Background

Diatoms play a particularly important role in the biogeochemical cycle [1] of primary elements such as carbon, nitrogen, phosphorus, and silica, contributing about 20–25% of global primary production [2]. Diatoms are ubiquitous and diverse species of single-celled, eukaryotic, photosynthetic microorganisms [3], and are often the dominant primary producers in marine and freshwater ecosystems [4]. Therefore, diatoms in such ecosystems may be remarkably dissimilar either in phylogenetic composition or biogeographic distribution [5, 6]. Freshwater bodies typically consist of lentic (particularly lakes and wetlands) and lotic waters (including streams and rivers), which are often dominated respectively by planktonic algae and benthic species [7].

Accurate identification of diatoms depends on the reliability of the methods used. Morphological analysis requires extensive taxonomic expertise and may exhibit shortcomings in characterizing specific diatoms in rivers [8]. With the development of high-throughput sequencing (HTS) technology, DNA metabarcoding has become a rapid, accurate, and reliable method for diatom detection [9]. Various DNA barcoding studies have been successfully conducted, based on different maker genes, including COI [10], ITS [11], and 18S rDNA [9, 12]. Malviya et al. [13] provided a new estimate of diversity and distribution of marine planktonic diatoms based on the V9 region of eukaryotic 18S rDNA. As a result, the most widespread and diverse diatom genera are derived from 46 marine stations. Recently, the V4 region of 18S rDNA was proposed for diatom barcoding in studies of diatoms in river and deltaic systems [9, 12].

Comparing the numerous studies of diatoms and eutrophication in oceans [14, 15] and lakes [16, 17] to date, it is clear that the present understanding of diatoms is relatively poor for lotic and oligotrophic rivers [7]. In fact, previous reports on the dynamics of riverine diatoms have mostly focused on tributaries, small rivers, reaches, stations, and estuaries [18,19,20]. Many studies examined the diversity and composition of planktonic [19, 20] or benthic diatoms [18, 21, 22] based on morphological identification. For example, Centis et al. [20] investigated planktonic diatoms dominated by physical constraints at two stations of the River Adige, Italy. Liu et al. [18] investigated the community structure of benthic diatoms in the Dong River, one of the three main tributaries of the Pearl River, China. Although Kireta et al. [23] observed that both planktonic and periphytic diatoms could be used as bio-indicators of river conditions, little is known about the distinction between planktonic and benthic diatoms regarding their spatiotemporal distributions.

Biogeography studies aim to reveal the spatial and temporal distribution of biodiversity and provide insight into the mechanisms that generate and sustain diversity [24]. Spatial dispersal and environmental selection processes are regarded as essential drivers for the biogeographical pattern of bacterial community [25]. The former promotes movement of species and their establishment at a new location, whereas the latter alters the abundance and composition of species, according to the ability to survive and reproduce under local environmental conditions. A similar explanation has been proved to apply to the biogeographical pattern of planktonic or benthic diatom communities in small rivers using morphological analysis [26,27,28]. However, it remains unclear how the integrated spatiotemporal distributions of planktonic and benthic diatom communities are shaped by spatial dispersal and environmental selection processes in large rivers subject to complex natural and anthropogenic impacts.

To close the above gap, we implemented large-scaled synchronous monitoring of diatom communities at 62 hydrologic stations over a 6030 km continuum of the Yangtze River in China. Consequently, we provided the first molecular biogeographic pattern of both planktonic and benthic diatoms in the largest river in Asia (Fig. 1). Meanwhile, environmental drivers of diatom communities were interpreted in terms of photosynthetic active radiation, temperature, channel slope, and nutrient conditions under varying landforms.

