4. DISCUSSION
Hydrologic connectivity disruption triggered by dams can affect flow regimes, habitat and physicochemical parameters of river systems (Tiemann et al., 2004; Yan et al., 2013; Hitchman et al., 2018). Previous studies have shown that low-head dams may modify local habitats in streams, including deeper water, slower flow, smaller substrate in impoundments upstream, and faster flow and larger substrate in the plunging areas downstream (Yan et al., 2013; Fencl et al., 2015; Li et al., 2021). Furthermore, the PERMDISP analyses suggested that the local habitat heterogeneity did not significantly differ between the impoundments and free-flowing segments, and the habitat differentiation in impoundments was slightly higher than that of the free-flowing segments. Habitat conditions are highly heterogeneous along the upstream-downstream gradients of streams, and the larger spatial scales should encompass greater environmental heterogeneity (Vannote et al., 1980; López-Delgado et al., 2019). When all spatial units share the same range of habitat characteristics at relatively large scales, the common changes of habitat conditions in impoundments (i.e., water-depth increasing, flow slowing and substrate-size decreasing) may reduce the spatial heterogeneity and further cause habitat differentiation among different impoundments (Bu et al., 2017). In contrast, dams would result in greater environmental differences and induce habitat differentiation across impoundments at smaller scales (Smith & Mather, 2013; Liu et al., 2019), which is due to that the degree to which local habitat is modified often varies with dam size and operation (Poff & Hart, 2002; Li et al., 2021). In this study, all the 53 low-head dams surveyed were located within the first-order headwater streams in Wannan Mountains areas, therefore, the slight habitat differentiation in impoundments was observed at finer spatial scales.
As low-head dams are relatively small with weaker storage capability, the extent and degree to which low-head dams altered the local fish assemblages could be relatively weak (Yan et al., 2013; Fencl et al., 2015; Brewitt & Colwyn, 2020). Although a fraction of native species lost and native-invasive species gained in impoundments, a large number of the remaining species (especially for abundant species) often tend to decline or increase in abundance (Tiemann et al., 2004; Liu et al., 2019). In terms of species composition, twenty-nine species occurred in both the impoundments and free-flowing segments, and just a few species gained (C. idella and M. swinhonis ) and lost (L. guilinensis and P. assimilis ) compared with that of free-flowing segments (Table 1). However, changes in species abundance can also affect the degree to which communities differ in composition despite pairwise communities with the same species identities (Cassey et al., 2008; Barwell et al., 2015). We also found the significant changes in community composition between the impoundments and free-flowing segments. Compared with that of free-flowing segments, some native species such as Z. platypus , Ctenogobius sp. , and V. stenosoma decreased in abundances, and other native-invasive species such as M. anguillicaudatus , O. potamophila , and C. auratusincreased (Table 2). Earlier works have shown that the transformation of lotic to lentic habitats in impoundments created by low-head dams can not only cause native fish species population declining and even local extinguishing but also favor the occurrence and spread of native invaders into upland streams (Chu et al., 2015; Smith et al., 2017; Liu et al., 2019).
In this study, our results demonstrated the overall tendency of fish taxonomic differentiation for abundance-based approaches in impoundments relative to free-flowing segments (Figure 4a), which is consistent with the slight habitat differentiation observed in impoundments. Surprisingly, taxonomic homogenization detected is often observed at relative large scales (inter-ecoregion or basins), which is the results of both the establishment of alien/cosmopolitan species and the loss of native species (Toussaint et al., 2016; Daga et al., 2020; Kirk et al., 2020). However, our results are not consistent with those studies at broader spatial scales. First signs of changes in fish diversity patterns affected by low-head dams may be mainly reflected in the abundance changes (Tiemann et al., 2004; Fencl et al., 2015; Liu et al., 2019). If the number of shared species between communities does not change, but the abundances do, the Bray-Curtis index based on abundance approaches will show changes that depend on the magnitude of shifts in abundance (Cassey et al., 2008; Baselga, 2013). Apart from the decline in abundance of shared dominant species (Z. platypus andCtenogobius sp.), formerly dominant species became rarer (V. stenosoma ) and some unique species abundance markedly increased (M. anguillicaudatus and C. auratus ) in impoundments, which may result in taxonomic differentiation of current fish assemblages based on abundance approaches. Additionally, previous studies also revealed that fish assemblages experienced a transition phase from taxonomic differentiation to homogenization over time (Petesse & Petrere, 2012; Pool & Olden, 2012). Given that the alien/cosmopolitan species introduction and colonization may take considerable time to negatively affect native species (Ding et al., 2017), the observed taxonomic differentiation of current fish assemblages is probably an early warning sign of further homogenization.
We found that changes in functional similarity of fish assemblages tended to demonstrate homogenization (Figure 4b), which showed opposite trends to changes in taxonomic similarity. Such patterns are often the result of the decrease in abundance of shared native species with unique functional roles and the establishment and colonization of alien/cosmopolitan species with similar functional roles (Olden & Rooney, 2006; Pool & Olden, 2012). Previous research in this region (Li et al., 2021) has shown that some native species (i.e., typical of lotic species) could bring unique functional traits, and the native-invasive and native species potentially have functional redundancy. In other words, the native species (Z. platypus andV. stenosoma ) with unique functional traits decreased sharply in abundance and some native-invasive species (M. anguillicaudatusand C. auratus ) increased, both of which contributed to functional homogenization despite diverging taxonomies based on abundance approaches. More importantly, even if a significantly positive correlation was observed between changes in mean taxonomic and functional similarities (Figure 5b), approximately a third of pairwise assemblages presented contrasting outcomes (Table 3). The pattern of fish faunal homogenization and differentiation depended in large part on spatiotemporal changes in species abundance, and ecological functions of those species were gained and lost among communities (Cassey et al., 2008; Villéger et al., 2014; Socolar et al., 2016). Taxonomic differentiation but functional homogenization can occur when shared native species with unique functional traits decline in range or abundance, whereas Taxonomic homogenization but functional differentiation can occur when unshared species having similar traits are lost (Villéger et al., 2014; Campbell & Mandrak, 2020). This discrepancy between changes in taxonomic and functional similarities could be explained by functional redundancy among different species, which was supported by previous studies (Villéger et al., 2014; Su et al., 2015).