Comparison of environmental factors between wet and dry seasons
The environmental factors are summarized in Additional file 2: Table S1. Half of the 24 environmental factors showed a significant difference between wet and dry seasons (Additional file 2: Table S2). In general, the mean values of temperature, turbidity, suspended solids, electrical conductivity (EC), TOC, NO3-N, and arsenic (As) were significantly higher in the wet season than those of the dry season. On the contrary, pH, dissolved oxygen (DO), total phosphorus (TP), PO4-P, and cadmium (Cd) showed higher mean values in the dry season.
Comparison of microeukaryotic community alpha diversity between wet and dry seasons
Rarefaction curves for individual and combined set of 60 samples showed that most samples from either wet or dry seasons tended to approach saturation, and wet microeukaryotic communities showed higher species richness with the same sequencing effort (Additional file 1: Figure S1). Good’s coverage ranged from 97.68 to 99.31% in the individual samples, and the index of all 60 samples combined was 99.98% (Additional file 2: Table S3). The rarefaction curves, extrapolated species richness indices (ACE and Chao 1), and Good’s coverage indices indicated that the majority of the microeukaryotic taxa had been recovered from the studied metacommunity (Additional file 1: Figure S1 and Additional file 2: Table S3). In general, the microeukaryotic metacommunity in the wet season showed higher alpha diversity compared with those in the dry season.
A total of 17,416 microeukaryotic OTUs were identified from 6,623,640 high-quality sequences at 97% similarity level for the 60 samples, and all microeukaryotic taxa were divided into five categories (conditionally abundant taxa, moderate taxa, always rare taxa, conditionally rare taxa, and conditionally rare and abundant taxa) since no OTU was detected as always abundant taxa (Additional file 2: Table S4).
Comparison of microeukaryotic community composition between wet and dry seasons
Overall, most diverse OTUs in both wet and dry seasons were assigned to the groups of Stramenopiles, Alveolata, and Fungi, and the OTUs of Stramenopiles were significantly more diverse in the wet than in the dry seasons. In addition, most abundant OTUs were assigned to Alveolata, Animalia, other eukaryota (incertae sedis eukaryota), and Stramenopiles. The proportions of sequences of Alveolata, Animalia, Discoba, Holozoa, and Stramenopiles were significantly higher in wet than in dry seasons, whereas Chloroplastida, other eukaryota, and Fungi showed higher relative abundance in dry than in wet seasons (Additional file 1: Figure S2A).
Our PCoA ordinations and ANOSIM tests showed that the microeukaryotic community compositions (including total, dominant, always rare, and conditionally rare taxa) from wet and dry seasons exhibited a clear separation (R > 0.423, P = 0.001; Fig. 2). Venn diagrams indicated that most OTUs obtained were shared between the two seasons (71.72% for total, 100% for dominant, 64.99% for always rare, and 98.88% for conditionally rare taxa; Additional file 1: Figure S2B). Taken together, there were significant differences in microeukaryotic community composition between wet and dry seasons, although a large proportion of taxa were shared with both seasons in the Tingjiang River.
Environmental and spatial factors related with microeukaryotic community composition
The Mantel tests indicated that turbidity, electrical conductivity (EC), total carbon (TC), nitrogen (TN and NH4-N), phosphorus (TP and PO4-P), and arsenic (As) were significantly associated with the microeukaryotic communities (including total, dominant, always rare, and conditionally rare taxa; P < 0.05, Additional file 2: Table S5) in the wet season, while water temperature, turbidity, suspended solids, carbon (TOC), nitrogen (TN, NH4-N, NO2-N, and NO3-N), total phosphorus (TP), zinc (Zn), and arsenic (As) were significantly correlated to the microeukaryotic communities in the dry season (P < 0.05, Additional file 2: Table S5).
The two different spatial predictors (dendritic distance and Euclidian distance between sampling sites) of microeukaryotic community compositions showed similar results (Fig. 3, Additional file 1: Figure S3, and Additional file 1: Figure S4). There were significant and negative relationships between the geographical distance and Bray-Curtis similarity of microeukaryotic communities (including total, dominant, always rare, conditionally rare taxa) in both wet and dry seasons (P < 0.01; Fig. 3, Additional file 1: Figure S3). Further, the biogeography of both the always rare taxa and conditionally rare taxa was similar to that of the dominant and total microeukaryotic communities. Additionally, Spearman’s rank correlations between the Bray-Curtis similarity of river microeukaryotes (from genus to kingdom levels) and two types of spatial distances (dendritic and Euclidian distances) also showed a significant distance-decay relationship (Additional file 1: Figure S4). All of the microeukaryotic subcommunities exhibited a robust distance-decay pattern, whereby community dissimilarity increased with geographical distance, indicating that these distinct microeukaryotic taxa under different hydrologic regimes (i.e., wet and dry seasons) exhibited general similar spatial pattern. Further, among the correlations between spatial distances (either dendritic or Euclidian) and microeukaryotic communities, all of the microeukaryotic communities exhibited a stronger distance-decay pattern in the dry season than in the wet season (Fig. 3, Additional file 1: Figure S3, and Additional file 1: Figure S4).
Relative contribution of environmental and spatial factors on microeukaryotic distribution
The variance partitioning analyses (VPA) results showed that only a small proportion of microeukaryotic community variation was explained by environmental and spatial factors (Additional file 2: Table S6, Additional file 2: Table S7, and Additional file 2: Table S8). Both variance partitioning analyses (VPA) using environmental vs. principal coordinates of neighboring matrices (PCNM), and environmental vs. asymmetric eigenvector map (AEM) variables revealed similar results in both the wet and dry seasons. The principal coordinates of neighboring matrices (PCNM) analysis, represents the geographical distance matrix as Euclidean distance between each pair of sampling sites, while the asymmetric eigenvector map (AEM) analysis represents the network distance matrix between two sites along the river.