Fig. 1
figure 1

Flowchart of the study. a Two hundred seventy-nine water and sediment samples at 62 hydrologic stations in the Yangtze River covering the actual sinuous channel reach of length 6030 km (equivalent to 1.83 times the 3290 km straight line from start to end sampling sites). b Metabarcoding analysis provides insights into biogeographic pattern of diatoms along the mainstream of the Yangtze River, represented by the spatial distribution of Shannon diversity. c Interpretations on the biogeographic patterns of diatom communities, with main influencing factors such as photosynthetically active radiation (PAR), channel slope, and nutrients characterized by ratio of total nitrogen to total phosphate (TN:TP)

Full size image

Results

Our study generated a total of 8,602,620 V4 18S rDNA reads from 279 samples. All sequencing reads were classified into 3947 operational taxonomic units (OTUs) at a 97% similarity threshold, with 3144 OTUs well matching 454 diatom species in our reference database. Rarefaction curves (Additional file 1: Figure S1) together with high values of Good’s coverage ranging from 0.9854 to 0.9992 illustrated that OTUs obtained by the current sequencing depth gave a reasonable representation of the diatom communities. The phylogeny tree, constructed by representative OTUs (accounting for > 90% sequence in all samples) and reference sequences (Additional file 1: Figure S2), further confirmed the accuracy of taxonomic assignment.

Alpha and beta diversity of diatom communities

Molecular barcoding based on high-throughput sequencing (HTS) provided a detailed diatom directory for the whole Yangtze River at different taxonomy levels, i.e., 4 classes, 37 orders, 60 families, and 152 genera.

HTS is of particular use in detecting nano-sized diatoms (2–20 μm) in the Yangtze River, confirming the presence of Fragilaria perminuta, Achnanthidium minutissimum, Achnanthidium saprophilum, Amphora pediculus, Fistulifera saprophila, Mayamaea permitis, Sellaphora seminulum, Encyonema minutum, Fragilaria famelica, Fragilaria rumpens, Gomphonema pumilum, Staurosirella pinnata, Planothidium frequentissimum, Craticula buderi, and Craticula molestiformis.

Six types of environmental samples were taken along the Yangtze River, including water and sediment samples from the river source region (i.e., water-plateau (12 samples) and sediment-plateau (12 samples)) and those from the mainstream in the non-plateau area (i.e., water-spring (38 samples), water-autumn (46 samples), sediment-spring (87 samples), and sediment-autumn (84 samples)). Planktonic diatoms exhibited the highest alpha-diversity (Chao1 and Shannon indices) and benthic diatoms the lowest richness (Chao1) in the plateau (Additional file 1: Figure S3). In the non-plateau area, no significant differences were observed in the alpha richness and diversity of diatom communities in the four sample types.

Non-metric multidimensional scaling (NMDS) analysis of the compositional dissimilarities between diatom communities demonstrated not only a clear spatial differentiation in diatoms between the plateau and the main body of the Yangtze, but also a division between planktonic and benthic groups (Additional file 1: Figure S4). Water and sediment samples between spring and autumn in the non-plateau area were used for further seasonal analysis. Seasonal difference in planktonic diatoms was found much more significant than in benthic diatoms, as further confirmed by an analysis of similarity (ANOSIM) test (Additional file 1: Figure S5). Moreover, one-way analysis of variance (one-way ANOVA) indicated that more planktonic diatoms (42.75 ± 13.98% relative abundance, primarily belonging to Cyclotella, Stephanodiscus, and Skeletonema) than benthic diatoms (16.58 ± 5.06% relative abundance, primarily belonging to Pinnularia and Stephanodiscus) exhibited significant seasonal sensitivity (Additional file 1: Figure S6).

Biogeographic patterns of diatom communities

A variety of diatom species have been found closely relevant to carbon export [29]. In the Yangtze River, the planktonic diatoms such as Asterionella formosa, Diatoma vulgare, Lindavia viaradiosa, Gomphonema pumilum, and Thalassiosira nordenskioeldii were significantly strongly associated with dissolved carbon dioxide (pCO2, see the “Methods” section) (Spearman r > 0.3, P < 0.05), while the benthic diatoms Asterionella formosa, Encyonema prostratum, Eucocconeis laevis, Fistulifera saprophila, and Nitzschia sigmoidea were highly correlated with pCO2 (Additional file 1: Figure S7).

