From July 6 to September 13, 2020, 28 stations were sampled along two transects by the R/V Dayanghao and Dayangyihao. The east transect was located along 135°E, between 13°N and 28.5°N, and the west transect was located along 129°E, between 13°N and 22°N. Another 15 and 22 stations were sampled by the R/V Shenhaiyihao from January 17 to January 28 and October 27 to November 27, 2021, respectively ( Figure 1 ). Ichthyoplankton samples were collected using a WP2 plankton net (mesh size of 200 μm, mouth area 1 m²), towed vertically from a depth of 200 m to the surface at a speed of 1 m/s. Each sample was divided into two subsamples with a Folsom plankton splitter. The two stochastic sub-samples were preserved in 5% (v/v) buffered formalin-seawater solution and 95% alcohol. The larvae stored in formalin-seawater were counted and identified in the laboratory using morphological characteristics (Moser, 1996; Okiyama, 2014; Wan and Zhang, 2016). The specimens were identified to the lowest taxon possible. Damaged specimens were grouped as unidentified. Larvae stored in alcohol were used to aid the identification of the species. The counts were standardized to the density of individuals per 100 m³.

Temperature and salinity data were collected using the 911 PLUS CTD system. Content of Chlorophyll a (Chl a), net primary productivity (NPP), and suspended particulate matter concentration (SPM) (based on satellite remote sensing data with 4 km resolution) and sea level anomaly (SLA) (based on physics reanalysis data with 0.25 degrees resolution) and Mix layer depth (MLD) (based on physics reanalysis data with 0.083 degrees resolution) for August 2020, January 2021, and November 2021 were downloaded from the E.U. Copernicus Marine Service (https://marine.copernicus.eu/).

Dominant taxa were defined as those taxa for which the index of relative importance (IRI) was greater than 100 (Zhang et al., 2019a). The IRI value was calculated by the following formula:

where N% and F% represent the relative abundance and frequency of occurrence. Ichthyoplankton diversity was measured using the Margalef species richness index, Pielou’s evenness index, and the Shannon-Wiener index. The indices were calculated based on abundance data in Primer (Version 6.0, PRIMER-e, UK), and log_e was used in diversity index calculations. All calculations were at the species level; samples not identified to the species level were treated as a species unless otherwise specified.

One-way ANOVA in R was used to analyze differences in abundance between summer, autumn, and winter in the same sampled areas (south of 18°N on the east transect). To clarify the spatial distribution of the assemblages in summer and autumn, cluster analysis based on Bray-Curtis similarity with log (x + 1) transformed abundance data was performed in Primer (Version 6.0, PRIMER-e, UK).

The environmental variables that were found to influence the fish larvae assemblages were selected to analyze the community-environment relationships in the Philippine Sea. These were the average temperature from 0 to 200 m (T_0-200), average salinity from 0 to 200 m (S_0-200), Chl a, NPP, MLD, SPM, and SLA. Canonical correspondence analysis (CCA) was used to explore the effects of spatial/environmental variables on ichthyoplankton assemblage based on the maximum gradient length of detrended correspondence analysis (DCA) being 5.04. CCA was operated in CANOCO software (Version 5.0, Wageningen University & Research, NL) based on the abundance of predominant species (abundance >1%). T_0-200, S_0-200, NPP, and SLA were chosen to be analyzed based on the forward selection results. The larval abundance was transformed by log (x + 1) to reduce dominance effects.

The oceanographic structures were determined by a temperature-salinity (T-S) diagram during the cruises. Stations south of 20°N were affected by the NEC, with a higher surface temperature and a higher salinity at 150–200 m ( Figures 2 , S1 ). The sea surface temperature (SST) in the Philippine Sea is warmer in summer compared to autumn and winter ( Figure S2 ). The temperature of the southern part of the Philippine Sea is higher than in the northern part, but there is a small temperature difference in summer ( Figure S2 ). The surface salinity (SSS) was relatively lower in the southern Philippine Sea, and the salinity was > 34.8 psu as deep as 100 to 200 meters ( Figures 2 , S2 ). Chl a in the surface Philippine Sea was low (< 0.15 mg/m³), and lower Chl a levels were observed in summer compared to autumn and winter ( Figure S2 ).