Our VPA performed on environmental model and directional spatial model (AEM) showed that over the four communities (total, dominant, always rare, and conditionally rare taxa) in the wet season, the pure environmental variables (e.g., temperature, pH, As), the pure spatial processes (the spatial structuring of the microeukaryotic community characterized by AEM variables), and their shared effect explained mean 5.6%, 11.7%, and 8.2% of the community variations, respectively. For dry season, the pure environmental, the pure spatial variables, and the shared effect explained mean 7.9%, 18.2%, and 2.6% of the community variations, respectively. Consequently, a large amount of the variance (on average 74.5% for wet season and 71.3% for dry season, respectively) remained unexplained by the environmental and spatial variables (Additional file 2: Table S6). The VPA based on environmental and PCNM variables (Additional file 2: Table S7 and Additional file 2: Table S8) revealed similar results with VPA based on environmental and AEM variables. The high proportion of unexplained variation in microeukaryotic taxa in wet and dry seasons indicated the potential importance of neutral or stochastic processes for community assembly, especially in wet season.
Fit to the neutral model of community assembly
The neutral community model (NCM) successfully estimated a large fraction of the relationship between the occurrence frequency of OTUs and their relative abundance variations (Fig. 4), with 89.9%, 88.5%, and 89.6% of explained community variance for wet, dry, and both seasons, respectively. Further, the NCM of microeukaryotic community at supergroup levels showed large explained community variance for wet (ranged from 78.8 to 99.3%, with mean value 91.3%) and dry seasons (varied from 67.5 to 97.0%, with mean value 88.8%), respectively (Additional file 1: Figure S5).
More importantly, the explained variation of NCM tended to remain relatively large and constant along taxonomic ranks from genus to kingdom taxonomic resolutions (all of the R2 over than 88.8% with mean value 92.7%), indicating that taxa within the same phylogenetical lineages generally show similar responses to stochastic processes (Additional file 1: Figure S6). These results indicated that stochastic processes were very important in shaping the microeukaryotic community assembly in both seasons. The Nm-value was higher for microeukaryotic taxa in the wet season (Nm = 53,309) than the dry season (Nm = 49,567). Since the number of sequences in both samples was 110,394, the m value was estimated to be 0.483 in wet season and 0.449 in dry season, respectively. These results indicated that species dispersal of planktonic microeukaryotes was higher in the wet than dry seasons.
Neutral and non-neutral partitions are ecologically distinct
For any season, there were a number of microeukaryotic taxa that occurred more or less (above or below neutral prediction) frequently than predicted by NCM given their overall abundance in the metacommunity (Fig. 5). We would consider the taxa that differ significantly from the neutral prediction were due to their different migration or dispersal ability. Points above the prediction represent taxa that are found more frequently than expected, suggesting that they have higher migration ability and can disperse to more locations. Points below the prediction represent taxa found less frequently than expected, suggesting their lower dispersal ability in the river (Additional file 2: Table S9). Another possibility is that this could be due to taxa responding more strongly to local environmental conditions (Additional file 2: Table S10). Partition fractions above the neutral prediction were mainly composed of Centrohelida, other eukaryota, and Holozoa (the estimated migration rate over than 0.6), while the main supergroups of below partition were Animalia, Discoba, and Rhodophyceae (the estimated migration rate less than 0.3, Additional file 2: Table S11).
Taxa from neutral and non-neutral fractions (above and below neutral prediction) showed contrasting community composition, diversity, and abundance (Fig. 5). For wet, dry, and both seasons, we separated the metacommunity into three fractions comprising those OTUs found above, below, and neutral partitions, and analyzed the non-metric multidimensional scaling ordination (NMDS) based on the Bray-Curtis similarity among partitions. The results clearly demonstrated that above, below, and neutral partitions of taxa showed significantly different compositions among wet (global R = 0.891), dry (global R = 0.895), and both (global R = 0.976) seasons at P < 0.01 (Fig. 5a). Further, non-neutral fractions of the metacommunity were also much more heterogenous than the neutral fractions. In addition, the proportions of microeukaryotic OTUs and sequence numbers at supergroup level showed that the neutral fraction accounted for much higher richness and abundance proportions than above and below fractions in both wet and dry seasons (Fig. 5b, c).
Neutral and non-neutral partitions’ microeukaryotic community compositions showed different responses to environmental and spatial factor change. The Mantel test between environmental factors and three microeukaryotic community partitions predicted by the NCM showed that neutral and non-neutral partitions of the metacommunity were significantly associated with both common and different environmental variables in wet and dry seasons. For example, EC and As in the wet season and nitrogen (NH4-N, NO3-N, and NO2-N) and As in the dry season were significantly correlated to both the neutral and non-neutral partitions’ microeukaryotic communities. However, other environmental factors such as water temperature, suspended solids, current velocity, TN, NO2-N, TP and Cd in the wet season, and water temperature, PO4-P and Zn in the dry season were only significantly associated with the non-neutral partition. Further, turbidity was only significantly correlated to neutral partition in the dry seasons (P < 0.05, Additional file 2: Table S10). The neutral and non-neutral partitions’ microeukaryotic taxa also showed distinct degrees of distance-decay patterns, although all of the relationships between microeukaryotic community and geographical distance were significant. And the relationship was stronger in the dry than wet seasons (the absolute r value of Mantel tests between microeukaryotic community and geographical distance was higher in the dry than wet seasons; Additional file 2: Table S10).