Stable isotopes reveal food web reliance on different carbon sources in a subtropical watershed in South China

Limnologica xxx (xxxx) xxx–xxx

Contents lists available at ScienceDirect

Limnologica journal homepage: www.elsevier.com/locate/limno

Stable isotopes reveal food web reliance on different carbon sources in a subtropical watershed in South China ⁎

Yanyi Zeng , Zini Lai, Wanling Yang, Haiyan Li Pearl River Fisheries Research Institute, Chinese Academy of Fishery Science, Guangzhou, 510380, People’s Republic of China

A R T I C L E I N F O

A B S T R A C T

Keywords: Pearl River Watershed IsoSource Submersed water grass Stable isotope Submersed macrophyte bed

Fish species in the Pearl River Watershed (PRW), the largest subtropical watershed in southern China, have declined dramatically in recent decades. To protect river habitats and maintain the integrity of river ecosystems, we need to know how different carbon sources contribute to aquatic food webs. We used information about stable carbon and nitrogen isotopes and IsoSource software to investigate the relative contributions of different carbon sources to the food web of the PRW. We found that the stable signatures of C3 plants (C3P), particulate organic matter (POM), submersed water grasses (SWG), and C4 plants (C4P) were significantly different and fell into four distinct groups. Of these, C3P was a high and stable source of carbon to certain consumers, while C4P was a low but stable carbon source. In comparison, the contributions from the POM and SWG groups were unstable and varied widely, between the 1st and 99th confidence intervals. The results suggest that the C3P, POM, and SWG groups might be important carbon sources for the PRW food web. Analysis of the 99th confidence interval contribution of each group that provided more than 50% showed that consumers in the PRW food web might be divided into those that mainly relied on C3P, those that mainly relied on SWG, those that potentially relied on C3P-SWG, and those that potentially relied on POM-SWG. This study provides basic information that will help support the protection and management of the PRW ecosystem, and will be especially useful to help restore submersed macrophyte beds and riparian buffer strips in this watershed.

1. Introduction Rivers are considered the cradle of human society and economic development. However, because of excessive exploitation and abuse, river water quality has deteriorated and river habitats are increasingly fragmented, with the result that river ecosystem functions are in decline worldwide (Nilsson et al., 2005; Azrina et al., 2006; Miyazono and Taylor, 2013). We therefore need to have information about the contribution of various sources of carbon that contribute to aquatic food webs to protect river habitats and maintain the integrity of river ecosystems. Three conceptual models can be used to describe the contributions of various carbon sources to river food webs. The river continuum concept holds that the majority of organic sources that support large river food webs originate from terrestrial plants in the headwater and middle reaches, while in-stream primary production is more important in the lower reaches of the river, where it may be limited by turbidity and light attenuation as depth increases (Vannote et al., 1980). The flood pulse concept proposes that lateral river floodplain exchanges, rather than organic matter subsidies from upstream, are the main source of carbon for the food web (Junk et al.,

⁎

1989). The riverine productivity model highlights the importance of local autochthonous production, such as phytoplankton, benthic algae, water grasses, and direct organic inputs from the riparian zone (Thorp and Delong, 1994; Thorp et al., 1998). In recent decades, numerous studies have demonstrated that the contributions of carbon sources to food webs depend on a range of factors including hydrological conditions, latitude, habitat characteristics, and human activities (Caraco et al., 2010; Babler et al., 2011; Kaymak et al., 2015; Delong and Thoms, 2016) and that the contributions of various carbon sources may differ significantly between river systems (Pingram et al., 2012). The Pearl River Watershed (PRW) is the largest watershed in southern China. Historically, the PRW has been known for its wide range of habitats that were home to more than 380 species of fish and other macroinvertebrates needed to support fish reproduction, growth, and fattening. In the early 1980s, the number of fish species in the lower reaches of the PRW had declined to 161 (Lu, 1990). However, in recent years, the number of fish species has declined further and now there are only 10 main species, the biomass of which accounts for 90% of the total fishing catches (Li et al., 2010). With these dramatic declines in fish species in the past 20 years, many species endemic to the

Corresponding author at: No. 1 Xingyu Road 1, Xilang, Liwan District, Guangzhou, 510380, Guangdong Province, People’s Republic of China. E-mail address: [email protected] (Y. Zeng).

https://doi.org/10.1016/j.limno.2017.11.003 Received 14 March 2017; Received in revised form 23 October 2017; Accepted 10 November 2017 0075-9511/ © 2017 Published by Elsevier GmbH.

Please cite this article as: Zeng, Y., Limnologica (2017), https://doi.org/10.1016/j.limno.2017.11.003

Limnologica xxx (xxxx) xxx–xxx

Y. Zeng et al.

debris on the river banks. More than 90% of the study watershed was wide-open-water channel, and less than 10% was submersed macrophyte beds and riparian buffer strip. The riparian buffer strips were usually on the sides of the river where the water was up to 1 m deep, and submersed macrophyte beds were usually on the sides of the river where the water was up to 3 m deep. Water deeper than 3 m was generally part of the wide water channel area, and ranged from 3 m to more than 10 m deep.