Obvious difference in species composition was observed in planktonic and benthic diatoms. In the Yangtze River, diatoms mainly consisted of Coscinodiscophyceae, Fragilariophyceae, Bacillariophyceae, and Mediophycea. Planktonic diatoms were dominated by Coscinodiscophyceae (about 43.76% of the total number of sequences) and Mediophyceae (17.91%), while benthic diatoms were dominated by Bacillariophyceae (54.88%) and Coscinodiscophyceae (30.96%) (Additional file 1: Figure S8). Planktonic and benthic diatoms were not always consistent in dominant genera (top 20, the relative abundance ranged from 55.6 to 83.6%) (Fig. 2a). The dominant genera were found to be Cyclotella, Stephanodiscus, Pinnularia, and Paralia, represented 12.2, 8.6, 7.3, and 6.6% of total sequences, respectively, in water samples. Meanwhile, Navicula, Pinnularia, and Cyclotella became the dominant genera, represented 14.4, 9.1, and 6.9% of total sequences, respectively, in sediment samples, in which Navicula was dominant in either sediment-plateau (17.1%), sediment-autumn (13.1%), or sediment-spring samples (13.2%).

Fig. 2
figure 2

a Circular visualization of dominant diatoms at genus level in six sample types. Inner circular diagram shows relative abundance of different diatom genus in six sample types. Only the dominant genus with a mean relative abundance of ≥ 1% in all samples is depicted. The width of ribbons for each diatoms is directly proportional to their relative abundance in each sample type. Similarly, different colored ribbons of different width for each sample type describe the distribution of different genera. b Representative diatoms genera in different landform types from the river source to mouth along the Yangtze River

Full size image

Over the 6030 km continuum from river source to mouth, landform played a significant role in spatial differentiation of both planktonic and benthic diatom communities. Referring to a previous study on landform types in the Yangtze [30], planktonic diatoms were represented by Cymbella, Asterionella, Stephanodiscus, Melosira, Cyclotella, and Conticribra in the plateau, mountain, foothill, basin, foothill-mountain, and plain regions, respectively, while benthic diatoms were abundant by Cymbella, Navicula, Melosira, Conticribra, Cyclotella, and Surirella, respectively, in the corresponding regions (Fig. 2b).

A completed description on biogeographic pattern over a large river requires to identify the difference in diatom compositions among different types of samples and their spatiotemporal heterogeneity. Using the indicator species analysis, the diatoms that were responsible for the observed community differences among the six types of samples could be well identified (Additional file 1: Table S1). The number of indicator diatom species in the river ranged from 6 (sediment-spring) to 41 (water-plateau). Diatom communities in the plateau region were quite different from those in the non-plateau region of the Yangtze River, as evidenced by the higher percentage of top indicator species in water-plateau and sediment-plateau samples (Additional file 1: Figure S9). The average relative abundance of indicator species in the source area exceeded 40%, and planktonic indicator species contributed more reads than benthic indicator species. Furthermore, a number of indicator species belonging to Tabellariales and Hemiaulales occurred in water-plateau and sediment-plateau samples, respectively.

Diatom composition in terms of ecological guilds showed spatial dissimilarity in water and sediment samples (Additional file 1: Figure S10). Diatoms were divided into four ecological guilds according to their biological traits, including low-profile, high-profile, motile, and planktic guilds in terms of different responses to nutrients and dynamic disturbances [31,32,33] (see “Ecological guilds classification” section). Benthic diatoms in the motile guild prevailed at most stations along the whole river, whereas those in high-profile and planktic guilds dominated upstream and downstream reaches, respectively. In addition, planktonic diatoms in the planktic guild were predominant at most stations along the Yangtze River.

