Diel, seasonal, and annual variations of fish assemblages in intertidal creeks of the Changjiang River estuary*

Jiayi FAN, Jiasheng YANG, Yuanwen HE, Xiaodong JIANG

State Key Laboratory Estuarine and Coastal Research, School of Life Sciences, East China Normal University, Shanghai 200241, China

Abstract The use patterns of salt marsh habitats by fish assemblages were investigated in two welldeveloped intertidal creeks in the Changjiang (Yangtze) River estuary on spatial and temporal scales.Samples were collected using fyke nets at two sites during day and night in each season throughout four years.Notable changes in fish assemblages were detected over seasonal and interannual cycles, with many more marine species present in winter assemblages, whereas freshwater and estuarine species characterized the other seasonal samplings.The appearance of catadromous species in intertidal creeks mainly depended upon their specific physiological requirements, such as spawning migration.No significant diel shifts were observed in fish assemblages.This diel pattern may be a combined consequence of avian predation pressure and other drivers (i.e., inherent living behaviors, food availability, and predation by piscivorous fishes).The spatial difference in fish community was insignificant even though the two intertidal creeks were different in size.Intertidal creeks with different size could possess similar ecological value for fish fauna.Redundancy analysis (RDA) showed that biological factors exerted greater effects on fish assemblages than physical variables.Temporal variations were strongly facilitated by food availability, but negatively impacted by avian predation.Salinity, water temperature, and river runoff had weak influence on temporal variations of fish communities.The lengthfrequency distribution showed that the fish collected in these two sites mainly consisted of young-of-year and juvenile fish, which confirmed that the creeks provided important nursery and forage roles in the process of fish ontogenetic development.Further, we proposed that salt marsh conservation management should be equally applied to all well-developed intertidal creeks regardless of creek size.

Keyword: fish; temporal variation; intertidal creeks; Changjiang (Yangtze) River estuary

1 INTRODUCTION

Salt marshes are important components of coastal and estuarine ecosystems, because they attract an array of animals to a variety of complex sub-habitats,including vegetated marsh areas, non-vegetated marsh ponds and pools, intertidal creeks, and subtidal creeks(Minello et al., 2003).During periods of inundation,intertidal sections become linked with subtidal zones as a considerable amount of mass and energy are exchanged, which underpins the intense primary productivity of salt marsh systems (Novakowski et al., 2004).Many aquatic animals and migratory birds visit salt marsh creeks and utilize these intertidal networks as indispensable habitats for various life stages and physiological processes,which further highlights important nursery and forage services of these habitats (Botton et al., 1994;Kimball and Able, 2007).

Effects of diel, seasonal, and annual cycles on variations in fish assemblages have been extensively examined in many types of ecological habitats worldwide.Some studies have reported huge diel shifts in fish assemblages and more fish abundance at night in subtidal and intertidal creeks (Rountree and Able, 1993) and in nearshore estuaries (Yeoh et al., 2017); other studies, however, have provided opposite evidence, and no clear diel patterns of dominant species were recorded in sandy beaches(Pessanha and Araújo, 2003; Pessanha et al., 2003)or in seagrass beds (Ribeiro et al., 2006).Studies on seasonal variations have elucidated strong seasonal patterns of fish diversity and abundance in intertidal creeks (Koutsogiannopoulou and Wilson, 2007; Green et al., 2009).Other study, however, has demonstrated clear seasonal patterns in species richness and fish composition, but not in species abundance and biomass, in subtidal creeks and estuaries (Veiga et al.,2006).In terms of interannual variations, a number of studies have demonstrated consistently stable interannual trends in nekton communities in intertidal salt ponds, sandy shores (Lazzari et al., 1999), marsh creeks (Laffaille et al., 2000), and estuaries (Desmond et al., 2002).

The use of intertidal creeks by fish has been recognized in some field investigations, but has not been addressed for a long time.Some studies have documented temporal variations of fish community structure in intertidal habitats, including strong seasonal patterns in structuring nekton communities in intertidal creeks (Cattrijsse et al., 1994; Koutsogiannopoulou and Wilson, 2007), whereas other studies have shown no significant seasonal disparity in three African intertidal salt marsh creeks (Paterson and Whitfield, 2003), and less important diel shifts in faunal composition in unvegetated mudflats, eelgrass beds (Gross et al., 2019), and intertidal creeks(Hampel et al., 2003).A 4-year study has demonstrated little interannual changes in fish communities in intertidal salt marsh creeks (Laffaille et al., 2000).The use patterns of intertidal creeks by fish communities are inconsistent and temporal characters of fish assemblages are relatively uncertain.Therefore, it is necessary to gain more comprehensive insight into the ecological value and function of ichthyofaunal communities.