A total of 882 fish larvae representing 81 distinct taxa in 23 families were recorded. Most specimens (91.5%) were identified to genus or species level. Less than 5% of the specimens were unidentified due to damage caused by the trawling process. Among the 23 families, only five exceeded 1% of the total abundance, Phosichthyidae (31.4%), Myctophidae (29.8%), Engraulidae (18.3%), Gonostomatidae (9.0%), and Paralepididae (1.6%). Among these families, 10 were represented by only one taxon and were only collected once. Most larvae belonged to the order Myctophiformes, followed by the order Stomiiformes ( Table 1 ). These two orders dominate the global mesopelagic fish fauna. We collected myctophid larvae in 49 stations and identified 32 myctophid taxa (20 identified to species) belonging to 16 genera ( Table 1 ). Genus Diaphus was the most abundant, accounting for 22.9% of all myctophid larvae, followed by the genera Lampanyctus (13.1%), Hygophum (12.8%), and Ceratoscopelus (12.1%). In addition, we collected stomiiformes fish larvae in 40 stations and identified 17 stomiiformes taxa (10 identified to species) belonging to seven genera ( Table 1 ). Genus Vinciguerria was the most abundant, accounting for 74.7% of all stomiiformes fish larvae, followed by genus Cyclothone (16.8%).

According to the IRI index, Vinciguerria nimbaria, Encrasicholina punctifer, and Cyclothone sp. were the dominant taxa in the summer, accounting for 63.6% of the abundance. In the autumn, V. nimbaria and Ceratoscopelus warmingii were the dominant species accounting for 50.8% of the abundance. In the winter, Diaphus spp., D. garmani, and Sigmops sp. were the dominant taxa accounting for 43.4% of the abundance ( Tables 1 , S2 ).

The Shannon–Wiener index for the fish larvae ranged from 0–2.32 in summer, 0–2.14 in autumn, and 0–1.59 in winter. The Shannon–Wiener index for 15 stations was zero because only one species was caught in seven stations during the summer, five stations during autumn, and three stations during winter. The Shannon–Wiener index did not differ significantly among the summer, autumn, and winter based on the Wilcoxon test of the same stations sampled during the three seasons ( Figure 3A ).

The abundance of ichthyoplankton in the tropical and subtropical Philippine Sea changed between August 2020, January 2021, and November 2021. We compared the same stations during summer, autumn, and winter. Our results showed that ichthyoplankton abundance in the Philippine Sea was higher in summer than in autumn and winter, but there was no significant difference between autumn and winter ( Figure 3B ). The fish larval abundance of most stations was low (0 –45.60 ind./100 m³), and no ichthyoplankton were collected in two stations in winter ( Figure 4 ). The mean abundance of the fish larvae was 11.35 ind./100 m³ (0.68–45.60 ind./100 m³) in summer, 5.83 ind./100 m³ (0.83–29.97 ind./100 m³) in autumn, and 2.57 ind./100 m³ (0–11.85 ind./100 m³) in winter.

At the species level, V. nimbaria was the dominant fish larva in both summer and autumn. In addition, strong seasonal shifts in community structure were observed. In summer, we collected 15 species that were not recorded in autumn, for example the dominant species E. punctifer. Meanwhile, in autumn we collected 22 species that were not recorded in summer, for example, the dominant species C. warmingii. ( Figures 4 , 5 ).

The components of fish larvae assemblages changed with the seasons, and their abundances had a definite latitudinal distribution ( Figure 5 ). Totals of 98.7% and 87.5% of the larvae of V. nimbaria were collected north of 20°N during the summer and autumn, respectively, indicating that the distribution range was limited in the Philippine Sea. In summer, the relationship between the fish larval species diversity and latitude was not significant (P = 0.7686), but larval species diversity showed a significant positive correlation with latitude in autumn (P< 0.001) ( Figure 5 ). The species diversity increased along with latitude in autumn.

Significant spatial structure was recognized based on the cluster analysis. The larval assemblages corresponded well with stations sampled in the northern and southern parts of the Philippine Sea. In the summer, the cluster analysis of Bray-Curtis similarity based on the predominant species (abundance > 1%) divided the larval assemblage into three spatial groups ( Figures 6A, B ). Group NPSS I (northern part of the Philippine Sea in the summer I) and NPSS II (northern part of the Philippine Sea in the summer II) consisted of seven and two stations sampled from the northern region of the Philippine Sea, respectively. The group SPSS (southern part of the Philippine Sea in the summer) consisted of six stations sampled from the southern region of the Philippine Sea. The dominant characteristic taxa of group NPSS I were V. nimbaria and Cyclothone sp., and Cyclothone sp. was dominant in group NPSS II. The group SPSS was defined by the dominant taxa E. punctifer and D. garmani. In the autumn, the cluster analysis of Bray-Curtis similarity divided the larval assemblage into three spatial groups ( Figures 6C, D ). Group NPSA I (northern part of the Philippine Sea in autumn I) and NPSA II (northern part of the Philippine Sea in autumn II) each consisted of four stations sampled from the northern region of the Philippine Sea. The group SPSA (southern part of the Philippine Sea in autumn) consisted of five stations sampled from the southern region of the Philippine Sea. The dominant characteristic taxa of group NPSA I were V. nimbaria, and Hygohum reinhardtii of group NPSA II. The group SPSA was defined by the dominant taxa Ceratoscopelus warmingii, Diaphus sp., and Lestidium orirhtale.

