relative Importance (IRI) is calculated based on Pinkas (1971), Hyslop (1980) and Osman, El Ganainy & Amin (2019) and applied only for E. tetradactylum. ${\displaystyle {\mathrm{IRI}}_V=\left(\text{\%}N+\text{\%}V\right)\times \text{\%}FO}$

where IRI_V = Index of Relative Importance by volumetric, %N = percentage of numerical, %V = percentage of volume and %FO = percentage frequency of occurrence. Percentage of Index of Relative Importance (%IRI_i) is applied for E. tetradactylum and expressed as the percentage for each food group based on Cortés (1997) and Osman, El Ganainy & Amin (2019). ${\displaystyle \text{\%}{\mathrm{IRI}}_i=\frac{{\mathrm{IRI}}_i}{\sum _{i=1}^n{\mathrm{IRI}}_i}\times 100}$

where %IRI_i = percentage of Index of Relative Importance, i = number of specific food category and n = the total number of food categories.

Diet breadth (Bi) is assessed, based on %V data, by the formula of Levin’s standardized index (Krebs, 1989; Labropoulou & Papadopoulou-Smith, 1999). ${\displaystyle Bi=\left(\frac{1}{n-1}\right)\left(\left(\frac{1}{\sum _{i,j=1}^n{P}_{ij}^2}\right)-1\right)}$

where B_i = the Levin’s standard index for predator ‘i’, ‘ P_ij’ = the proportion of diet of predator ‘i’ that is made up of food ‘j’ and ‘n’ = the number of food category.

Diet overlap (C_H) is calculated to assess dietary inter-specific relationship between different fish species inhabiting Pattani Bay. Simplified Morisita-Horn index (Horn, 1966), based on %V, data is calculated for diet overlap between fish species: ${\displaystyle C_{\mathrm{H}}=\frac{2\left(\sum p_{ij}{p}_{ik}\right)}{\sum \underset{ij}{\overset{2}{p}}+\sum \underset{ik}{\overset{2}{p}}}}$ where ‘C_H’ = Morisita-Horn index of overlap between species ‘j’ and ‘k’; ‘ p_ij’ = the proportion of food ‘i’ of the total food quantity by species ‘j’; ‘ p_ik’ = the proportion of food ‘i’ of the total food used by species ‘k’. The degree of overlap was classified as low overlap (0.0–0.29), moderate overlap (0.30–0.59) and high overlap (0.60–1.00) (Langton, 1982).

Trophic level (TL) is calculated, based on %V, to assess the trophic rank of fishes based on the proportion of each prey component in fish diet using the following equation: ${\displaystyle TL=1+\sum _{i=1}^n\left(V_i\ast T_i\right)}$

where TL = trophic level, V_i = the proportion of prey “i” in the diet of fish and T_i = the mean TL of prey based on Froese & Pauly (2000).

Statistical analysis

One-way analysis of variance (ANOVA) was used to analyze whether size class, sex, season and study site affected the fullness index (FL) and average number of food type (AF) in the stomach E. tetradactylum. Once the difference is detected, Tukey’s range test was applied. The raw data were Log (X+1) transformed to reduce the non-normality prior to statistical tests. Regression analyses were conducted to analyze the relationship between fish standard length (SL in cm) with mouth opening area (MA) of E. tetradactylum, standard length with overall prey dimension and specific prey dimension, mouth area with overall prey dimension and specific prey dimension (Erzini et al., 1997; Paul et al., 2017; Kyritsi & Moutopoulos, 2018). The raw data were square root transformed prior to analysis. Cluster analysis based on Bray-Curtis similarity matrix (%) and group average linkage method was conducted to assess how E. tetradactylum and other co-existing species interacted with each other for food. A cluster dendogram was created among eleven fish species caught together with E. tetradactylum in Pattani Bay. Analysis of similarity (ANOSIM) was later used to assess the significant level of the difference between groups. Additionally, a similarity percentage (SIMPER) was used to identify type of food responsible for the formation of each cluster group on the dendogram (Hajisamae & Yeesin, 2014; Soe et al., 2021). All these statistical tests were performed by PRIMER software package version 5.0 based on %V data (Clarke & Gorley, 2001).