PRW have gone extinct, and aquatic ecosystem function has decreased. The declines are most likely attributable to water pollution, and the high degree of river modification through dredging, channelization, and the construction of extensive levee systems that disrupt fish habitats and natural connections between the river and its floodplains (Li et al., 2010). Given the severity of the situation, we urgently need information about the contributions of different carbon sources and habitats to the food web of the PRW, as this information can be used to develop strategies to ensure fishery sources are protected, and that aquatic ecosystem function is maintained. Stable isotopes have been frequently used to improve our understanding of aquatic food webs and aquatic ecosystems (Vander Zanden et al., 1999; Gu et al., 2011; Saigo et al., 2015), and, in particular, to support quantitative evaluations of the contributions of different carbon sources to various consumers in food webs (Zeug and Winemiller, 2008; Nyunja et al., 2009; Giarrizzo et al., 2011; Tue et al., 2014). However, because of a lack of studies in recent years, we have very little information about the stable carbon and nitrogen signatures in the PRW (Wei et al., 2008; Zhang and Ran, 2014). Statistical modules or packages, such as IsoSource software and Bayesian frameworks, that can be used to analyze the contributions of different carbon sources to consumers based on stable isotopes (Benstead et al., 2006; Cole and Solomon, 2012; Wang et al., 2014a) are increasingly available, though these different methods should be evaluated for their advantages and disadvantages before application (Moore and Semmens, 2008; Parnell et al., 2010; Fry, 2013). The aim of the present study was therefore to estimate the contribution of potential carbon sources to the main consumers in the PRW food web using information about stable carbon and nitrogen isotopes and IsoSource software. From this study we hope to collect basic data that will support the management, protection, and restoration of the PRW and the maintenance of ecological function; we also hope that our data will supplement existing information on carbon fluxes in river systems worldwide.

2.2. Sample collection and analysis We collected samples of potential sources of organic carbon and of the main fish and macro-invertebrates present at the four sampling sites in March and August of 2015. The potential sources of organic carbon included particulate organic matter (POM) from the main stem of the wide-open water area; submersed water grasses (SWG) including V. denseserrulata, N. marina, H. varticillata, Ceratophyllum demersum, and Potamogeton crispus from the submersed macrophyte beds; emerged C3 plants, such as P. trigonocarpum, C. palustris, Oenanthe sinense, and Gratiola japonica, and C4 plants, such as C. rotundus, H. compressa and Carex biuensis from the riparian buffer strip. Surface water was filtered through coarse sieves (100 μm) and then was further filtered onto Whatman GF/F glass fiber filters (0.7 μm) that had been pre-combusted at 450 °C for 5 h to collect the fine POM. Using this method we collected POM that mainly comprised planktonic algae (Wang et al., 2014a); our POM samples were made up mostly of diatoms, which dominate POM in the Pearl River (Wang et al., 2014b, 2016). The filters with the retained POM were acidified with hydrochloric acid vapor in a dryer system for 24 h to remove any carbonate, and then the filtered samples were dried at 55 °C and kept in a dryer until stable isotope analysis. The submersed macrophyte beds were usually under water that was between 0.5 and 1.5 m deep, so we hired a local fisherman to collect water grass samples from between three and five sampling frames at each (each with an area of 0.25 m2), which were then combined into one sample each sampling site. The water grasses were classified and the dominant species were rinsed, packed separately, stored in a portable refrigerator, and transported back to the laboratory. In the laboratory, the algae or seston attached to the leaf were scraped off the water grasses, and then the grasses were rinsed and dried at 55 °C. The dominant C3 and C4 plant species were identified and collected from riparian buffer strips around each site and transported back to the laboratory, where they were treated in the same way as the SWG samples. The individual species collected at each site were combined into a single sample and ground, and then passed through a 60-mesh screen before stable isotope analysis. During the first investigation in March, the dry season with relative high biomass of benthic algae and epiphytes to wet season of August, we scrapped the benthic algae and the epiphytes from the stones or SWG, and treated them in the same way as POM. Their stable carbon and nitrogen isotope values were in the same range as those of SWG and their biomass was negligible relative to the biomass of SWG. Given that the stable isotope values of benthic algae, epiphytes, and SWG overlapped, we just collected SWG samples during the second sampling in August 2015. We did not collect the coarse POM or benthic detritus as they may have been complex mixtures of algae, C3P, C4P, and submersed water grasses as previous study demonstrated (Wang et al., 2014a), and may have caused confusion in the IsoSource analysis. Most of the fish and macroinvertebrates species were bought directly from local fishermen on their boats, who caught them around the sampling sites by various methods including seine nets, gillnets, bottom supporting nets, or shrimp traps. We hand-collected species of Limnoperna fortunei, an invertebrate that is attached to big stones, rocks, piers, wharves, and macrophytes, around the sampling sites, stored them in a portable refrigerator, and transported back them to the laboratory. The lengths and weights of the fishes were measured before dissection. The dorsal white muscle tissues from fishes, shrimp muscles

2. Material and methods 2.1. Study area The PRW drains an area of 442,100 km2 and is the largest watershed in southern China. The Pearl River has an annual runoff of about 300 billion m3 and an annual average flow of 11,000 m3 per second. It has three main tributaries, the West River (Xijiang River), the North River (Beijiang River), and the East River (Dongjiang River). The present study was conducted in the lower reaches of the West and North Rivers (Fig. 1), where the main tributaries are anastomosing by forked channels of various widths. The lower reaches of the PRW, the key habitat for the fattening and growth of many riverine fishes and macroinvertebrates, are characterized by these linked channels. The PRW has a subtropical monsoon climate, with frequent light rain through spring, wet hot summers, and frequent typhoons in autumn and dry winters. The annual average water temperature ranges from 20 °C to 22 °C (Yan, 1989; Lu, 1990). Water pollution and engineering works, such as river dredging, channel expansion, and the construction of revetments and dams, have resulted in dramatic changes in the aquatic environment, habitats, and communities of the PRW, particularly in the spawning grounds of aquatic species (Li et al., 2010; Tan et al., 2012). We established four sampling sites in the linked channel that flows through Xiaotang, Beijiao, Zuotan, and Waihai Counties (Fig. 1). The sampling sites had sandy sediment beds and habitats that were representative of the conditions in the lower reaches of the Pearl River. Beds of submersed macrophytes were dominated by, among others, Vallisneria denseserrulata, Najas marina, and Hydrilla varticillata, and riparian vegetation was made up of C3 plants dominated by Polygonim trigonocarpum and Callitriche palustris, and C4 plants dominated by Cyperus rotundus and Hemarthria compressa. There was very little woody 2