The Changjiang (Yangtze) River estuary is one of the most typical meso-tidal estuaries in the world(Zhuang et al., 2006).As this estuary empties into the junction of the Yellow Sea and the East China Sea, huge amounts of silts and sediments brought by the upper Changjiang River deposit in the mouth of the estuary (Dagg et al., 2004), and then greatly expedite the formation of various topographies of wetland habitats, including shoals, sand bars, tidal flats, tidal creeks, and salt marshes (Tian et al.,2008).Moreover, these diversified ecological zones have made them optimal destinations, allowing survival, growth, and reproduction activities of various terrestrial and aquatic fauna (Rountree and Able, 2007).A study on seasonal variations of the demersal nekton community off the Changjiang River estuary demonstrated a clear seasonal succession in species composition and confirmed the important ecological function of spawning, nursing, and overwintering for the fish community (Jiang et al.,2014).In addition to the general biophysiological functions of this estuary, it has been considered an important migration corridor for endangered fish species, which further underpinned the importance of the ecological and economical features of the Changjiang River estuary (Zhang et al., 2019).Additionally, relevant researches concerning the relationship between environmental factors and fish assemblage structure have been conducted in the Changjiang River estuary and its adjacent waters(Jiang et al., 2006; Yu and Xian, 2009).

The Chongming Dongtan wetland is a typical coastal wetland that lies in the easternmost section of the Chongming Island and borders the Changjiang River estuary, providing multiple ecological functions for various species (Sun, 2020).For migrating birds,these intertidal areas, which are located on the East Asia-Australasian flyaway, function as intermediate stopover sites for migratory waterfowls to feed and rest (Ma et al., 2002).Furthermore, for aquatic organisms, intertidal creeks are vital migration channels for nekton species when exposed to subtidal zones during inundation and play a crucial role in fish nursery and feeding activities (Quan et al., 2009; Jin et al., 2014).Additionally, similar studies on fish utilization of salt marsh intertidal creeks have been surveyed in the Shanghai Jiuduansha wetland, which is adjacent to the Dongtan wetland (Jin et al., 2007,2010).However, since the last century, a widespread invasive grass,Spartinaalterniflora, has been intentionally or accidentally introduced into numerous littoral wetlands, which altered native salt marsh plant communities, threatened aquatic and bird habitats, and reduced the ecological integrity and function (Ma et al., 2014).In the Dongtan wetland,as one of the most heavily infested areas, a dikebuilding project has been developed to eradicate this invasive plant and make the tidal wetland a favorable habitat for various species (Wang et al., 2021).Because most studies on fish community dynamics were conducted prior to the completion of the ecological dike program in 2016 (Quan et al., 2009;Jin et al., 2014; Zhang et al., 2020), it is essential to ascertain how fish assemblages utilize the intertidal wetland after enclosure at different temporal and spatial scales.

The present study aimed to: (ⅰ) investigate diel,seasonal, and annual variations in fish communities that utilize intertidal habitats; (ⅱ) test whether fish communities change at two mature intertidal creeks with different geomorphological characteristics; and(ⅲ) explore the mechanism of temporal distribution by analyzing the relationship between fish communities and relevant biological and environmental variables.

2 MATERIAL AND METHOD

2.1 Study area

In our study, two representative well-developed intertidal creeks were sampled to investigate the fish assemblages (Fig.1).Tuanjiesha creek is located at the southern section of the Dongtan marshes and is a relatively small intertidal creek, with a length of 1 727 m and a mean width of 14.8 m.Xiaonangang creek is situated in close proximity to the seawalls constructed in the northeast shoal of the Dongtan wetland and empties into the center of the river mouth.It is one of the typically largest intertidal creeks in the Dongtan salt marshes, with a length of 4 879 m and an average width of 44.4 m.Furthermore,these two intertidal creeks are core regions colonized by migratory birds to feed and rest during their migration seasons (Botton et al., 1994), and there are long-term monitoring stations for plankton dynamics,which help facilitate comparison of different organisms.The Dongtan tidal marshes are characterized by the Asian monsoon climate and typical semi-diurnal tidal pattern, with two high tides and two low tides within one diel shift, and a mean tidal range of 2.45–4.96 m from neap tide to spring tide (Xu and Zhao,2005).