Diet compositions and trophic attributes of E. tetradactylum

In general, E. tetradactylum predated mainly on penaeid shrimp, (%IRI = 50.20%), fish (33.28%) and Acetes sp./shrimp post larvae (8.41%). Details of food compositions based on %N, %FO, %V and %IRI of E. tetradactyum from all sites along the coast of Thailand are shown in Table 1. The %IRI of penaeid shrimp were 76.81%, 84.70% and 70.02% in the stomachs of fish from Samut Prakan, Satun, and Nakorn Si Thammarat Provinces, respectively (Table 2). About 70% of the penaeid shrimp consumed was Fenneropenaeus merguiensis. Samples of E. tetradactylum collected from Pattani Bay fed mainly on a combination of fish (%V = 43.79%), followed by penaeid shrimp, especially F. merguiensis (35.68%) (Table 3). The prey fish were mainly Clupeiformes, Perciformes, Mugilliformes and Atheriniformes. These included Ambassis sp., Atherinomorus lacunosus, Dussumieria elopsoides, Encrasicholina devisi, Hilsa kelee, Leiognathus equulus, Planiliza subviridis, Stolephorus commersonii, Thryssa setirostris and Trichiurus lepturus. Different food compositions between size class, sex, season and site of collection were observed. Details of food compositions and attributes for all size class, sex, season and study site are illustrated in Tables 2 and 4. The main food types of different sex stages were observed. The male and female fishes fed mainly on shrimp and fish, respectively, but fish at transitional sex stage predated almost entirely on other fishes. Results of analysis of variance (ANOVA) indicated that value of fullness index (FL) varied significantly based on size class, sex, season and study site (Table 5). Average number of food type (AF) was influenced by fish size, sex and study site but not by season (Table 5). Results of post-hoc analysis are given in Table 6. The FL value of transitional sex fish was not different from male or female fishes, but the variation of AF between transitional sex fish (1.09 ± 0.30) and female (1.53 ± 0.65) was significant (p = 0.006).

Ontogenetic shift of diet compositions

Based on samples from Pattani Bay, juvenile fish fed largely on Acetes/shrimp post larvae (%IRI = 69.54%), followed by fishes (21.89%) and polychaete (4.78%) (Table 2). The small sized-fish turned its diet to penaeid shrimp (71.30%) with some proportions of fish (21.61%) and squid (1.18%). In the medium-sized fish, the reduction of penaeid shrimp was observed to 43.62% and the increment of fish and squid was found to be 43.97% and 4.40%, respectively. Importance of penaeid shrimp reduced gradually when the fish grew to larger size (24.20%), with a concurrent increase of fish and squid (30.12% and 30.34%, respectively).

Relationship between standard length and mouth area of E. tetradactylum with prey dimensions

A total of 128 fish samples were analyzed with the standard length of 5.90 cm and maximum length of 46.11 cm. The maximum and minimum mouth areas of fish sample were 32.20 cm² and 0.26 cm², respectively. The slope of regression showed a significantly positive relationship between standard length and mouth area (R² = 0.88, P < 0.001), indicating that mouth opening of the fish was positively related to the standard length (Fig. 2A).

Both standard length and mouth area showed positive relationships with overall prey dimensions (R² = 0.3281, P < 0.001; R² = 0.3594, P < 0.001), indicating that the larger fish consumed the larger prey (Figs. 2B and 2C). Results of a specific analysis on the relationship between standard length and mouth opening with prey dimensions of three major prey types (fish, squid and shrimp) are shown in Figs. 3, 4 and 5. The dimensions of the consumed prey items increased significantly for all prey types with the increase of standard length and mouth opening of the predator. This confirms that the mouth dimension of this species controls the size of all three prey types (fish, squid and shrimp).

Inter-specific diet relationship between E. tetradactylum and other co-existing species

A total of 792 fish samples from eleven fish species were examined (Table 3). Three species feeding on shrimp as the main food item included E. tetradactylum, J. belangerii and L. russellii. Fish was considered a dominant prey in the stomachs of S. commerson, O. ruber, S. lysan and E. coioides. Two species including H. kelee and S. argus fed dominantly on detritus. Zooplankton and phytoplankton were the dominant food for P. niger and E. jonesi, respectively.

Results of trophic attributes of eleven fish species including FL, VI, total food type, average number of food type and B_i are provided in Table 4. The FL ranged from 2.50 ± 0.89 for E. jonesi to 3.27 ± 0.88 for S. commerson, VI varied from 7.14 for E. coioides to 34.78 for S. commerson, total number of food type was between 5 for the three fish species to 8 for E. tetradactylum, AF was from 1.35 ± 0.57 for E. tetradactylum to 2.38 ± 0.51 for E, coioides, and B_i ranged from 0.25 for H. kelee to 0.80 for P. lysan. The TL values ranged from 1.31 for H. kelee to 4.25 for O. ruber.

Based on the data from Pattani Bay, twenty-three out of fifty-five pairs of analyses were shown to have high diet overlaps (C_H >0.60) (Table 7). The C_H values for E. tetradactylum indicated significant overlaps with six co-existing species including S. commerson, J. belangerii, L. russellii, O. ruber, S. lysan and E. coioides. The low C_H values, considering no overlap for food, were found between E. tetradactylum and E. jonesi, H. kelee, P. niger and S. argus.

The above result was also confirmed by the cluster analysis. Two trophic clusters, at 45% similarity, were formed by eleven fish species caught together with E. tetradactylum (Fig. 6). Result of analysis of similarity (ANOSIM) indicated the difference of diet composition among the two groups (P = 0.003, Global R = 0.907). The first group (G1) included O. ruber, L. russellii, E. coioides, J. belangerii, S. commerson, S. lysan and E. tetradactylum. A similarity percentage (SIMPER) indicated that teleost fish (47.8% contribution) and shrimp (44.5%) were the main food types contributing to the formation of this group. The second cluster (G2) was formed by four fish species including P. niger, E. jonesi, S. argus and H. kelee. Contribution of zooplankton, teleost fish, detritus and phytoplankton to the formation of this group were 29.6%, 25.7%, 21.0% and 17.89%, respectively.