Limnologica xxx (xxxx) xxx–xxx

Y. Zeng et al.

Fig. 1. Sampling sites in the lower reaches of the Pearl River.

matrix case labels arrayed as each species with their average values of δ13C, δ15N and C/N were input into HCA in SPSS 16.0. Cluster analysis was based on Ward’s cluster method and the squared Euclidean distance for interval data. Before entering the isotope data into the IsoSource model, the trophic enrichments of the δ15N and δ13C values for each consumer were corrected depending on the consumer’s trophic level (TL) and the respective isotope enrichment factors. The TL was calculated as follows (Post, 2002):

without intestines, and soft tissues of the other invertebrates were collected, freeze-dried, ground to powder, and then passed through a 60-mesh screen before isotopes analysis. Stable isotope analysis was completed in the Public Service Center of Instruments and Equipment in the South China Sea Institute of Oceanology, of the Chinese Academy of Sciences. The 15N/14N and 13 C/12C ratios of organic samples were determined by isotope-ratio mass spectrometry (Thermo Scientific Flash EA 1112 HT Delta V). Stable isotope data were expressed as the relative difference ratios between the heavy to light isotopes and an internationally accepted reference standard, calculated as: δX = [(Rsample/Rstandard) − 1] × 1000

TL = [(δ15Nconsumer − δ15Nbaseline)/Δ15N + 2

(2)

where δ Nconsumer was the average δ N of the consumer species whose TL is estimated; δ15Nbaseline was the average δ15N of L. fortunei, a filter mussel commonly-found in the PRW that represented primary consumers with a TL of 2, and Δ15N was the mean trophic fractionation factor of 2.54‰, obtained from a meta-analysis of previous studies (Vanderklift and Ponsard, 2003; Caut et al., 2009). As in previous studies, we used IsoSource to estimate the contributions of different carbon sources to consumers in the PRW (Phillips and Gregg, 2003; Wang et al., 2014a). Potential groups of organic carbon sources were entered into IsoSource as the average value of each group. To prepare the isotope data for the IsoSource model, the TLs of the consumers above the primary producer (i.e. TL-1) were multiplied by isotope enrichment factors of 0.4‰ and 2.54‰ for δ13C and δ15N, respectively, to obtain the taxon-specific isotope enrichment levels. These values were then added to each production source. The mean δ13C and δ15N values for each taxon and the four groups of carbon sources (corrected for isotope enrichment) were entered into the IsoSource mixing model, with increments of 1%. The tolerance was initially set at 0.1‰; if the mixture isotope values were out of bounds, the tolerance value was incrementally increased by 0.1‰. We restricted the use of the IsoSource output to the median (50th confidence interval) and ranges (i.e. 1st to 99th confidence interval) of the source contributions for each source and consumer species. Further details of the method have been published elsewhere (Phillips and Gregg, 2003; Vanderklift and Ponsard, 2003; Benstead et al., 2006; Zeug and 15

(1)

where X is 15N or 13C, and R is the corresponding 15N/14N or 13C/12C ratio of the sample and the standard. The δX was expressed as the per mil (‰) deviation of that sample from working standards (urea and glycine for 15N, glycine and cellulose for 13C) and the measurement precision was approximately 0.3‰ and 0.1‰ for 15N/14N and 13C/12C, respectively.

2.3. Data analysis The δ15N and δ13C values from the two samplings at the four sampling sites were pooled for analysis because the stable isotopes of food sources are generally integrated in consumers over the long-term and pre-statistical analysis showed that there were no significant differences among the δ15N and δ13C of consumers between sites or seasons. We used ANOVA to determine whether there were significant differences between the mean δ13C and δ15N values of the four nominal groups of organic carbon sources. Hierarchical cluster analysis (HCA) was used to determine whether it was appropriate to classify the species of organic carbon sources into four groups. We carried out HCA by species for SWG, C3P, and C4P and by sample for POM in different sites and seasons based on their δ13C, δ15N, and C/N ratios. The POM samples with their δ13C, δ15N, and C/N were averaged and compiled into the four sites for spring and autumn (eight cases altogether) data matrix. Other data 3

15

Limnologica xxx (xxxx) xxx–xxx

Y. Zeng et al.

Table 1 Summary statistics (mean ± SE) of δ13C, δ15N, C/N in the different groups of dominant organic carbon sources in the Pearl River watershed. Organic matter group

δ13C

δ15N

C/N

Riparian C3 plants(C3P) (n = 16) Particulate organic matters (POM)(n = 16) Submersed water grasses (SWG)(n = 18) Riparian C4 plants(C4P) (n = 12)

−30.84 ± 0.56a

6.41 ± 0.42

7.14 ± 0.82b

−26.71 ± 0.39b

7.34 ± 1.01

2.18 ± 0.25a

−22.68 ± 0.42c

6.84 ± 0.47

6.53 ± 0.26b

−13.48 ± 0.54d

5.10 ± 0.75

16.47 ± 1.45c

Note: mean values in the same column with different letters indicate significant differences, while mean values with the same letter indicate the difference is not significant. α = 0.05. Fig. 2. Hierarchical dendrogram of different species of carbon sources based on their δ13C, δ15N, and C/N ratios. The POM samples were from different sites and the two sampling times.