2.2 Sampling strategy

In this study, field sampling in Xiaonangang and Tuanjiesha creeks was carried out in diel, seasonal,and annual periods.Diel sampling began at ebb tides as the water receded from the intertidal habitats.Seasonal sampling took place in May (spring), August(summer), October (autumn), and December (winter)in 2015.To assess interannual consistency in the fish community, sampling was conducted in May and October from 2015 to 2018.Fish were caught with fyke nets erected for three consecutive days,using a method modified from a previous study(Cattrijsse et al., 1994).The fyke net was 8 m long with 4-mm mesh and had a mouth of 1 m width×1 m height and wings of 8 m long×1 m height.The nets were placed in the middle of the creeks at the start of each sampling episode; the net mouth faced into water current and was set up at the start of the ebb tides to trap as many fish as possible (Laffaille et al., 2000).

All fish samples were fixed with 10% formalin solution at each station and then identified to species and counted in the laboratory.Weight was measured to the nearest 0.01 g and size (total length, TL) was measured to the nearest centimeter for each species(Veiga et al., 2006).All individuals were grouped into four ecological categories: marine, estuarine,freshwater, and catadromous (Paterson and Whitfield,2000).

2.3 Data analysis

The relative abundance (%N) and biomass (%B)were calculated based on the total fish catches and expressed as percentages of each species.The relative occurrence frequency (%Fo) was denoted by the percentage of the occurrence frequency of each species in the whole community.The index of relative importance (IRI) integrates the numbers of individuals, the biomass, and the occurrence of each species into a formula to provide a more accurate evaluation of the fish community (Pinkas et al., 1971).

IBI=%Fo(%N+%B).(1)

IRI together with the relative abundance (%N),biomass (%B), and frequency of occurrence (%Fo)eliminate the bias from considering a single parameter alone, and contribute to characterizing fish community structure and identifying the numerically prominent species in the community (Golani, 1993).

One-way analysis of variance (ANOVA) was performed to compare differences at the seasonal and annual scales among three parameters: number of species, abundance, and biomass.The least significant difference test was conducted because it is considered a valid means of a posteriori comparison to infer differences among several groups through multiple comparisons.Pairedt-tests were applied to diel and spatial comparisons of the three parameters.For all analyses,P＜0.05 was considered statistically significant.Sine function transformation (sin) was used, if necessary, to meet the assumptions of independence, normality, and homoscedasticity prior to statistical analysis to ensure accuracy of the results (Pessanha and Araújo, 2003).All statistical analyses were conducted in SPSS software (Version 25.0 for Windows, Chicago, IL, USA).One-way analysis of similarity (ANOSIM), as a nonparametric test, was expressed byR-statistics andP-values that were tested to evaluate the extent of similarity/difference in community composition and to assess the statistical significance level of variations between groups.R-statistics has a value from -1 to 1.Values closer to 1 indicate greater dissimilarity between groups, whereas values closer to 0 reflect higher similarity between samples.Additionally, a permutation test was performed to obtain the significance level using PRIMER 5.2 (Clarke and Warwick, 1994).

Redundancy analysis (RDA) was used to examine the relationships between fish assemblages and biological and environmental variables (salinity,water temperature, chlorophylla, river runoff, bird abundance, bird species richness, zooplankton Shannon-Wiener index (H′)).Seasonal variations of bird species richness and abundance in intertidal habitats were obtained from Chongming Dongtan National Nature Reserve.Environmental factors (salinity and water temperature) and seasonal changes in chlorophylla, zooplankton species richness and abundance were derived from our laboratory.River runoff data were obtained from http://www.cjw.gov.cn.To reduce the bias from the most dominant species, all fish abundance data were log-transformed; biological and physical parameters were also log-transformed to reduce magnitude effects prior to statistical analysis.Rare species with fewer than 10 individuals were excluded from the analysis.Monte Carlo test was conductd to test the statistical significance (P＜0.05) between species and impact parameters.All statistical analyses were carried out and plots were constructed using CANOCO 4.56 (Lepš and Šmilauer,2003).

3 RESULT

3.1 Fish community

A total of 4 700 fish representing 35 species and 22 families were caught using fyke nets in Xiaonangang and Tuanjiesha creeks during the 4-year study (Table 1).Of the 22 fish families found in the nekton community, the two most species-rich families were Cyprinidae and Gobiidae, with seven and five species, respectively.Among the 35 species caught, 19 species frequently occurred in the creeks,and the remaining of 16 species was regarded as rare with no more than 10 juveniles of each species every sampling period.Based on distinctive living behaviors, all species could be assigned to four ecological categories, including 3 catadromous species,5 marine species, 12 freshwater species, and 15 estuarine species.