The CCA analysis revealed the relationship between fish larvae and environmental factors. The first two axes explained 15.9% of the variation in species and 74.5% of the variation in species-environment ( Table S3 ). T_0-200, S_0-200, and NPP were environmental factors affecting ichthyoplankton assemblages based on the Monte Carlo tests employed in this study ( Table 2 ), and the factors T_0-200, S_0-200, and NPP explained 9.0%, 5.1%, and 4.2% variation in the ichthyoplankton assemblage, respectively. T_0-200 and S_0-200 were strongly correlated with the first axis and the second axis, respectively. T_0-200 was positively correlated with the first axis, and S_0-200 was negatively correlated with the second axis. In addition, NPP was negatively correlated with the first axis ( Figure 7 and Table 2 ).

As shown in the CCA ordination plot, the dominant larvae were mainly located at the ends of the two axes ( Figure 7 ), indicating that these species were niche specialists. The dominant larvae V. nimbaria showed a positive relationship with NPP but a negative relationship with T_0-200. The abundant larvae Diaphus garmani, and Diaphus spp. were positively related to T_0-200. Cyclothone sp. was positively related to S_0-200, and C. warmingii was negatively related to S _0-200 ( Figure 7 ). By contrast, the common species were largely situated close to the origin of the ordination plot, indicating that these species were generalists.

The abundance of fish larvae was in the range of values from the Philippine Sea (Sassa et al., 2004, Sassa et al., 2007a), and lower than the abundance in the Kuroshio waters (Su et al., 2011; Hsieh and Lo, 2019) ( Table S1 ). The dominant species V. nimbaria, E. punctifer, C. warmingii, D. garmani, Cyclothone sp., Sigmops sp., and Diaphus spp. have also been reported to have high abundance in the Philippine Sea and tropical and subtropical western Pacific (Sassa et al., 2004; Sassa et al., 2007a; Huang et al., 2017; Hsieh and Lo, 2019). Myctophid and stomiiform fish dominate fish assemblages in global mesopelagic fish fauna (Sutton et al., 2020). These fish spend their larval stages in the epipelagic zone. Myctophid fish are preyed upon by gadiform fish, squid, tuna, dolphins, and whales (Ohizumi et al., 1998; Phillips et al., 2001; Yamamura and Inada, 2001), and as such, they play a key ecological role in the western Pacific marine food web (Sassa et al., 2002; Catul et al., 2011). In this study, myctophids had the highest species occurrence in the study area. We collected myctophid larvae from 49 stations and identified 32 myctophid taxa. Taxa and numbers of myctophid larvae in this study were less than those found in the investigation in 1994 (Sassa et al., 2004; Sassa et al., 2007a). We believe this is because we sampled at a 0–200 m depth, while Sassa et al. towed in deeper waters. We collected stomiiform fish larvae from 40 stations. We identified 18 stomiiform taxa belonging to seven genera ( Table 1 ). Vinciguerria was the most abundant genus, accounting for 74.7% of all stomiiform fish larvae, followed by Cyclothone (16.8%). The genus Vinciguerria has always been considered to be abundant in global oceans (Dove et al., 2021). In the California Current region, the main species was V. lucetia (Moser and Smith, 1993; Gerardo et al., 2018), whereas the main species of genus Vinciguerra in the Philippine Sea in this study was V. nimbaria. V. nimbaria was reported as a dominant species in the Kuroshio region and in the eastern central Atlantic (Sassa, 2019; Dove et al., 2021).

Engraulidae was third in abundance despite being represented by only two taxa, because Engraulidae including Encrasicholina punctifer constituted 18.7% of the larvae. E. punctifer plays an important ecological role in the western Pacific as a major forage source for yellowfin tuna (Itano, 2000). We collected the E. punctifer larvae mainly in the NEC region ( Figure 4E ), and the distribution patterns of E. punctifer were generally consistent with the previous research (Ozawa and Tsukahara, 1973).