Winemiller, 2008; Wang et al., 2014a). 3. Results

3.2. δ13C-δ15N niche of the main aquatic organisms

3.1. δ13C and δ15N of diverse basal carbon sources

A total of 120 samples of aquatic animals were analyzed for their δ13C and δ15N values. There were significant differences among the δ13C-δ15N niches of the dominant fish and macroinvertebrate species, and there was some overlap between the species, as shown in Fig. 3. We divided the niche map into four quadrats to display the distributions of the various species as follows: quadrat a contained enriched δ13C and enriched δ15N, quadrat b contained depleted δ13C and enriched δ15N, quadrat c contained depleted δ13C and depleted δ15N, and quadrat d contained enriched δ13C and depleted δ15N. On the niche map, M. terminalis, a main endemic commercial fish species in the Pearl River, was mainly distributed from the center to the top of the niche (shown in the orange oval), implying a relatively high trophic level with medium δ13C values. Different-sized individuals of M. terminalis had similar isotopic carbon ratios but different isotopic nitrogen ratios. The δ15N values were highest in the M. terminalis population (Mt 3+), with a mean of 13.27‰, implying that they held the highest trophic position in the collected food web. The δ15N values were lower in the 1+ and 2+ M. terminalis populations than in the older 3+ populations. The other six dominant fishes, namely P. pekineisis, S. curriculus, C. carpio, C. molitorella, and X. argentea, all native species, and O. niloticus, an invasive species, were scattered through the niche map, but, as shown in the green oval, were mainly in quadrant a and quadrant d, indicating that the carbon ratios of these individuals were relatively enriched, and they might utilize enriched δ13C sources. The dominant molluscs, L.

A total of 62 samples of potential carbon sources were analyzed for their δ13C and δ15N values. They were nominally classified into four groups, namely C3P, POM, SWG, and C4P. The four dominant C3 plant species collected from the riparian buffer strip on the side of the river had δ13C values between −32.35‰ and −29.69‰, the lightest in this study, and δ15N values between 5.24‰ and 7.05‰. The δ13C and δ15N of POM ranged from −28.69‰ to −23.73‰, and from 0.13‰ to 10.89‰, respectively. The δ13C in the five main SWG ranged from −25.77‰ to −19.80‰, while the δ15N ranged from 2.75‰ to 9.30‰. The three dominant C4 plant species in the riparian buffer strip on the side of the river had δ13C values between −15.36‰ and −12.00‰, the most enriched δ13C values in this study, and δ15N values that ranged from 3.92‰ to 8.32‰. The average δ13C values differed significantly among the different groups of the main organic carbon sources (ANOVA, p < 0.001, Table 1), and the mean values were ranked from lowest to highest as C3P < POM < SWG < C4P (Table 2). The average δ15N values of the different groups of dominant organic carbon sources were not significantly different (ANOVA, p > 0.05, Table 1). The HCA grouped the potential carbon sources into four clusters; cluster one comprised the four species of C3P, cluster two comprised POM from the four sites from both sampling times, cluster three contained five species of SWG, and cluster four contained three species of C4P (Fig. 2). The results are consistent with the results of ANOVA analysis, and were suitable for analysis by the IsoSource mixing model.

Table 2 Median (1st–99th) confidence interval (CI) contributions of the 4 potential groups of organic carbon sources to the main fish (*) and macroinvertebrate (**) species in the lower reaches of the Pearl River Watershed. Species

Megalobrama terminalis 3 + * Macrobranchium nipponense ** Squaliobarbus curriculus* Xenocypris argentea* Megalobrama terminalis 2 + * Megalobrama terminalis 1 + * Cyprinus carpio * Cirrhinus molitorella * Anodonta woodiana ** Corbicula fluminea ** Oreochromis niloticus * Limnoperna fortunei ** Parabramis pekineisis * Gammarus sp. **

Tolerance (‰)

0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.2

Median (1th–99th) CI contribution proportion POM

C3P

SWG

C4P

13(0–35) 19(0–43) 24(0–53) 20(0–46) 15(0–36) 15(0–36) 29(0–62) 22(0–49) 13(0–32) 11(0–26) 22(0–49) 8(0–20) 25(0–54) 7(0–28)

61(55–68) 44(37–51) 27(19–35) 39(31–46) 56(49–63) 56(49–63) 8(0–16) 32(25–40) 61(55–68) 74(67–80) 34(26–41) 84(77–89) 25(17–33) 1(0–3)

16(0–38) 23(0–50) 29(0–63) 24(0–54) 18(0–42) 19(0–43) 36(0–75) 27(0–59) 16(0–38) 11(0–25) 26(0–58) 6(0–14) 29(0–64) 59(30–72)

9(1–14) 14(5–21) 19(8–28) 16(7–24) 10(2–16) 9(1–16) 26(14–37) 18(8–26) 9(1–14) 4(0–9) 17(7–25) 2(0–5) 20(9–29) 32(27–39)

4

Limnologica xxx (xxxx) xxx–xxx

Y. Zeng et al.

Fig. 3. Scattergram of δ13C-δ15N for the main aquatic animals (individual) and the potential groups of carbon sources (M ± SE) in the Pearl River Watershed. The species names were abbreviated as follows: Mt3+ (Megalobrama terminalis populations of 3-years or older), Mn (Macrobranchium nipponense), Sc (Squaliobarbus curriculus), Xa (Xenocypris argentea), Mt2+ (Megalobrama terminalis population of 2-year old), Mt1+ (Megalobrama terminalis population of 1-year old), Cc (Cyprinus carpio), Cm (Cirrhinus molitorella), Aw (Anodonta woodiana), Cf (Corbicula fluminea), On (Oreochromis niloticus), Lf (Limnoperna fortunei), Pp (Parabramis pekineisis), Gs (Gammarus sp.). The symbol * represents fish species, ** represents macroinvertebrate species. The intersection of the dotted lines is the center point (−25.48‰, 10.91‰) of the δ13C–δ15N niche map and the dotted lines divide the δ13C–δ15N niche map into four quadrants.