Among the 35 species captured, only five estuarine fish species were defined as dominant species in the fish assemblage according to the IRI(＞1 000), and these species represented 83.36% of the total abundance.Based on the rank ordination in decreasing abundance, these species wereLizahaematocheila,Boleophthalmuspectinirostris,Acanthogobiusommaturus,Taenioidesrubicundus,andLateolabraxmaculatus.The remaining 16.64%of the total fish catch were species present in relatively low numbers, and they were mainly rare species.Furthermore, the same five species contributed 81.12% of the total biomass, but with a slight change in rank ordination in descending biomass:L.haematocheila,B.pectinirostris,T.rubicundus,A.ommaturus, andL.maculatus.

Table 1 Estuarine association category, species composition, percentage numerical abundance (%N), percentage biomass(%B), percentage frequency of occurrence (%Fo), and index of relative importance (IRI) for fishes caught by fyke nets in two intertidal creeks of the Changjiang River estuary

3.2 Spatial distribution

The ANOSIM test confirmed that the fish community composition was similar between Xiaonangang and Tuanjiesha creeks, albeit with conspicuous topographic features (ANOSIM,P=0.568,R-statistic=-0.018).A total of 22 species were common in both intertidal marsh creeks, but only 19 species were frequently present and considered generalists, with at least 10 juveniles collected during the study (Fig.2).Moreover,from the perspective of ecological guilds, the two intertidal creeks were primarily dominated by estuarine(nine species) and freshwater species (six species),but several catadromous and marine species were also recorded.However, some of the most dominant species (e.g.,B.pectinirostrisandT.rubicundus)were captured in higher abundance in relatively wider creeks, whereasA.ommaturushad an affinity for narrower marsh creeks.

In general, no significant differences of the number of species (pairedt-test, df=19,t=1.502,P=0.150),abundance (t=-0.855,P=0.403), and biomass (t=1.123,P=0.275) were obtained between Xiaonangang and Tuanjiesha creeks (Supplementary Fig.S1).In terms of the five most dominant species, none revealed a clear difference in the total number of individuals(pairedt-test, df=19,t=1.017,P=0.322 forL.haematocheila;t=0.037,P=0.971 forT.rubicundus;t=-0.159,P=0.875 forB.pectinirostris; andt=0.018,P=0.985 forL.maculatus, Supplementary Fig.S2a),with the exception ofA.ommaturus, which showed a pronounced difference (t=-3.069,P=0.006).The dominant species displayed similar biomass between the two creeks (pairedt-test, df=19,t=0.338,P=0.739 forL.haematocheila;t=-1.007,P=0.327 forT.rubicundus;t=-0.293,P=0.772 forA.ommaturus;t=-0.418,P=0.680 forB.pectinirostris;t=-0.791,P=0.439 forL.maculatus, Supplementary Fig.S2b).Therefore, because these data for the two creeks were similar in our investigation, we considered the sampling at these two sites as repeated measurements, and thus combined the site data for the following analyses.

3.3 Temporal variation

No obvious diel distinctiveness in fish species richness (pairedt-test, df=19,t=-1.558,P=0.136),abundance (t=0.105,P=0.917), and biomass (t=1.279,P=0.216) was observed between diurnal and nocturnal activities (Supplementary Fig.S3).Moreover,the five most abundant species did not show diel patterns throughout the study period (Supplementary Fig.S4a–b).The species composition was similar between day and night according to ANOSIM test(P=0.912,R-statistic=-0.078), and the majority of fish were collected throughout the whole sampling episodes (Fig.3).Twenty-one species were present during the day and night, whereas eight species only occurred during the day, and six species only occurred at night.

No distinct seasonal differences were detected for species richness and abundance (one-way ANOVA,F3,12=0.828,P=0.503;F3,12=1.288,P=0.323,respectively), but a pronounced seasonal difference was observed in biomass (F3,12=4.910,P=0.019), and particularly for autumn vs.spring and autumn vs.winter(F3,12=4.910,P=0.010, andF3,12=4.910,P=0.004,respectively).In general, the species composition entirely differed between seasons according to the ANOSIM test (P=0.019,R-statistic=0.688), with higher dissimilarity detected between winter and spring,and summer, and autumn assemblages, with 57.09%,51.7%, and 60.01% dissimilarities, respectively.

