Larval Fish Composition , Distribution and Assemblages by Scientific Sounder from Eight Stations off Northeastern Taiwan

With the help of SIMRAD EK-400 scientific sounder, a Maruchi-type larval net was used to sample fish larvae from 8 stations of which the water · depth is 104 to 1210 m along shelf and slope off northeastern Taiwan, by R.V. Ocean Researcher 1, during 254 cruise, September; 1990. In all, 1992 fish larvae were collected which represented, at least, 54 families and 143 species from which 66 species could be named. The greatest contribution was the En­ graulidae (33.6%), the Myctophidae (33.0%) and the Leiognathidae (4.8%). The most numerous identified species were Engraulis japonicus (33.5%), Benthosema pterotum (30.9%), Leiognathus nuchalis (4.6%). Higher abundance (noJ250 m3) of E. japonicus and L. nuchalis were found in shelf areas, while B. pterotum were found in shelf and slope areas. From the nutrient (N03) transect and NOAA-9 SST, higher larval diversity was near the edge of and away from the upwelling center. Using the Bray-Curtis dissimilarity values and the UPGMA algorithm, the most abundant 10 identi­ fied species from 8 stations were classified into three species and three station groups respectively. Interpretations of larval diversity and distribution in relation to hydrography were discussed.


INTRODUCTION
The East China Sea Shelf (ECSS) off northeastern (N.E.) Taiwan is influenced by the Kuroshio Current year round.It has been known that the hydrography in this area was mainly governed by the persistent upwelling of the Kuroshio subsurface water off the shelf of N.E.
A lot of fish species inhabiting the ECSS are exploited commercially (Ann., 1990;Okamura, 1985).In addition, the shelf region off N .E. Taiwan is a traditional fishing groW1d for local fisherman.Thus, fishing pressure of pelagic and demersel fishery in this area is heavy.Because little information is available on the adult and juvenile stages of pelagic and demersel fishes, the study on the planktonic stages of fish species is important for • Department of Aquaculture, National Pingtung Polytechnic Institute, Pingtung, Taiwan, R.0.C.
The intensity of the upwelling has a great infl uence on the abundance, distribution, spawning habits, and reproductive strategies of herring-like fishes (Parri sh et al., 1983;Bakun, 1985) and many other species (Hamann et al., 1981).There have been few studies dealing with larval fish species of the ECSS off N .E. Taiwan (Chiu, l 991a;Chiu and Lee, 1991).A complete picture of larval fish composition has not yet been made so far.There are also few studies elucidating the distribution and abundance of larval fish in relation to hydrography (Chiu and Lee, 1991).Furthermore, the examination in 'among species' patterns of larval fish distribution has not been made before.Knowledge of these patterns is very important for understanding processes affecting larval survival and subsequent recruitment • (Richardson and Stephenson, 1978;Richardson et al., 1980).
This study is part of an integrated project: Keep Edge Exchange Processes (KEEP) sponsored by the National Science Council, R.0.C.The purpose of this paper is to examine the composition, distribution and assemblages of larval fish and their possible relation to hydrographic features along shelf and slope off N.E.Taiwan.
Hydrographic data, sea surface temperature (SST) and salinity of every station were recorded by a SEABIRD SBE 9/11 CTD probe.Water samples at a specific depth were taken to measure its D.O., nutrient and fl uorescence.Based on the echo trac� (Figure 2) of SIMRAD EK-400 scientifi c sounder, a Maruchi type larval net (diameter: 1.3m, length: 4.5m) fitted with 0.5mm mesh was towed obliquely at 8 specific stations.Once the echo trace was recorded, the larval net was casted to the recorded depth and then heaved obliquely.The net was towed at a speed under 2 knots.Depending on the sampling depth, the oblique tow lasted 9 to 12 minutes.The basic sampling infonnation is listed in Table 1.           .>•t�:•.f... �...,.

2-2. Treatment of samples
All larval fish samples were fixed in seawater with 5-10% formalin.Larvae were sorted from the plankton in a rotatable sorting ring.The larvae were then identified under a dissecting microscope, to the lowest taxonomic level possible, and counted.The larval fish abundance was expressed as number (no.) per 250m 3 calculated from volume filtered and the depth traversed by the net.The average volume of water sampled by each net at 8 surveyed stations was 772.5m 3 • Identifications of larvae were based on the meristics and descriptions of Chen and Huang (1985), Leis and Rennis (1982), Leis and Tmski (1989), Moser et al. (1984), Okamura (1985), Okiyama (1988), and Zhang et al. (1985).