Fig. 4. Estimation of trophic levels for the dominant fish and macroinvertebrate species in the Pearl River Watershed with Limnoperna fortunei as the baseline. Fish species names are indicated by *, while macroinvertebrate species are indicated by **.

carbon to all the main consumers. There was huge variation in the 1st to 99th CI contributions of POM and SWG to almost all the aquatic consumers, with 99th CI contributions to certain aquatic consumers that exceeded 50%, which implies that there were wide variations and considerable uncertainty associated with their contributions to certain consumers; the contribution of SWG, with a median of about 25%, to all consumers was steady (Table 2).

fortunei, A. woodiana, and C. fluminea, were mainly distributed throughout quadrant c, as shown in the blue oval, which implies their trophic levels were low and their carbon ratios were depleted with potentially depleted δ13C sources. The δ13C was most depleted in species of the widely-distributed mussel L. fortunei; we used this species as a baseline for trophic estimation because there were only minor differences between the mean δ13C and δ15N values for the different individuals. Gammarus sp., with a value of −19.73‰, had the most enriched δ13C, and was distributed in quadrant d, implying that its trophic level was low and its carbon ratio was enriched (Fig. 3).

4. Discussion We analyzed samples and determined the stable carbon and nitrogen signatures for different consumers in the PRW, and estimated the basic structure of the food web. We found that the food chain length in the PRW was 3.67, which is consistent with the food chain length of 3.79 ± 0.89 reported for rivers in a review of global aquatic food webs (Vander Zanden and Fetzer, 2007). This confirms that the method we used to estimate the trophic level was appropriate. These results can serve as basic information about energy flows and material cycling in the PRW food web, and can form the basis for future isotopes studies in the PRW. Potential carbon sources in river food webs are often divided into two groups: autochthonous (e.g., phytoplankton, benthic algae, biofilms, and macrophytes) and allochthonous (e.g., processed organic matter from upstream, terrestrial inputs derived from floodplain interactions, and anthropogenic sources) (Pingram et al., 2012). Sometimes, however, it is difficult to distinguish whether specific sources are autochthonous or allochthonous when they are inside the gastrointestinal tracts of aquatic animals, e.g. it is difficult to know whether certain C3 or C4 plants represent autochthonous or allochthonous sources. As in previous studies, we used stable carbon and nitrogen isotope ratios to differentiate the potential carbon sources into different groups (Herzka, 2005; Zeug and Winemiller, 2008; Wang et al., 2014a). Our current study showed that the δ13C values of the nominated source groups of C3P, POM, SWG, and C4P were significantly different. The mean δ13C of these groups increased as follows: C3P < POM < SWG < C4P, which, even though they did not include SWG samples, was similar to the rank orders reported for the Brazos and Yangtze Rivers (Zeug and Winemiller, 2008; Wang et al., 2014a). Because of their specific habitat in the submersed macrophyte bed, and also because their δ13C values differed significantly from the values in the other sources guilds, we separated the SWG into a separate source

3.3. Trophic level (TL) estimates of the main aquatic consumers The δ15N data indicated that the maximum trophic level in the PRW was 3.67, so there were about 4 trophic levels (Fig. 4). The mean relative TLs for the crustacean gammarid, mussel L. fortunei, clam C. fluminea, and two fish species P. pekineisis and O. niloticus were around 2, which indicates they were primary consumers in the PRW food web. The mollusk A. woodiana had a trophic level of 2.55, which was similar to those of the common fish species C. molitorella and C. carpio. The TLs of the fish species S. curriculus, the crustacean M. nipponense, and the native fish species M. terminalis were highest and were above 3. It is noteworthy that the M. terminalis species occupied various trophic levels between 2.75 and 3.67 as their population grew from 1+ to 3+, indicating significant changes in food items as the species aged. 3.4. Contribution of potential organic carbon sources to the main consumers Stable δ13C and δ15N values together provided good discrimination among the four potential groups of carbon sources, as shown by the statistical differentiation of at least 4‰ between the mean δ13C for the four groups of sources; the mean δ15N did not differ statistically among the 4 organic carbon sources (Table 1). The 50th percentile and the ranges of the proportions (1st to 99th confidence intervals, CI) contributed by the four potential carbon sources to consumers in the Pearl River are listed in Table 2. The IsoSource model results showed that C3P was a relatively fixed source that supported some aquatic animals, such as M. terminalis (Guangdong bream), A. woodiana, O. niloticus, and L. fortunei, while C4P accounted for a smaller fraction of the assimilated 5