The NEC (between about 10°N and 20°N) is driven by the easterly trade winds and Ekman transport (Lukas, 2001; Hu et al., 2015), and it is one of the main currents flowing westward over the upper ocean in the western North Pacific. Based on the cluster analysis, the fish larvae assemblages were perennially divided into two spatial groups in the Philippine Sea. In this study, the latitudinal distributions of fish larvae assemblages showed distinct latitudinal variations. The boundary between the southern and northern fish larvae assemblages was around 20°N latitude, which corresponded well with the northern boundary of the NEC current. The larvae of E. punctifer, and C. warmingii are abundant in the NEC region, while V. nimbaria, and H. reinhardtii are largely distributed on the northern side of the NEC. Along the same transect, (Sassa et al., 2004) found similar distribution patterns in myctophia larvae assemblages. The myctophid larvae were divided into a Kuroshio Countercurrent assemblage and a Subtropical Countercurrent-North Equatorial Current assemblage, with a boundary of 23°N during June in the tropical Philippine Sea (Sassa et al., 2004). The same distribution patterns of latitudinal variation were also found in the phytoplankton (Han et al., 2004) and zooplankton communities (Nagai et al., 2015; Yang et al., 2017). A significant latitudinal variation of phytoplankton and zooplankton is always correlated with main water masses in the subtropical and tropical Western Pacific. Zooplanktons were reported as important prey and predators for larval fish (McClatchie et al., 2012; Nunn et al., 2012). Furthermore, Yang et al. (2017) reported the spatial patterns in zooplankton communities associated with currents and divided the zooplankton community into three assemblages related to the STCC, NEC, and North Equatorial Countercurrent in the Philippine Sea. Nagai et al. (2015) also found that chaetognaths had significant differences between subtropical areas and the NEC. The same spatial patterns indicated that fish larvae may be determined by prey availability and predation pressure from zooplankton.

In autumn, species diversity showed a significant positive correlation with latitude (P< 0.001), but the latitudinal trend was not found in summer. One possible explanation of seasonal differences in the latitudinal trends of the species richness is that along the latitude the lower surface sea temperature weakens thermal stratification in the autumn, thereby allowing upwelling to move nutrients into the upper ocean. In terms of myctophid larvae assemblages, the abundance was significantly higher in the north of the Philippine Sea than those in the southern region, and the numbers of species did not clearly differentiate between the northern region and the NEC in the summer (Sassa et al., 2004). The latitudinal pattern was different from the fish larvae in the Atlantic, which showed a decreasing trend from the equator to the temperate regions (Dove et al., 2021). More similar distribution patterns were found in NPP and zooplankton; the NPP showed an increasing trend from the equator to the temperate zones in the Philippine Sea. Sun and Wang (2017) reported that both the biomass and species number of zooplankton were significantly higher in the subtropical zones than in tropical areas. This suggests that the latitudinal gradient pattern of pelagic communities in the tropical and subtropical western Pacific Ocean may differ from that of the Atlantic Ocean at the same latitude.

The maximum current speed of the NEC at times reaches 30 cm/s near the surface (Wang et al., 2019), and the NEC bifurcates into the KC and MC on the Philippine coast (Hu et al., 2015). In this study, larvae of E. punctifer and C. warmingii, the two predominant larvae of the NEC assemblages, have also been reported as the most abundant larvae in the Kuroshio waters off eastern Taiwan island (Hsieh and Lo, 2019) and the South China Sea (Hsieh et al., 2011; Huang et al., 2017). We proposed that these fish larvae were probably transported by the NEC westward to the Kuroshio waters off eastern Taiwan island and the South China Sea. The fish larvae transported by ocean currents were common; the Kuroshio Current carried the fish larvae to the Kuroshio-Oyashio transition region (Sassa et al., 2002; Sassa and Kawaguchi, 2004), and the Japanese eel larvae were transported by the NEC and KC to east Asia (Tsukamoto, 2006). Meanwhile, larvae of V. nimbaria were collected in the Kuroshio waters off eastern Taiwan island (Hsieh and Lo, 2019) and the South China Sea (Hsieh et al., 2011). However, in the Philippine Sea, larvae of V. nimbaria were distributed on the northern side of the NEC. This suggests that V. nimbaria in the Kuroshio waters off eastern Taiwan island and the South China Sea were not transported from the NEC, and V. nimbaria as a cosmopolitan species may spawn in their own waters (Dove et al., 2021). Based on the abundance difference and the predominant fish larvae, a significant seasonal difference in the Philippine Sea fish larvae assemblages was observed in this study. These seasonal changes for most taxa suggest that each species may have a fixed spawning period in the Philippine Sea. Seasonal changes in ichthyoplankton assemblages in coastal and estuarine water are relatively common (Korsman, 2013; Zhang et al., 2019a; Zhang et al., 2019b; Jiang et al., 2021). Seasonal differences were also found in the California Current (Loeb et al., 1983; Moser and Smith, 1993; Peiro-Alcantar et al., 2016) and Kuroshio Current regions (Moku et al., 2003; Sassa et al., 2007b; Sassa and Hirota, 2013). Gonadal maturation and spawning of fish depend on the appropriate external environment to ensure that larvae are produced when conditions are most suitable for their survival.