Environmental effects on diatom biogeography

Significant distance-decay in diatom similarity was observed along the geographical distance (Additional file 1: Figure S11), with a greater slope of the curve for water (slope = − 0.042) than for sediment (slope = − 0.038) using least squares linear regression. The partial Mantel test demonstrated that both geographical and environmental distances played important roles in constraining diatom composition and distribution (Additional file 1: Table S2). Canonical correspondence analysis (CCA) showed significant correlations between diatom communities and specific environmental and spatial factors such as water temperature, pH, suspended solids, and PCNM-1 (Additional file 1: Table S3). Variation partitioning of diatom composition showed that a greater percentage (14.6–21.2%) could be explained by a purely environmental component than that (3.4–6.0%) of the total variation by a pure spatial component (Additional file 1: Figure S12), and a minor portion (0.4–5.4%) explained by spatially structured environmental heterogeneity, leaving the majority of the total variation (68.7–79.0%) inexplicable. As a deterministic process, environmental selection played a critical role in the biogeography of planktonic and benthic diatoms. Although environmental differentiation seemed more important than spatial dispersing in sha** a diatom community, neither could fully explain the total variation in diatom composition. Among others, the typical environmental components such as photosynthetically active radiation, temperature, channel slope, and nutrients condition are essential to diatom community accompanied with the spatial dispersal.

Photosynthetically active radiation (PAR, 400–700 nm) is utilized by diatoms to synthesize biomass through photosynthesis [34]. Spatially, the annual-averaged PAR exhibits four stages along the Yangtze River [35], i.e., the highest in the upper reach located in Qinghai-Tibet Plateau region (above 32 mol m−2 d−1), the higher in the reach located in the Hengduan Mountains, the lowest in the reach located in the Sichuan Basin (below 23 mol m− d−1), and the moderate in the lower reach (Fig. 3c, see the “Photosynthetically active radiation (PAR) divisions” section). For a better understanding of the spatial heterogeneity of both planktonic and benthic diatom communities, LefSe analysis was used considering its advantages in identifying differentially abundant taxa under different environmental conditions [36]. Consequently, preferred planktonic and benthic diatom species in different PAR regions were identified (Fig. 3a, b). For example, the Caloneis, Cymbella, Fistulifera, and Fragilaria genera preferred very-high PAR zones, the Papiliocellulus genus favored medium PAR regions, and Conticribra and Cyclotella showed a preference for low PAR habitats. Planktonic Cymatopleura and Navicula, and benthic Asterionella, Biddulphia, Diatoma, and Encyonema genera preferred to high PAR conditions. Moreover, water temperature is a key environmental factor in structuring diatom community assemblages through its influence on diatom size and growth rate [37] in the Yangtze River (Additional file 1: Table S3). Although the richness of planktonic diatoms seemed to fluctuate with PAR, the richness of benthic diatoms tended to rise with increasing temperature (Fig. 3c).

Fig. 3
figure 3

LEfSe cladogram of planktonic (a) and benthic (b) diatom communities from four PAR regions. Diatom taxa with a mean relative abundance of ≥ 0.1% in all samples, assigned to kingdom (innermost), phylum, class, order, family, and genus (outermost), are used to determine taxa or clades most likely to explain differences between PAR regions. Differentially abundant taxa (biomarkers) are colored by their most abundant PAR regions, i.e., red, green, blue, and purple circles stand for biomarkers in regions of very high, high, medium, and low. Orange and blue circles display the average alpha-diversity (Chao1) of planktonic and benthic diatoms respectively in different photosynthetically active radiation (PAR) regions, and their sizes correspond to the Chao1 index (c)