Limnologica xxx (xxxx) xxx–xxx

Y. Zeng et al.

mainly from C3 plants, rather than C4 plants. In contrast, the riverineproductivity model emphasizes the importance of algae production to food webs in the channels of large rivers (Thorp and Delong, 1994). We collected POM instead of pure samples of algae for two reasons. First, it is difficult to collect pure samples of algae in rivers because phytoplankton, particulate matter, and micro-zooplankton are mixed as socalled POM with similar sizes. Second, the POM samples we collected with the pre-separating method were primarily comprised of phytoplankton. Algae were therefore represented by POM in this study. We found that the stable carbon signature of POM in this study was similar to that of another large river (Zeug and Winemiller, 2008), and so, along with SWG, may also be an important carbon source for certain consumers in the Pearl River watershed food web. Our results showed that SWG were the main carbon source for Gammarus sp. and potentially, along with POM and/or C3P, provided carbon to more than half of the collected dominant consumers from the PRW, with a 99th CI contribution of more than 50%. This indicates that SWG were a good source of feeding materials for fishes and invertebrates in the PRW food web, as reported in previous studies (Lu, 1990; Strayer et al., 2003). Submersed macrophyte beds also provide habitat for growing young fishes (Garner, 1996; Baras and Nindaba, 1999) and adult spawning and reproduction (Copp, 1992; Copp et al., 1994). We speculate that algae or SWG may be limiting in the PRW. If available in sufficient quantities, the contributions of algae or SWG biomass to the potentially reliant consumers would perhaps be stable and higher, but we need to investigate this further. It is relatively difficult to control algae biomass in a river but relatively easy to restore SWG beds. Therefore, given that the potentially dominant reliance of C3P and SWG by consumers, the submersed macrophyte beds and the riparian buffer strips that are habitat for the SWG and C3P in the PRW, respectively, should be protected and restored, which may help ensure the future function and integrity of ecosystems in the PRW.

guild. The IsoSource model results indicated that the C3P sources (mainly P. trigonocarpum, C. palustris, O. sinense and G. japonica) made relatively fixed contributions to the biomass of certain consumers in the PRW food web, and that, similar to the results of other studies, the contributions of C4P sources to aquatic animals in the PRW were minor (Zeug and Winemiller, 2008; Wang et al., 2014a). The earlier studies in the Yangtze River (the longest river in China) and the Brazos River (ranked 11th longest in the USA) showed that river food webs were mainly supported by terrestrial C3 plants (Zeug and Winemiller, 2008; Wang et al., 2014a), while other studies reported that large river food webs were mainly supported by algal carbon (Lewis et al., 2001; Douglas et al., 2005). While our results did not show fixed contributions from POM or SWG to the food web, POM and SWG may be important for certain consumers, e.g. the 99th confidence interval contribution proportion of SWG to C. carpio was up to 80%. The δ13C and δ15N values of benthic algae and epiphytic algae in the PRW were similar to those of SWG. In the present study, we could not use the δ13C and δ15N values to differentiate SWG from the benthic and epiphytic algae, which suggest that the contribution of benthic and epiphytic algae may be important, as previously observed in turbid rivers (Hamilton et al., 1992; Lewis et al., 2001). While these results are useful, we need to carry out more advanced combined studies of stable isotopes and fatty acid biomarkers in the PRW to confirm this is actually the case (Fujibayashi et al., 2016). Factors such as river size, latitude, fluvial dynamics, degree of seasonal variation, and human activities may alter the diversity of primary producers in different river systems, with consequences for the contributions of different carbon sources (Caraco et al., 2010; Babler et al., 2011; Kaymak et al., 2015; Delong and Thoms, 2016). However, given that there were few significant differences in the stable carbon and nitrogen signatures of the organic carbon sources between the two samplings and the four sites, these factors may not be important in the PRW, as suggested by a previous report of the particulate organic carbon-isotope composition in this area (Wei et al., 2008). Habitats in the PRW are characterized by wide open water channels, and have few riparian buffer strips that are dominated with emerged C3 or C4 plants, or submersed macrophyte beds covered by submersed water grasses. Therefore, our results reflect some combination of the previous theories about how large river food webs are supported by carbon sources. Realistically, as was the case in this study, several species will be mainly supported by certain carbon source guilds while other species will be mainly supported by other different carbon source guilds, which further confirms that the different habitats in the PRW had species-specific roles. From their maximum potential reliance (the 99th CI contribution of certain carbon sources over 50%) on different carbon source guilds, the consumers from the food web of the PRW fell into four functional groups, namely a group that mainly relied on C3P (represented by population M. terminalis 1+, 2+, 3+, A. woodiana, O. niloticus and L. fortunei), a group that mainly relied on SWG (represented by Gammarus sp.), a group that mainly relied on C3P-SWG (represented by M. nipponense), and a group that mainly relied on POM-SWG (represented by S. curriculus, X. argentea, C. carpio, C. molitorella, C. fluminea, P. pekineisis). The results of this study did not contradict previous theories for large river food webs (Pingram et al., 2012), but rather showed that the contributions from different carbon sources were mixed. In the floodpulse concept, Junk et al. (1989) claimed that consumers in the main channels of rivers were mainly supported by terrestrial carbon sources that originated on the floodplain. While the species of C3P in this study comprised a group of carbon sources that lived on the riparian buffer strip, they are representative of many C3P species, regardless of whether they originate from shallow wetlands or from upstream terrestrial sources, because the stable carbon isotopes of wetland C3P species and terrestrial leaves of C3 trees are similar (Peterson et al., 1985; Wang et al., 2014a). Our results highlight the importance of terrestrial material to biomass in the PRW food web. However, these sources were