Fig.2 Relative proportion of abundance (a) and biomass (b) of fish species in two intertidal creeks, and the total numbers of individuals (n)

Fish abundance by season (Fig.4) displayed a trend of lower abundance recorded in the early months of the year, particularly in spring and winter,and there was higher abundance in summer and autumn whenL.haematocheilaandB.pectinirostriswere dominant in the nekton fauna.Freshwater and estuarine species constituted a large portion of abundance in fish assemblages over the spring to autumn periods, whereas a considerable influx of marine and catadromous species entered the creeks and collectively became major parts of the fish fauna during the summer and winter periods.Although there were similar trends overall, fish biomass markedly peaked in autumn, which did not match the season of peak abundance.

Across the 4-year study, pronounced differences in species number, abundance, and biomass were observed in May (one-way ANOVA,F3,12=8.655,P=0.002 for species number;F3,12=9.312,P=0.003 for biomass) and October (F3,12=8.218,P=0.003;F3,12=7.169,P=0.005;F3,12=5.245,P=0.015, respectively).The change in fish abundance in May did not dramatically differ but was close to statistical significance (F3,12=3.308,P=0.057).

Fig.4 The number of species (a), abundance (b), and biomass (c) of fish assemblages in two intertidal creeks from May 2015 to December 2015

The species composition between years substantially differed based on the ANOSIM test (P=0.01,Rstatistic=0.938 for May;P=0.029,R-statistic=0.50 for October) (Fig.5).In May, average dissimilarity was higher between samplings from 2017 and those from 2015, 2016, and 2018, with 51.41%, 54.67%,and 58.35% dissimilarities, respectively.Moreover,higher average dissimilarity was observed in October 2015 vs.2018, with 44.96% dissimilarity, followed by October 2016 vs.2018 and October 2016 vs.2015, with 37.31% and 36.93% dissimilarity,respectively.Peak abundance was reached in October of each year.Individual fluctuations revealed substantially higher peaks in May 2016 and in October 2016 and 2018, whenA.ommaturusandL.haematocheilawere overwhelmingly abundant in the fish fauna.Results of comparison of biomass between years were consistent with the abundance data (Fig.5).

3.4 Body size

Most of the fish captured in our study were juveniles of approximately ＞1 years of age.Four abundant species, with a length range of 2–30-cm TL, dominated the fish assemblages on seasonal and annual scales.On a seasonal scale (Fig.6), the length frequencies of those four species generally showed increasing modal sizes from spring onwards.B.pectinirostriswas the smallest of them, with the smallest individuals recorded in winter, whereas the other seasons were characterized by catches of relatively larger sized individuals and the largest individuals were recorded in autumn.

From an annual perspective (Fig.7),L.haematocheilashowed a consistent body size range across the four years, with a modal size range of 6–30-cm TL;A.ommaturuswere relatively small in catches across the study period, except in 2017, when relatively larger-sized individuals frequently occurred in the fyke net catch;B.pectinirostriswere smaller in body size, with total lengths of 6–18 cm from year to year; andL.maculatuswere frequently smaller in 2015 and 2018, and frequently larger in 2016 and 2017.

3.5 Relationships between fish assemblages and environmental factors

RDA results indicated that the first and second axes collectively accounted for 87.2% of the total variance in fish assemblages.Among them, the eigenvalues of the first two axes were 0.634 and 0.238, respectively.The lower eigenvalues of the other two axes reflected lower correlation between fish communities and measured parameters.Therefore,the results derived from the first and second axes were visualized using RDA ordination plot (Fig.8).However, no statistical significance in first and all canonical axes was detected through Monte Carlo test (P=1.000 for both), which indicated that all of the variables in our study did not play a significant role in explaining the variation in fish assemblages but displayed higher correlation with biological and environmental factors.

Fig.6 Length vs.frequency distributions of the four most dominant fish species in two intertidal creeks from May to December 2015

Fig.7 Length vs.frequency distributions of the four most dominant fish species in two intertidal creeks from 2015 to 2018

The RDA ordination diagram showed that biological factors with the higher correlation coefficients exerted the stronger effects on fish community dynamics than physical variables.For biotic factors,zooplankton Shannon-Wiener index (R=0.683 9) and chlorophylla(R=0.813 0) were recognized as food sources of fish.The food availability was positively correlated with and largely contributed to fish community variations.On the other hand, the avian predation pressure denoted by bird abundance (R=-0.202 1) and bird species richness (R=-0.440 4) was negatively responsible for fish fauna variations.For abiotic drivers, water temperature (R=0.420 8) and salinity (R=-0.308 0) had a positive and negative impact on fish community respectively to some degree, but little effect was discerned in river runoff(R=0.032 7).