Full size image

Stream power, often simply characterized by the river channel slope or the product of channel slope and flow discharge (except in plateau regions) [38], is another important factor altering the spatial distribution of diatoms. The channel slope dramatically changes along the Yangtze River, primarily due to the basis of geology, climate and geomorphology. In the mountainous reaches (stations 1~2), the river channel slope can be higher than 400 × 10−5. In the upper reaches (stations 3~14), the channel slope has dropped sharply to 10–30 × 10−5. In middle-lower reaches (stations 15~24), the slope of the riverbed is nearly to zero as the channel widens and shallows in the estuarine region. In general, the varying channel slope along the Yangtze River could be simplified into three stages, steep slope in mountainous reaches, moderate slope in upper reaches, and mild slope in middle-lower reaches (Fig. 4c). During the wet season (autumn), the higher flow discharge weakens the correlation between planktonic community similarity and channel slope, although a stronger correlation between benthic community similarity and channel slope is maintained due to higher mobility of the streambed (Fig. 4a, b). In view of their relative abundance, planktonic diatoms were characterized by Psammothidium, Nitzschia, and Cymbella for steep slope environments, Papiliocellulus for moderate slope, Mayamaea, Pinnularia, and Surirella for mild slope environments. Benthic diatoms were represented by Cocconeis, Entomoneis, and Melosira for steep slope environments, Fallacia, Psammothidium, and Skeletonema for moderate slopes, and Actinocyclus, Aulacoseira, and Conticribra for mild slopes (Additional file 1: Figure S13). Furthermore, slope effects on diatoms might be identified in terms of ecological guilds. Regardless of the diatoms in an unspecified ecological guild, planktonic diatoms were dominant in the planktic guild. Interestingly, species in motile guild stably constituted the main component of benthic diatoms in the whole lotic river (Fig. 4c).

Fig. 4
figure 4

Relationships between community similarity and river channel slope for water-spring (a), water-autumn (b), sediment-spring (c), and sediment-autumn (d) samples. Values of Mantel Spearman correlations (r) and probabilities (P) are also provided. Gray lines denote ordinary least squares linear regression fits across all samples. Spatial distributions of ecological guilds for different channel slopes are shown in (e)

Full size image

Nutrient condition, represented by the total nitrogen to total phosphorus ratio (TN:TP), was considered as an indicator of ecosystem to support for algae biomass [39]. Since diatoms were reported to predominate in phytoplankton at high nitrogen to phosphorus ratio (N:P > 16) in water [40], we investigated the response of planktonic and benthic diatoms to TN:TP in the Yangtze River. Higher TN:TP was observed in water samples (13.8~45.63) than in sediment samples (0.004~0.65). The spatial distribution of TN:TP in water and sediment samples varied in different sampling stations. The lowest value of TN:TP in water samples was obtained in river reaches receiving large inflow from the Min River, and the highest value was obtained in both water and sediment samples near river estuary. Moreover, TN:TP exhibited a fluctuation between upstream and downstream of large dams (** and draining water while the syringe is pointing upwards. Dissolved CO2 was extracted by transferring 25 ml of ultra-high purity nitrogen at the field sites and then equilibrated with the headspace in the sample syringe by vigorous shake for 5 min. After equilibration, the headspace gas was immediately transferred to a pre-vacuum glass storage vial equipped with chlorobutyl septa. Finally, the CO2 partial pressure was measured using a gas chromatograph equipped with a thermal conductivity detector.

The information on channel slope in the Yangtze was sourced from Chen et al. [70]. Nutrients condition was represented by the atomic ratio of nitrogen to phosphorus. For water samples, we utilized dissolved total nitrogen (TN) and total phosphorus (TP) to calculate annual-averaged TN:TP for 2005 to 2014 and monitored TN:TP observed at 50 stations for 2014. For sediment samples, we utilized TN and TP to calculate monitored TN:TP observed at 50 stations for 2014.

