Habitat alteration and fragmentation resulting from anthropogenic disturbances greatly contribute to global freshwater biodiversity loss (Erős and Grant, 2015; Barbarossa et al., 2020; He et al., 2021). In a river network, headwaters have unique features and are extremely vulnerable to anthropogenic disturbances (Vannote et al., 1980; Meyer et al., 2007). Headwater streams not only export quantities of organic matter and nutrients to downstream ecosystems (Vannote et al., 1980; Eggert et al., 2012), but also provide heterogeneous habitats for a wide range of endemic species, which collectively contribute to the biodiversity of river networks (Meyer et al., 2007).

Low-head dams are ubiquitous human-made disturbances in mountainous headwater streams that are used for water supply, agricultural irrigation, and recreation (Yan et al., 2013; Jumani et al., 2020; Li et al., 2021). Dams can affect stream fish assemblages via multiple pathways, including blocking species dispersal between localities (Porto et al., 1999; Barbarossa et al., 2020) and changing the habitat and physicochemical conditions in streams (Smith et al., 2017; Hitchman et al., 2018; Turgeon et al., 2019). Specifically, the transformation from lotic to lentic conditions in impoundments can provide suitable conditions for potential invasive species to displace native species (Johnson et al., 2008; Liew et al., 2016; Turgeon et al., 2019). Further, habitat simplification can decrease resource and niche availability (Ruhí et al., 2016; Turgeon et al., 2019).

The localized effects of individual low-head dams have been relatively well documented, but the effects and conclusions differ among studies (see Fencl et al., 2017; Turgeon et al., 2019; Li et al., 2021). A potential reason for this inconsistency is that non-native or generalist native species can compensate for the difference in total local diversity in impoundments (Slawski et al., 2008; Liew et al., 2016; Turgeon et al., 2019). Invasions of non-native species caused by dams have a profound impact on global freshwater biodiversity and have drawn increasing attention from the public (Liew et al., 2016; Turgeon et al., 2019; Zhang et al., 2019). More importantly, in regions where native fish species are completely dominant, generalist native species that commonly occur in large streams would be potential invaders for upland streams and are usually termed as “native-invasive species” (Scott and Helfman, 2001; Dala-Corte et al., 2019). However, most previous studies on the ecological effects of low-head dams have ignored the invasion patterns of native-invasive species, as increased native diversity is generally viewed as a positive effect on ecosystem function.

Changes in species occurrences (i.e., local extinctions and colonizations) not only alter community α-diversity, but also affect community β-diversity and dynamics (Socolar et al., 2016; Erős et al., 2020), resulting in the formation of “novel communities” (Hobbs et al., 2006). A serious effect of native-invasive species invasion on native species is a change in trophic ecology through predation, competition, and the indirect effects in the context of environmental changes (Busst and Britton, 2017). Although the effects of low-head dams on habitat conditions and biodiversity have long been examined (Fencl et al., 2015, 2017; Turgeon et al., 2019), the responses in species interaction networks remain understudied. Recently, Momal et al. (2020) proposed a more robust method for inferring species interaction networks by incorporating environmental conditions (i.e., abiotic factors). Furthermore, native-invasive species typically have broad niches and environmental tolerances in headwater streams (Scott and Helfman, 2001). These native-invasive species with broad niches may contribute less to beta diversity than species with narrow or intermediate niches, because the latter species may occur in environmentally restricted conditions and contribute more to beta diversity (Heino and Grönroos, 2017). Therefore, exploring the contribution of each species to beta diversity may provide a platform for understanding the formation of community diversity patterns (Heino and Grönroos, 2017; Gavioli et al., 2019).

The Wannan Mountains of China, characterized by low-head dams becoming ubiquitous features of the landscape (Li et al., 2021), offer an excellent opportunity to evaluate the effects of low-head dams on fish assemblages. Empirical evidence has shown that spatial patterns in the ecological effects of low-head dams can be quantified effectively by selecting free-flowing reference segments (Fencl et al., 2017; Hitchman et al., 2018; Li et al., 2021). Based on fish assemblage data collected in the impoundments created by low-head dams (i.e., comparison sites) and free-flowing segments (i.e., reference sites) in the first-order streams of the Wannan Mountains, we used a quantitative approach to assess the patterns in low-head dams affecting fish assemblages by considering the distinction between native and native-invasive species. We aimed to: (1) determine the effects of low-head dams on within-community (i.e., α-diversity) and between-community (i.e., β-diversity) diversities, (2) quantify the contributions of each species to the β-diversity for each habitat type, and (3) assess the difference in species interaction networks between the habitat types.