Fig.8 Redundancy analysis (RDA) triplot of fish abundance,measured variables, and sampling seasons in 2015

In terms of species distribution,C.gracilis,S.dumerili, andP.nitiduswere much closer to the vector of the zooplankton Shannon-Wiener index(H′) and chlorophylla, and characterized the autumn sampling.Higher bird species richness and abundance in the summer period were strongly facilitated by the bloom ofL.haematocheila,T.rubicundus,H.leucisculus, andA.ommaturus.The presence ofL.maculatuswas positively associated with water temperature and constituted the majority of the spring assemblage along withC.auratusandB.pectinirostris.The winter assemblage was strongly modulated by the salinity and the considerable influx ofC.lucidus,H.nehereus, andCoilia mystus.Conversely, river runoff was less effective in structuring fish communities.

4 DISCUSSION

4.1 Fish communities of intertidal creeks

Of the 35 fish species found in the nekton community of the two intertidal creeks in the Changjiang River estuary, only five species were commonly recorded across the sampling periods and considered species that inhabited the salt marsh creeks throughout the year.The retention of these species in the creeks supports the idea that areas with ample food resources and favorable habitats provide nursery and refuge functions against predation pressure and species competition (Koutsogiannopoulou and Wilson, 2007).This perspective was reinforced by the presence and high abundance of only a small number of species that dominated the fish community in other studies (Pessanha et al., 2003; Veiga et al.,2006).Among the five fish species commonly present, all were categorized as estuarine, and thus were species that could dwell in the estuary through life stages.

Juveniles and small adults overwhelmingly formed the fish assemblages, indicating that these life stages made full use of resources.The prevalence of juvenile fishes and the temporal length-frequency distributions of the species are evidence of the function of intertidal creeks in the Changjiang River estuary as unique nursery and refuge grounds.In general, the observed patterns in number of species,abundance, and biomass indicate that a portion of the fish assemblages stay in the creek habitats until late autumn or winter as their size gradually increases.One explanation is that the greater food availability(Yeoh et al., 2017) during high tides considerably contributes to fish growth.However, they might migrate from these relatively shallow but productive creeks into deeper and warmer adjacent zones to escape the creek’s more rigorous environments.An alternate explanation links their utilization of the creek habitats to predation avoidance, as noted in a previous study (Akin et al., 2005).

4.2 Spatial variations in fish assemblages

The fish species composition in the two intertidal creeks in the Changjiang River estuary largely overlapped despite the difference of salt marsh channel size.Each creek contained a large proportion of ubiquitous species without distinctive site preferences and a few rare species (present in low numbers),which were only apparent in certain locations and contributed less to fish community composition.Our findings were consistent with those of previous studies that provided evidence that different size and topographic traits between two salt marsh creeks did not affect the fish community, and both creeks shared common fish composition (Varnell et al.,1995; Koutsogiannopoulou and Wilson, 2007).In contrast, a few comparable studies have drawn the conclusion that more fish species inhibit larger and wider creeks, probably because of more available niches and resources for them to select (Subrahmanyam and Drake, 1975; Rountree and Able, 1993; Nicolas et al., 2010).Therefore, our results suggested that salt marsh creeks offered a similar option for aquatic animals and further provided a reasonable basis for wetland conservation management; thus, equal attention should be paid to all mature intertidal creeks, irrespective of size difference.

4.3 Temporal variations in fish assemblages

Diel variations were not evident in the species composition and the three attributes of the fish communities (number of species, abundance, and biomass), especially for the most dominant fish species in the intertidal creeks.The observations in our study are contrary to those of studies that found dramatic diel shifts, with far higher values of the above attributes emerging for samples collected at night compared with those for samples collected during the day (Hoeksema and Potter, 2006; Miller and Skilleter, 2006).Some studies have attributed conspicuous diel changes to an increasing threat of diurnal predation from avian piscivory (Gawlik,2002; Steinmetz et al., 2003).Notably, most avian predators are visual hunters and prefer to prey upon fish under clear daylight (Sheaves, 2001; Tweedley et al., 2016).Most dominant water birds are primarily waders and cormorants in the Dongtan marshlands, such asTachybaptusruficollis,Ardea albus,Phalacrocoraxcarbo, andEgrettagarzetta(Zhang et al., 2014).However, the predation by piscivorous birds did not result in the significant diurnal changes, which highlighted other potential drivers (i.e., inherent living behaviors, food availability,and predation by piscivorous fishes) for diurnal pattern in fish assemblages.