DNA extraction, PCR amplification, and sequencing

DNA was extracted in triplicate using the FastDNA® SPIN Kit for Soil (MP Biomedicals, USA) following the manufacturer’s instructions. The triplicate DNA extracts were mixed together for later PCR amplification. Amplification of the V4 region of the 18S rDNA was performed by polymerase chain reaction (PCR) using barcoded primers DIV4for (5′-GCGGTAATTCCAGCTCCAATAG-3′) and DIV4rev3 (5′-CTCTGACAATGGAATACGAATA-3′) [12], where barcode is an eight-base sequence unique to each sample. Amplification was conducted under the following conditions: initial denaturation at 94 °C for 2 min, then 32 cycles of denaturation at 94 °C for 45 s, annealing at 50 °C for 45 s, elongation at 72 °C for 60 s, and final extension at 72 °C for 10 min, 10 °C until halted by user. PCR mixtures (20 μL volume) were prepared in triplicate contained 2 μL of 10 × buffer, 2 μL of 2.5 mM dNTPs, 0.8 μL of each primer (5 μM), 0.2 μL of rTaq polymerase, 0.2 μL of BSA, and 1 μL of 10 ng DNA sample. Amplicons were purified using the AxyPrep DNA Gel Extraction Kit (Axygen Bioscience, Union City, CA, USA) according to the manufacturer’s instructions and quantified using QuantiFluorTM-ST (Promega, USA). Adaptor was ligated onto the amplicons for the library construction. Afterwards, sample libraries were pooled in equimolar amounts and sequenced on Illumina MiSeq 2 × 250 PE platform (Majorbio Company, Shanghai, China).

Three negative control samples were used to monitor any contamination during the molecular workflow, negative filtration, DNA extraction, and PCR controls; however, no quantifiable DNA was detected for further analysis.

Bioinformatics analysis

Sequences of diatom 18S rDNA were quality-filtered using QIIME [71] as follows: (i) minimum sequence length of 300 bp, and minimum threshold quality score of Q20; (ii) maximum mismatches of 2 for matching the primer; any reads with ambiguous bases were removed; and (iii) merged pair-ended sequences that overlapped longer than 10 bp into a single sequence. UCHIME was used to remove chimeric sequences and UPARSE was used to cluster operational taxonomic units (OTUs) with 97% similarity cutoff [72].

We built a reference database of 18S rRNA reads composed of 4573 unique diatom sequences. First, we extracted all diatom sequences of 18S rRNA reads from GenBank (http://www.ncbi.nlm.nih.gov/). Second, short reads (less than 100 nucleotides) were refused access to the reference database, and redundant reads were eliminated by cd-hit to increase the taxonomy identification accuracy. Third, sequence alignment was performed by Mafft (ver 7.310) [73], then the sequences were analyzed to construct an approximately-maximum-likelihood phylogenetic tree using FastTree (ver 2.1.10) [74], and any incorrect reads discarded. Finally, a total of 4573 unique sequences were retained in our reference database.

To identify taxonomically OTUs obtained in this study against known diatom species, the BLASTN [75] program was applied to align clean 18S rRNA reads to the corrected diatom database. Those OTUs with the best BLAST hit scores, not only an e value ≤ 10−5 but also identity ≥ 80% with respect to the reference sequence were firstly selected. Then, the selected OTUs were checked by means of the phylogenetic tree, and only OTUs with correct taxonomical assignment were retained for further analysis. Clean reads were further assigned to known diatom species based on our reference database.

To estimate the community structure for each site, the Mothur program [76] was used to normalize all data sets with respect to the least-well-represented data set (11049 sequences). Alpha diversity indices (chao1, Shannon, and Goods coverage) were calculated using QIIME.

Statistical analysis

Diatom species that characterize each sample group were identified with indicator species analysis using labdsv and indval packages in R software [77]. Indicator values were calculated based on the relative frequency and relative average abundance of a given species in six types of environmental samples. Species with indicator value ≥ 0.3 and p value ≤ 0.01 were defined as indicator species at class, order, family, and genus levels. Nonmetric multidimensional scaling (NMDS) was performed to visualize the dissimilarity of different samples based on Bray-Curtis similarity matrices. Analysis of similarity (ANOSIM) was conducted to test the significance of differences among a priori sampling groups based on environmental parameters. NMDS and ANOSIM statistics were carried out using the vegan package in R. The linear discriminant analysis effect size (LEfSe) [36] was used with Kruskal-Wallis and Wilcoxon tests to discover high-dimensional biomarker and explain taxa difference at different environment conditions of PAR or channel slope. The LEfSe biomarker detection was performed in QIIME [71] using the logarithmic LDA threshold > 4 and the statistical parameters of P < 0.05. One-way analysis of variance (one-way ANOVA) was carried out to test significance of group differences using the vegan package in R.