Funding This research was funded by the National Natural Science Foundation of China (Grant No. 31500434), and the Natural Science Foundation of Guangdong Province, China (Grant No. 2015A030310297), and by part of the Special Scientific Research Funds for Central Non-profit Institutes, Chinese Academy of Fishery Sciences (No. 2015A01YY02). Acknowledgements Special thanks are given to Dr. Songyao Peng, Yuefei Li, and Fangcan Chen for classifying the macroinvertebrates and fishes. We also thank Dr. Yongzhan Mai, Yuan Gao and all of the colleagues and students for assisting with fieldwork. We are grateful to Dr. Shouhui Dai for the stable isotope analysis. We thank Deborah Ballantine, PhD, from Liwen Bianji, Edanz Editing China (www.liwenbianji.cn/ac), for editing the English text of a draft of this manuscript. References Azrina, M., Yap, C., Ismail, A.R., Ismail, A., Tan, S., 2006. Anthropogenic impacts on the distribution and biodiversity of benthic macroinvertebrates and water quality of the Langat River, Peninsular Malaysia. Ecotoxicol. Environ. Saf. 64, 337–347. Babler, A.L., Pilati, A., Vanni, M.J., 2011. Terrestrial support of detritivorous fish populations decreases with watershed size. Ecosphere 2, 1–23. Baras, E., Nindaba, J., 1999. Seasonal and diel utilization of inshore microhabitats by larvae and juveniles of Leuciscus cephalus and Leuciscus leuciscus. Environ. Biol. Fish. 56, 183–197. Benstead, J.P., March, J.G., Fry, B., Ewel, K.C., Pringle, C.M., 2006. Testing IsoSource: stable isotope analysis of a tropical fishery with diverse organic matter sources. Ecology 87, 326–333. Caraco, N., Bauer, J.E., Cole, J.J., Petsch, S., Raymond, P., 2010. Millennial-aged organic carbon subsidies to a modern river food web. Ecology 91, 2385–2393. Caut, S., Angulo, E., Courchamp, F., 2009. Variation in discrimination factors (δ15N and δ13C): the effect of diet isotopic values and applications for diet reconstruction. J.

6

Limnologica xxx (xxxx) xxx–xxx

Y. Zeng et al.

Parnell, A.C., Inger, R., Bearhop, S., Jackson, A.L., 2010. Source partitioning using stable isotopes: coping with too much variation. PLoS One 5, e9672. http://dx.doi.org/10. 1371/journal.pone.0009672. Peterson, B.J., Howarth, R.W., Garritt, R.H., 1985. Multiple stable isotopes used to trace the flow of organic matter in estuarine food webs. Science 227, 1361–1363. Phillips, D.L., Gregg, J.W., 2003. Source partitioning using stable isotopes: coping with too many sources. Oecologia 136, 261–269. Pingram, M.A., Collier, K.J., Hamilton, D.P., David, B.O., Hicks, B.J., 2012. Carbon sources supporting large river food webs: a review of ecological theories and evidence from stable isotopes. Freshw. Rev. 5, 85–103. Post, D.M., 2002. Using stable isotopes to estimate trophic position: models, methods, and assumptions. Ecology 83, 703–718. Saigo, M., Zilli, F.L., Marchese, M.R., Demonte, D., 2015. Trophic level, food chain length and omnivory in the Paraná River: a food web model approach in a floodplain river system. Ecol. Res. 30, 843–852. Strayer, D.L., Lutz, C., Malcom, H.M., Munger, K., Shaw, W.H., 2003. Invertebrate communities associated with a native (Vallisneria americana) and an alien (Trapa natans) macrophyte in a large river. Freshw. Biol. 48, 1938–1949. Tan, X., Li, Y., Li, X., Li, J., Wang, C., 2012. The status of fish spawning ground in the East River with cascade dams duress. J. Lake Sci. 24, 443–449 (in Chinese). Thorp, J.H., Delong, M.D., 1994. The riverine productivity model: an heuristic view of carbon sources and organic processing in large river ecosystems. Oikos 70, 305–308. Thorp, J.H., Delong, M.D., Greenwood, K.S., Casper, A.F., 1998. Isotopic analysis of three food web theories in constricted and floodplain regions of a large river. Oecologia 117, 551–563. Tue, N.T., Hamaoka, H., Quy, T.D., Nhuan, M.T., Sogabe, A., Nam, N.T., Omori, K., 2014. Dual isotope study of food sources of a fish assemblage in the Red River mangrove ecosystem, Vietnam. Hydrobiologia 733, 71–83. Vander Zanden, M.J., Fetzer, W.W., 2007. Global patterns of aquatic food chain length. Oikos 116, 1378–1388. Vander Zanden, M.J., Casselman, J.M., Rasmussen, J.B., 1999. Stable isotope evidence for the food web consequences of species invasions in lakes. Nature 401, 464–467. Vanderklift, M.A., Ponsard, S., 2003. Sources of variation in consumer-diet δ15N enrichment: a meta-analysis. Oecologia 136, 169–182. Vannote, R.L., Minshall, G.W., Cummins, K.W., Sedell, J.R., Cushing, C.E., 1980. The river continuum concept. Can. J. Fish. Aquat. Sci. 37, 130–137. Wang, J., Gu, B., Huang, J., Han, X., Lin, G., Zheng, F., Li, Y., 2014a. Terrestrial contributions to the aquatic food web in the Middle Yangtze River. PLoS One 9, e102473. Wang, C., Li, X., Lai, Z., Li, Y., Dauta, A., Lek, S., 2014b. Patterning and predicting phytoplankton assemblages in a large subtropical river. Fund. Appl. Limnol. 185 (3–4), 263–279. Wang, C., Li, X., Wang, X., Wu, N., Yang, W., Lai, Z., Lek, S., 2016. Spatio-temporal patterns and predictions of phytoplankton assemblages in a subtropical river delta system. Fund. Appl. Limnol. 187 (4), 335–349. Wei, X., Zhuo, M., Guo, Z., 2008. Particulate organic carbon-isotope composition and its spatio-temporal variability of suspended sediment in the Xijiang River, South China. Ecol. Environ. 17, 2127–2131 (in Chinese). Yan, S., 1989. A study on the aquatic vascular plants of the Pearl River basin in Guangdong. Acta Hydrobiol. Sin. 13, 305–311 (in Chinese). Zeug, S.C., Winemiller, K.O., 2008. Evidence supporting the importance of terrestrial carbon in a large-river food web. Ecology 89, 1733–1743. Zhang, Y., Ran, Y., 2014. Stable carbon and nitrgen isotopic, amino acids and lignin compositions and geochemical significance of particulate organic matter from the middle and lower reaches of the Pearl River. Geochemica 43, 114–121 (in Chinese).