Fish usually avoid avian predation by physiological activities, such as frequent nocturnal movements at night (Miller and Skilleter, 2006), tight schooling of small-sized fishes during the day (Becker et al.,2011), daytime burrowing in sediments and sheltering in seagrass beds, especially in Gobiidae (Grabowska and Grabowski, 2005), and visual evasion from fyke nets by highly mobile fishes (Guest et al., 2003).Food availability such as zooplankton, phytoplankton,and benthic macroinvertebrates (Gill and Potter,1993) and predation risk by piscivorous fish (Becker and Suthers, 2014) might exert a potential influence on diurnal changes (Yeoh et al., 2017).The specific influences of predator-prey and behavioral interactions on diel pattern of fish assemblages need further studies.

Seasonal variations in fish abundance, and biomass similarly varied throughout the year, dropping to lower levels in winter and spring, and reaching substantially higher levels in summer and autumn;the same patterns were consistent with those in most American and European studies on salt marsh creeks(Koutsogiannopoulou and Wilson, 2007; Green et al.,2009).Such patterns were likely explained by the physiological response to temperature shifts throughout the year (Desmond et al., 2002).Warmer months promoted the temperature-driven reproductive and ontogenetic periods of most fish larvae, whereas freezing months made a great amount of fish leave the salt marsh creeks and move into offshore areas in response to the sharp drop in temperature (Lazzari et al., 1999).Maximum biomass was recorded in autumn, which was not consistent with the pattern of fish abundance.It was more likely that intertidal habitats were inhabited by remarkably larger-sized fish (e.g.,L.haematocheila,B.pectinirostris, andL.maculatus) based upon the seasonal length-frequency distributions.Higher food diversity and primary production in autumn (i.e., zooplankton Shannon-Wiener index (H′) and chlorophyll-aconcentration in the water column, Fig.8) may provide high food availability for fish.Therefore, these observations echo the hypothesis that salt marsh creeks act as important nursery areas for fish growth (e.g.,hyperbenthic fauna, Cattrijsse et al., 1994), which might attract the majority of fishes to occupy these habitats.

As for seasonal changes in fish ecological guilds,estuarine fish were consistently present in the Changjiang River estuary and made a large contribution to the whole community year round because they were capable of adapting to such harsh and dynamic conditions (Zhuang et al., 2006).Additionally, we found evident abundance and species richness peaks in summer, which is in part due to the arrival of some marine species, especiallyC.lucidus; such a pattern has also been observed in European marsh creeks (Veiga et al., 2006).One possible explanation of the occurrence ofC.luciduswas that it utilized the intertidal habitats or shoals as the breeding grounds from the April to June period (Zhuang et al., 2006) and had intense foraging activities around the estuary in summer (Wang et al., 2016).Additionally,marine species were also obtained in winter assemblages, possibly because of the higher salinity in intertidal creeks, where the environment was suitable for them to temporarily dwell since salinity has been a strong factor governing the fluctuation of fish fauna (Akin et al., 2003).In contrast, freshwater species frequently occurred during the spring to autumn period when the water salinity was lower in the marsh creeks.Three catadromous fishes occurred in the creeks because of their specific life-history behaviors or occasional flush by tidal flows, such asC.mystus, who used the estuary for spawning every spring or summer (Zhuang et al., 2006).

Notable interannual differences in species composition, number of species, abundance, and biomass were detected, except for fish abundance in May.This was consistent with previous findings of fishes in other types of tidal habitats in the United State (e.g., salt ponds and sandy beaches, Lazzari et al.,1999).The fish community structure was primarily comprised of five dominant estuarine species plus several other species that exhibited irregular presence/absence, either among months or years.

4.4 Mechanism of fish assemblage in intertidal creeks

Fish distribution pattern was highly subject to multiple biological and physical variables.Biological factors exerted the greater impacts on fish community dynamics than physical variables.Specifically, the temporal variations were strongly facilitated by food availability, but negatively impacted by avian predation.They were weakly affected by salinity, water temperature, and river runoff.Wading birds are widely considered to be a threat affecting the structure of fish assemblages in tidal systems (Bretsch and Allen, 2006; Žydelis and Kontautas, 2008).In our study, the maximal abundance ofA.ommaturus,H.leucisculus,L.haematocheila, andT.rubicundusduring the summer period coincided with the peak value of species richness and abundance of water birds (Supplementary Fig.S5).Fish juveniles are suitable dietary items for most ferocious piscivorous water birds, such asE.garzetta,Larusargentatus,andChlidoniasleucopterus(Cao et al., 2017).