As a common anthropogenic disturbance in headwater streams, low-head dams may modify the local habitat in stream, including deeper water, slower flow, smaller substrate in impoundments upstream, and faster flow and larger substrate in the plunging areas downstream (Tiemann et al., 2004; Yan et al., 2013; Fencl et al., 2015). Due to the backwater effect of dams, the wetted width and water depth are greater in impoundments than that in free-flowing segments (Fencl et al., 2015). As the water depth increases in impoundments, current velocity is reduced and its sediment transport ability in the water column increases, generally resulting in an increase of sediment deposition (Tiemann et al., 2004; Yan et al., 2013). In this study, we found that low-head dams significantly decreased substrate coarseness, current velocity, and increased water depth, wetted width, and substrate heterogeneity in impoundments, which is consistent with the findings of other studies (Yan et al., 2013; Fencl et al., 2015).

Habitat modifications associated with anthropogenic activities not only cause the native species population to decline and even cause local extinctions, but also facilitate the invasion of alien/native-invasive species into upland streams (Slawski et al., 2008; Dala-Corte et al., 2019; Turgeon et al., 2019). A serious consequence associated with low-head dams is the addition of native-invasive species and exclusion of native species (Liew et al., 2016; Smith et al., 2017; Liu et al., 2019). Our study clearly documented that the species richness of native species significantly decreased in impoundments, while that of native-invasive species increased (Table 2). Changes in local diversities associated with anthropogenic activities often depend on the relative number of lost and gained species and their abundances (Turgeon et al., 2019). Unsurprisingly, increases in native-invasive species richness can largely compensate for the loss of native species, which could explain this pattern of changes in the total species richness in impoundments; this effect masked changes in the whole fish assemblage. When the loss of native species (abundance change: 14.89) outnumbers the gain of native-invasive species (abundance change: 0.21), changes in the species abundance patterns of whole assemblages may be driven by native species (Table 2). Overall, low-head dams favor native-invasive species, while negatively affecting native species with high sensitivity to environmental change (Slawski et al., 2008; Liu et al., 2019).

The variation in both the identity and abundance of species between different habitats can also affect the degree to which local fish community compositions differ (Barwell et al., 2015; Li et al., 2021). More importantly, our results showed that the local community changes in fish assemblages (i.e., the total β-diversity) associated with low-head dams were significantly higher for abundance-based approaches than for occurrence-based approaches (Table 3), suggesting that the latter could underestimate the localized effects of individual low-head dams. As low-head dams are less likely to result in local-scale species extinctions than large dams, most of the remaining species often experience changes in abundance (decreased, increased, or unchanged; Yan et al., 2013; Fencl et al., 2017). Previous studies have shown that changes in fish diversity patterns may be primarily driven by changes in the abundance of native species (Tiemann et al., 2004; Liu et al., 2019) in communities or regions where native-invasive species are still not predominant. If the number of shared species between communities does not change, but the abundance does, the dissimilarity index based on abundance approaches will show changes that depend on the magnitude of shifts in abundance (Cassey et al., 2008; Baselga, 2013; Legendre, 2014). In this study, 29 of the 32 species were found in both impoundments and free-flowing segments. However, besides the variations in the abundances of dominant native species, we also found that some native species (e.g., Vanmanenia stenosoma and Aphyocypris chinensis) drastically decreased and other native-invasive species (e.g., Carassius auratus and Odontobutis potamophila) markedly increased in abundance in the impoundments compared to in the free-flowing segments (Table 1). Therefore, the results of our study are not surprising when formerly rare species become dominant or vice versa, given that species abundances can change drastically enough to produce large differences relative to the changes in species composition from one community to another.