2-3. Data analysis
A Simpson's (Simpson, 1949) diversity and evenness indice of each station, based on the number of larvae of each species caught, was.calculated.Assemblages (defined as station groups) and species association (defined as species groups) were classifi ed by a Bray-Curtis dissimilarity values (Bray and Curtis, 1956) as the correlation coefficient, clustered by the Unweighted Pair Group Method Using Arithmetic Averages (UPGMA) method (Sneath and Sokal, 1973).A log transformed values of species abundance (log(abundance+ 1)) were used for the classfication of station and species.Any taxa over 1 % of the the sample larvae was taken as a variable to classify hierarchically.Rare species were eliminated from the analysis due to their little classificatory information (Richardson et al. 1980).Prior to performing classification, the 8 stations were grouped according to their bottom depth into three main classes: coast (from 0 to llOm), shelf (110 to 300m) and slope (more than 300m).

3-2. Larval fish composition
In all, 1992 fish larvae were caught which represented at least 54 families and 143 species (Table 2).And 143 species could be assigned to 66 particular species, 51 genus, 1 subfamily and 25 families.Five larvae could not be identified due to the severe body damage.
Engraulis japonic us and Stolephorus sp. were the species within the Engrauli dae (Table 2).The Myctophidae contained 7 species, of which Benthosema pterotum dominated and Neoscopelus microchir was the second dominant (Figure 7).The Leiog nathidae comprised two species, Leiognath us nuch alis and L. ri vulatus.The Gobiidae and Sciaenidae consisted of 20 and 5 species respectively (Table 2).The larvae of the 10 most abundant species contributed more than 1.0% and were approximately 79% of TOL (Figure 7).These 10 species contributed from 1.1 % of Ni bea al biflora to 33.6% of E. japonicus to TOL.

3-3. Larval fish distribution and abundance
Table 3 exhibits the relative abundance (no./250m 3 ) of 10 major species and total larvae by depth grouped stations.It indicates that shelf stations had higher larval abundance followed by slope and coastal stations.Stations 5 l 2A and 5315 each occupied the highest and lowest abundance respectively in 8 stations.
Among the 10 species of larvae, B. pterotum existed in all stations.This species had higher abundance in slope stations than that in shelf and coastal stations (Table 3).The distribution of E. japonicus was more abundant at shelf stations than at slope and coastal stations.This similar phenomenon was also observed for the other 8 species, except Platycephalidae spl.(Figure 8) and T. lepturus.Platycephalidae spl.and T. lepturus had higher abundance in coastal stations.

3-4. Larval fish taxa and diversity
The number of tax.on by station is shown in Figure 8.The tax. on of 512A was about four times more than that of sts.6020, 6115 and 5315.Referring to the diversity and evenness indices (Figure 9), the highest indice was at st. 6115 followed by st.6014, 512A, 5323 and 5121.And the lowest indice appeared at st. 5315.