Because the distribution of ichthyoplankton assemblages is affected by environmental factors, most research has shown that the temperature, salinity, content of Chl a, zooplankton, currents, and water mass are correlated with the distribution and structure of ichthyoplankton (Zhang et al., 2019a; Dove et al., 2021; Hou et al., 2021; Jiang et al., 2021). We built the relationship between ichthyoplankton assemblage structure with environmental factors by CCA analysis. Our results revelated that ichthyoplankton in the Philippine Sea were correlated with T_0-200, S_0-200, NPP, and SLA.

Temperature affects the development and metabolism of larvae; larval daily growth rates increased by approximately 0.01 per °C increase in temperature (Houde, 1989). The optimum temperature range for fish larval survival is narrow, and the lethal limits are broad (Ehrlich and Muszynski, 1982; Dove et al., 2021). Therefore, the distribution of larvae is associated with temperature changes. In this study, the temperature variation was associated with large-scale latitude change. As shown in the CCA ordination plot, larvae of D. garmani and E. punctifer showed a positive relationship with T_0-200, indicating that the two fish larvae showed a preference for warmer oceans. Ocean warming as the greatest impact of climate change has been reported to influence the ichthyoplankton in abundance, metabolism, and geographical distributions (Hsieh et al., 2009; Asch, 2015; Walsh et al., 2015; Zhang et al., 2019c). For example, in the southern California region, both warm and cold-water taxa showed a significant increase in abundance from the cold to warm period based on 50-year-long ichthyoplankton investigation surveys (Hsieh et al., 2009). In the present study, temperature was a key factor in structuring the fish larvae assemblages in the Philippine Sea. Accordingly, the temperature-dependent fish larvae would migrate toward higher latitudes as global warming.

The mesoscale eddy modulates the abundance and distribution of marine organisms (Bertrand et al., 2014). Eddy processes play an important role in controlling ocean productivity and the timing of phytoplankton blooms (Mahadevan et al., 2012). Godø et al. (2012) studied the impact of mesoscale eddies on the patchiness of biomass and found that they provided rich feeding habitats for higher trophic marine life. Meanwhile, pelagic predators increased inside anticyclonic eddies along with the mesopelagic prey abundance in anticyclone cores (Arostegui et al., 2022). Mesoscale eddies could move nutrient-rich deep water into the surface layer, promoting primary productivity and supporting more fish larvae (Falkowski et al., 1991). The mesoscale eddy could create greater prey availability for Hygophum reinhardtii, as it has a positive relationship with SLA. In addition, the mesoscale eddy is an important dispersal pathway and as such determines the distribution of fish larvae (Tiedemann et al., 2018). The Philippine Sea is one of the most mesoscale eddy-energetic regions (Yang et al., 2013). Most eddies propagate westward, and the westward speed can reach 19.2 cm/s (Yang et al., 2013). SLA as an explicit expression of a mesoscale eddy, was another factor determining the distribution of fish larvae in this study. Mesoscale eddies may be involved in the transport of the fish larvae and the determination of the distribution.

The common species had a broader degree of environmental adaptation than the dominant species in the present study. Previous studies have generally focused on the effects of environmental factors on the fish larvae community, but environmental adaptation was often ignored (Hsieh and Lo, 2019; Zhang et al., 2019a; Dove et al., 2021; Hou et al., 2021). The dominant larvae were largely located at the ends of the two axes. By contrast, the common species were primarily situated close to the origin of the ordination plot. This suggested that dominant species had more response to the environmental factors than common species. Meanwhile, larvae of V. nimbaria showed a negative relationship with T_0-200 but a positive relationship with NPP. This pattern indicated that as a cosmopolitan species, V. nimbaria was mainly distributed in subtropical rich feeding water. Larvae of E. punctifer, Diaphus garmani, and Diaphus spp. were positively related to T_0-200, and these larvae were largely distributed in lower latitudes.

The four selected environmental factors only explained 21.3% of the variability of ichthyoplankton ( Table 2 ), indicating that more factors should be considered. Because of the dynamics of fish larvae assemblages and marine ecosystems, the relationship between the environment and fish larvae is extremely complex. More environmental data and fish larvae samples are needed to improve our understanding of ichthyoplankton in the Philippine Sea.