For occurrence- and abundance-based approaches, we also found that native species had higher SCBD values than native-invasive species in both habitat types, but significant variation was observed only in abundance-based approaches (Figure 4). Headwater streams with highly heterogeneous habitats nourish unique species assemblages (Meyer et al., 2007), where native species with small niche breadths occurring under environmentally restricted conditions may contribute more to β-diversity (Heino and Grönroos, 2017). As is generally observed in natural ecosystems, communities often exhibit heterogeneous species–abundant distributions (Magurran et al., 2011). In this study, the RA of most native species across sites was higher than that of native-invasive species, especially in regions where native fish species were completely predominant. Previous studies have shown that species with high total abundance across sites contribute most to the abundance-based β-diversity (Heino and Grönroos, 2017; Gavioli et al., 2019). When the identities of native and native-invasive species are considered, the high SCBD values can reveal which native-invasive species are the most abundant and widespread, whereas the low SCBD values can identify rare native species. Although some rare native species have a relatively lower contribution to β-diversity, such species may still need conservation actions owing to their local extinction risk (Gavioli et al., 2019). Meanwhile, the native-invasive species with high abundance should be strictly controlled to mitigate the ecological consequences of low-head dams.

Complex ecological networks of interacting species are crucial for maintaining biodiversity in local ecosystems and the effects of species interactions can differ either spatially or temporally (Tylianakis and Morris, 2017). Normally, a fraction of native species may primarily contribute to local community dynamics (Meyer et al., 2007), especially in headwater streams where native-invasive fish species are still not predominant. However, changes in species diversity and basal food resources resulting from habitat modifications can affect the species interaction networks (Brimacombe et al., 2021; Danet et al., 2021). Previous studies have shown that impoundments created by dams can act as powerful environmental filters that may select sets of species that are better adapted to particular habitat conditions (Liew et al., 2016). In this study, we found that the betweenness centrality of native-invasive species (i.e., omnivorous fish) was the highest in the impoundments (Figure 5), which reflected the importance of species for the interaction network structure. Indeed, betweenness centrality measures the ability of nodes (i.e., species) to act as a bridge in the interaction network, which can efficiently identify the “keystone species” in ecosystems (González et al., 2010; Momal et al., 2020). High betweenness centrality values of species represent high network connectance among species, which can strengthen the importance of species interaction networks (Thébault and Fontaine, 2010). In addition, the transformation of natural lotic habitats into artificial lentic habitats not only facilitate the establishment of native-invasive species (Chu et al., 2015; Liew et al., 2016), but also greatly change basal food resources (Ruhí et al., 2016). Feeding habits can play a crucial role in maintaining the stability of species interactions, and tropical streams are typically dominated by omnivorous mechanisms (Ruhí et al., 2016; Brimacombe et al., 2021). Therefore, native-invasive species (omnivorous fish) can take full advantage of niches specific to lentic habitats in impoundments, which are essential for maintaining the stability of interaction networks.

In this study, we used a quantitative approach to assess the patterns of low-head dams affecting stream fish assemblages by distinguishing between native and native-invasive species. Our results suggest that low-head dams affect stream fish diversity in two ways: by adding native-invasive species and by excluding native species typical of headwaters. However, changes in species richness and abundance not only alter the local community composition (Barwell et al., 2015; Li et al., 2021), but also affect the community stability (Brimacombe et al., 2021). Although species interactions were not directly observed in our system, we provide initial evidence that low-head dams influence the interaction network structure. Additionally, we found that occurrence-based approaches could underestimate the localized effects of low-head dams when fish assemblages differ primarily in abundance rather than species composition. SCBD values can determine which species contribute the most to β-diversity (Legendre and De Cáceres, 2013). We found that the abundance-based SCBD values were significantly higher for native fish species than for native-invasive ones in both habitat types. In communities or regions where native fish species are predominant, our results suggest that understanding which species contribute to β-diversity will offer new insights into the development of effective conservation strategies, by taking the identities of native and native-invasive species into account.

The original contributions presented in this study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding authors.

The animal study was reviewed and approved by the Animal Ethics Committee of the Anhui Normal University.

YY and QL designed the study. QL, XL, KT, CZ, and YY conducted the field and/or laboratory work. QL, XL, and CZ analyzed data. LC, HF, and YG contributed materials, reagents, and analytical tools. QL drafted the manuscript with the assistance of CZ and YY. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.