In addition, zooplankton species are known to constitute a large portion of juvenile fish diets and to be capable of driving shifts in fish composition.In our study, higher zooplankton Shannon-Wiener index (H′) during the autumn period prompted greater appearance ofC.gracilis,S.dumerili, andP.nitidus.Specifically, cladocerans and copepods are two main food resources and thus heavily preyed upon by these fish larvae (Zhuang et al., 2006).Furthermore, chlorophyll-aconcentration, as an important aspect of primary productivity, has been considered when examining the structure changes in fish assemblages (Valiñas et al., 2012).In this work,a strongly positive correlation between phytoplankton chlorophyllaand autumn fish assemblage structure was detected.A marked cascading effect might occur since higher chlorophyllahas corresponded to higher zooplankton diversity in autumn and thereby affects the upper fish community composition.

In terms of physical variables, water temperature and salinity had some impacts on fish community structure variations, positively and negatively, whereas river runoff had little effect on fish fauna.In some systems, temperature and salinity have been well documented to play important roles in structuring fish communities and in colonization patterns of species under seasonal fluctuations, especially in intertidal habitats (Kneib, 1997; Desmond et al.,2002; Akin et al., 2003).In our present study, more freshwater and estuarine fish species observed in spring-autumn assemblages, which was mainly linked to the constantly higher temperatures over spring to fall periods.Warmer temperatures may support the spawning (i.e.,C.auratus), survival(i.e.,B.pectinirostris), and growth activities (i.e.,L.maculatus) in intertidal habitats and thus recruit much greater abundance of these species (Zhuang et al., 2006).Conversely, substantially higher salinity was detected during the winter period, which was a consequence of the lowest river runoff occurring in winter, when a drought period was recorded in the Changjiang River estuary (Sun, 2020).Therefore,the winter assemblages were mainly characterized by the estuarine residents plus a wide number of marine species, especiallyC.lucidusandH.nehereus, who temporarily dwelled in the creeks.A similar pattern was also detected in European estuaries(Nicolas et al., 2010).Additionally, because river runoff is an essential indicator and regulates the changes in fish community structure, a comparable study has portrayed that maximum runoff and associated lowest levels in salinity are strongly related to the influx of freshwater species (Veiga et al., 2006).However, in our study river runoff weakly affected the fish community changes but slightly boosted the occurrence of some freshwater species to some extent, probably because of the heavy rainfall during our spring-summer period (Sun, 2020).

5 CONCLUSION

The fish assemblages in two intertidal creeks from the Changjiang River estuary were highly modulated by seasonal and interannual cycles, with substantially different fish composition and remarkable disparity in species richness, abundance, and biomass.However, the diel shift in the fish communities was substantially weaker than any of the above temporal dynamics, which can be explained by the combined responses to avian predator pressure coupled with other drivers (i.e., inherent living behaviors, food availability, and predation by piscivorous fishes) in fish fauna.Although two intertidal creeks were different in size, there was no spatial difference in fish communities.The fish from two sites mainly consisted of young-of-year and juvenile fish, which indicated that both of the creeks provided crucial nursery and feeding roles for fishes.Wetland conservation management should be equally applied to all well-developed intertidal creeks, regardless of creek size.The mechanisms of fish community changes showed the stronger effects by biological factors than physical variables.The temporal variations in fish assemblages were positively facilitated by food availability, but negatively impacted by avian predation.The contributions of salinity, river runoff,and water temperature were weak when structuring fish communities.In summary, our results demonstrated that intertidal habitats were of great ecological value for estuarine fish.

In recent decades, there have been substantial anthropogenic activities in and biological influences on the Changjiang River estuary, such as wetland reclamation, resource overexploitation, pollutant discharge, and invasive species colonization.Consequently, the coastal habitats are going through dramatic degradation in biodiversity and ecosystem services.Although our study highlights the important nursery and trophic supporting roles of the intertidal creeks and further portrays the use patterns of these natural habitats by fish assemblages, more information should be obtained on how organisms respond to changes in environment and habitat structure.Moreover, further studies should compare aquatic communities between natural tidal habitats and impacted areas to investigate the specific distribution patterns.

6 DATA AVAILABILITY STATEMENT

The datasets generated and/or analyzed during the current study are available from the corresponding author on reasonable request.

7 ACKNOWLEDGMENT

We thank field assistance from the Yangtze Delta Estuarine Wetland Ecosystem Observation and Research Station, Ministry of Education and Shanghai Science and Technology Committee,Shanghai, China.