Appl. Ecol. 46, 443–453. Cole, J.J., Solomon, C.T., 2012. Terrestrial support of zebra mussels and the Hudson River food web: a multi-isotope, Bayesian analysis. Limnol. Oceanogr. 57, 1802–1815. Copp, G.H., Guti, G., Rovný, B., Černý, J., 1994. Hierarchical analysis of habitat use by 0+ juvenile fish in Hungarian/Slovak flood plain of the Danube River. Environ. Biol. Fish. 40, 329–348. Copp, G.H., 1992. Comparative microhabitat use of cyprinid larvae and juveniles in a lotic floodplain channel. Environ. Biol. Fish. 33, 181–193. Delong, M., Thoms, M., 2016. Changes in the trophic status of fish feeding guilds in response to flow modification. J. Geophys. Res. Biogeosci. 121, 949–964. Douglas, M.M., Bunn, S.E., Davies, P.M., 2005. River and wetland food webs in Australia’s wet- dry tropics: general principles and implications for management. Mar. Freshw. Res. 56, 329–342. Fry, B., 2013. Alternative approaches for solving underdetermined isotope mixing problems. Mar. Ecol-Prog. Ser. 472, 1–13. Fujibayashi, M., Sakamaki, T., Shin, W., Nishimura, O., 2016. Food utilization of shellattached algae contributes to the growth of host mud snail, Bellamya chinensis: evidence from fatty acid biomarkers and carbon stable isotope analysis. Limnol. Ecol. Manag. Inland Waters 57, 66–72. Garner, P., 1996. Microhabitat use and diet of 0+ cyprinid fishes in a lentic regulated reach of the River Great Ouse, England. J. Fish Biol. 48, 367–382. Giarrizzo, T., Schwamborn, R., Saint-Paul, U., 2011. Utilization of carbon sources in a northern Brazilian mangrove ecosystem. Estuar. Coast. Shelf Sci. 95, 447–457. Gu, B., Schelske, C.L., Waters, M.N., 2011. Patterns and controls of seasonal variability of carbon stable isotopes of particulate organic matter in lakes. Oecologia 165, 1083–1094. Hamilton, S.K., Lewis, W.M., Sippel, S.J., 1992. Energy sources for aquatic animals in the Orinoco River floodplain: evidence from stable isotopes. Oecologia 89, 324–330. Herzka, S.Z., 2005. Assessing connectivity of estuarine fishes based on stable isotope ratio analysis. Estuar. Coast. Shelf Sci. 64, 58–69. Junk, W.J., Bayley, P.B., Sparks, R.E., 1989. The flood pulse concept in river-floodplain systems. Can. J. Fish Aquat. Sci. 106, 110–127. Kaymak, N., Winemiller, K.O., Akin, S., Altuner, Z., Polat, F., Dal, T., 2015. Stable isotope analysis reveals relative influences of seasonal hydrologic variation and impoundment on assimilation of primary production sources by fish in the Upper Yesilırmak River, Turkey. Hydrobiologia 753, 131–147. Lewis, W.M., Hamilton, S.K., RodrÍguez, M.A., Saunders III, J.F., Lasi, M.A., 2001. Food web analysis of the Orinoco floodplain based on production estimates and stable isotope data. J. N. Am. Benthol. Soc. 20, 241–254. Li, J., Li, X., Jia, X., Li, Y., He, M., Tan, X., Wang, C., Jiang, W., 2010. Evolvement and diversity of fish community in Xijiang River. J. Fish. Sci. China 17, 298–311 (in Chinses). Lu, K., 1990. Fishery Resources of the Pearl River. Guangdong Science and Technology Press, Guangzhou, China (in Chinese). Miyazono, S., Taylor, C.M., 2013. Effects of habitat size and isolation on species immigration-extinction dynamics and community nestedness in a desert river system. Freshw. Biol. 58, 1303–1312. Moore, J.W., Semmens, B.X., 2008. Incorporating uncertainty and prior information into stable isotope mixing models. Ecol. Lett. 11, 470–480. Nilsson, C., Reidy, C.A., Dynesius, M., Revenga, C., 2005. Fragmentation and flow regulation of the world's large river systems. Science 308, 405–408. Nyunja, J., Ntiba, M., Onyari, J., Mavuti, K., Soetaert, K., Bouillon, S., 2009. Carbon sources supporting a diverse fish community in a tropical coastal ecosystem (Gazi Bay, Kenya). Estuar. Coast. Shelf Sci. 83, 333–341.

7