Distance-decay patterns of diatom community similarity were described by considering geographical distance and environmental distance from the site location to river mouth among sample sites. Mantel tests were used to examine the Spearman’s rank correlation between geographical and environmental distance and diatom community similarity using Bray-Curtis distance matrices with 999 permutations in R. The geographical distance of each sampling site was calculated using ArcGIS V10.3 software. The environmental distance matrix (normalized Euclidean distance) was generated with a normalized combination of environmental variables such as water temperature, COD, SS, DO, pH, NH4-N, NO3-N, TN, TP, and DOC for water samples as well as TOC, pH, NH4-N, NO3-N, TN, and TP for sediment samples. The rate of distance-decay of diatom communities was calculated as the slope of ordinary least-squares regression line fitted to the relationship between geographic distance and community similarity. Partial Mantel tests were conducted to assess the pure effects of geographical distance (controlling for environmental distance) and environmental distance (controlling for geographic distance) on diatom community similarity with 9999 permutations.

A set of spatial variables was generated through the use of principal coordinates of neighbor matrices (PCNM) analysis based on the longitude and latitude coordinates of each sampling site [78]. The function “envfit” was run with 999 permutations to select significant environmental variables (P < 0.05). Significance testing was then assessed using the “permutest” function based on 999 permutations in R, while canonical correspondence analysis (CCA) was performed to determine the effects of selected environmental and spatial variables on diatom communities (Additional file 1: Table S3). Partial canonical correspondence analysis (pCCA) was performed to decompose the total variation in diatom community into a pure environmental component, a pure spatial component, a spatially structured environmental component, and residual variation.

Ecological guild classification

Based on their ecological characteristics, diatom species are classified into four ecological guilds (low profile, high profile, motile, and planktic guilds) [31,32,33], which are expected to respond in different ways to nutrients’ conditions and physical disturbances. A low-profile guild is defined as having high reproduction rate, low nutrient and light availability, and slow-moving diatoms. A high-profile guild possesses characteristics of high resource availability and low disturbance. A motile ecological guild has the ability to move fast and choose the best microhabitat in a given circumstance. A planktic guild adapts to lentic environments and resists sedimentation. We extended these guilds by adding supplemented classifications used in other studies [79, 80].

Photosynthetically active radiation (PAR) divisions

Solar radiation with wavelengths (400–700 nm), called photosynthetically active radiation (PAR), is able to be utilized by plants and algae through photosynthesis to convert light energy into biomass [34]. Monteith reported the linear correlation between net primary production (NPP) and PAR absorbed by green foliage [81]. Zhu et al. [35] also suggested that the spatial distribution of annual-averaged PAR is complex and inhomogeneous across China, using calculated PAR spatial data for the period 1961–2007 provided by China Meteorological Administration.

Zhu et al. [35] calculated and spatialized PAR using data simulation method [82] based on three climatic datasets, i.e., daily sunshine duration data at 740 weather stations across China for 1961–2007 and global radiation data at 122 radiation stations across China for 1961–2000 from China Meteorological Administration, and PAR observatory data at 36 field stations across China for 2004–2007 from Chinese Ecosystem Research Network. Then, PAR along the Yangtze River could be further derived. Thus, we define four zones of PAR intensity in different regions across the Yangtze River basin as follows:

  1. I.

    Very high, PAR > 32 mol m−2 d−1;

  2. II.

    High, 26 < PAR ≤ 32 mol m−2 d−1;

  3. III.

    Medium, 23 < PAR ≤ 26 mol m−2 d−1;

  4. IV.

    Low, PAR ≤ 23 mol m−2 d−1.