3-5. Cluster analysis of larval fish abundance
Figures 10 and 11 are the cluster analysis and two-way coincidence table.Eight station groups were delimited from an examination of the complete dendrograms and two-way co incidence table.These groups were classified into three major groups (assemblages), coast, transition (shelf) and offshore (Figure 10).The offshore group C contained sts.5323 and 5022.The transitional group B had sts.6020, 5121 and 512A.The coastal group A was comprised of sts.6115 and 6014.These three groups occupied more than 93% of TOL.Station 5315 was isolated from the A, B and C groups.Table 3.The relative abundance (no./250m 3 ) of 10 major fish larvae species from 8 surveyed stations on OR254 cruise, September 1990 Major species     The classification method categorized the 10 major species from 8 stations into three species groups (Figure 11).The transitional g roup 2 consisted of three species, namely, B.
In this study, the Engraulidae and Myctophidae dominated within 8 surveyed stations (Figure 6) which agrees with the previous study in the East China Sea and its adjacent regions (Hattori, 1964;Ozawa and Tsukahara, 1971) _and in the area of southern Japan (Ida, 1972;K.idachi , 1983;Minami annd Tamaki, 1980).It is known that Engraulidae constitutes the major contribution in the pelagic area of inshore waters (Minami and Tamaki, 1980;Wu 1989).My study also exhibites the same result that the Engraulidae has the largest proportion, 33.5% of TOL (Figure 6).
The contribution of Myctophidae (32.9%) is next to Engraulidae (Figure 6).Chiu and Lee (1991) also found the.Myctophidae dominated along shelf and slope of N.E.Taiwan.This fish is widely spread in the Kuroshio Current and the area influenced by the Current.With vertical migration, Myctophidae was present everywhere in the water column of the ';!.Kuroshio water (Kawaguchi, 1974;Tzeng, 1989).Therefore , this fish is often used as an indicator species of Kuroshio water.
The result indicates that thirteen families occupy 87.0% of TOL offN.E.Taiwan.Within the same area, Chiu (1991a) studied the diurnal depth change of ichthyoplankton at one shelf station and fowid that 11 families accounted for over 72.0% of the total larvae in the nighttime mid-water collection and 8 families for over 49.0% in the daytime mid-water collection.Chiu (1991a) also identified 93 species comparing to 143 species (Table 2.) of my study.This different result mainly originated from the different sampling stations and schemes.Since Chiu's (1991a) study is the only available reference at present, it has valuable information to accompany my study to generate a large picture of the major families off N.E.Taiwan.As to the easteni waters of Taiwan, infl uenced by the-Kuroshio water, Tzeng (1989) showed that 7 families accounted for over 53.0%, while Chiu (199lb) fowid that 4 families accounted for over 50.0%.
The overall larvae distribution patterns (Table 3 and Figure 5), higher abundance in shelf stations and lower abundance in coastal stations, reflect the distribution of productivity in the study area.The distribution patterns of 10 major identified larvae are the most interesting, which suggest the hydrographic influences are involved.
Based on nutrient (Figure 4) and SST (Figure 5), sts.6115, 6014, 512A, 5121 and 5323 could be separated into two types.One included st.5323 and sts.512A and 5121 in the northern and southern edge of upwelling center.The other, sts.6115 and 6014, was away from the upwelling center.
Except for T. lepturus, other 9 major larvae occurred near the edge of the upwelling center.This result is consistent with the result found in the other upwelling regions (Hamann et al. 1981).In the Mauritanian upwelling area, higher concentration of fish larvae occ urred above the slope and above outer shelf areas (Hamann et al. 1981).Off California, most En graulis mordax larvae appeared in upwelling area (Ahlstrom, 1959).The other examples are E. encrasicolus larvae off North West Africa (John, 1985) and E. capensis larvae in the southern Benguela system (Shelton and Hutchings, 1982).It seems that Engraulidae larvae is more abundant near upwelling regions all over the world.Since the larval collection does not represent a record of all fish in the ecosystem, it is necessary to calculate larval diversity to reflect the complete ecosystem (Frontier, 1985).I found that st.6115 has the highest larval diversity but less tax.onnumber (Figure 8).Further, st.5121 occupies the largest number of taxon and less diversity (half of st.6115).This result is in accordance with Simpson (1949), Frontier (1985) and Peet (1974) that higher index indicates fewer species and a simpler ecosystem and vice versa.
Larval fish diversity has been related to hydrological conditions.Olivar (1987) reported that high larval diversity seemed to be related to hydrological stability, and low larval diversity was related to upwelling centers.Although this study did not collect fish larvae and CTD data at upwelling center, the hydrographic conditions are less stable at upwelling center than at the edge of and away from the center.Under this circumstance, the fact of higher larval diversity found in this study is consistent with the result of Olivar (1987).
Moreover, sts.5315, 6020, and 5022 all have less larval diversity.From numerical classification (Figure 10), I find that st.5315 was isolated from the identified three assem blages.In addition, this station has the lowest abundance.That may partly explain why st.5315 has the lowest larval diversity.As to low diversity at sts. 6020 and 50 22, there is no explanation but it may be attributed in part to the spawning areas and the current circulation.
It is interesting to note that these two stations are grouped together as slope stations.
Clustering analysis of the 10 most abundant larvae from 8 stations (Figure 11) suggests that three geographically distinct species groups exist in ECSS off N.E.Taiwan and their occurrence reflects the distribution patterns of the larval fish.The transitional group 2 is presented by E. japonicus, B. pterotum and L. nuchalis.The former two species appeared in all stations except for the absence of E. japonicus at sts. 6020 and 6115.This result is in accordance with the previous study (Hattori, 1964;Kawaguchi et al. 1972;Oawa and Tsukahara, 1971).According to depth category, B. pterotum and E. japonicus (Table 3) are more abundant at slope than shelf stations, while L. nuchalis is more abundant at shelf and slope stations.The result also reveals that B. pterotum occupied >14.0% of its total abundance at shelf stations.In other words, the contribution of this species in shelf area shall not be ignored.
myops are more abundant in shelf areas than in coastal and slope areas (Table 3).When reaching market size, N. albiflora, C. joyneri, A. argentatus and T. myops are all

Fig. 2 .
Fig. 2. •Echo trace of 8 surveyed stations from SIMRAD EK-400 scientific finder, September 1990.(Arror indicated the beginning of larval net cast) 5323 and 5022).The T-S diagram of 8 stations is shown on Figure 3.Most stations had high SST (> 25°C) and low salinity (< 34.10% ), except slope stations 5022 and 5323.The surface salinity fluctuated but exhibited an increasing trend toward deep water, which indicated the likely influence by the intrusion of warm and saline Kuroshio water.Among 8 stations, st.5323 had the lowest SST and highest surface salinity.

Fig . 5 .
Fig . 5. NOAA-9 satellite remote sensing of SST on September 19, 1990 in KEEP area (Courtesy of Taiwan Fisheries Research Institute).

Fig. 9 .
Fig. 9.The number of tax.on and the diversity and eveness indices in 8 stations.
. Summary of the cluster analysis results for 10 major larvae species from 8 stations during OR254 cruise, September, 1990, with emphasis on the distribution of species groups within a station group.The symbols in the two-way table summary represent the percentages of a species groups which occ ur in station groups; thus the percentages for a station group (row) sum to 100%.