The Himalaya spread over 3,200 km across seven countries, representing the tallest and youngest mountain range in the world. Western Himalaya extends from Pakistan to north-west India (west of Nepal), while E Himalaya extends from eastern Nepal to northern Myanmar (Rodgers et al., 2000). It has high biodiversity value as recognized by its four biodiversity hotspots, 60 ecoregions and 330 important bird areas (Myers et al., 2000). Riverine areas in the Himalaya harbor the highest diversity of specialist riverine birds in the world, among which several species are of conservation concern (Buckton, 1998; Sinha et al., 2019a; Menzies et al., 2021).

We conducted field surveys in the states of Arunachal Pradesh and Uttarakhand in the Indian part of the E and W Himalaya, respectively (Figure 1). Arunachal Pradesh is part of the Eastern Himalaya Biodiversity Hotspot, sharing its borders with Myanmar, China, and Bhutan (Mittermeier et al., 2011). It is drained by many rivers, such as the Noa Dehing, Kamlang, Lohit, Dibang, Siang, Subansiri, and Kameng which are important headwaters to the river Brahmaputra. In the west, Uttarakhand shares international borders with Nepal and China and is drained by torrential snow-melt rivers such as the Bhagirathi, Alaknanda, Mandakini, Pindar, Kosi, and Kali, several of which are headwaters to the Ganges. In the areas sampled in the E Himalaya, the vegetation is dominated by tropical and subtropical broadleaved evergreen forests. In the W Himalaya, the vegetation in riparian areas consists of conifers at higher elevations and subtropical vegetation at the foothills (Gaur et al., 2019).

Following Buckton (1998), we sampled river-dependent birds by walking along river banks and recording the number of individuals of different bird species detected. In the E Himalaya, we sampled 81 sites of variable lengths (average length = 1,285 m; range = 300 m–2 km; total effort = 101.3 km) across seven river drainages encompassing a wide elevational gradient of 60–2,000 m asl (Figure 1). In the W Himalaya, we sampled 53 river reaches of equal length (500 m; total effort = 26.5 km) in the Bhagirathi (main river and six first-order streams) and Amrut Ganga basins, important headstreams of the Upper Ganges, between an elevation gradient of 330–3,100 m asl. Surveys were carried out post-monsoon (mainly covering winter months) when several of the river bird species migrate to lower altitudes. Conducting surveys on foot during the monsoons is not feasible or safe since the rivers swell and prevent crossing for long stretches. Sampling was conducted between September to late January (2016–2018) in the W Himalaya and between late August to early March (2017–2019) in the E Himalaya. All species’ nomenclature follows the Clements Checklist by the Cornell Lab of Ornithology (Clements, 2007). For more details on the field survey design, please see Menzies et al. (2021) for the E Himalaya and Sinha et al. (2019a) for the W Himalaya.

We built a phylogenetic tree of 39 bird species recorded in our field surveys by pruning the global bird phylogenetic tree obtained from www.birdtree.org (Jetz et al., 2012). The original tree in Jetz et al. (2012) was assimilated from genetic data of 6,693 species of extant birds. The backbone of the tree was constructed using 15 genes (19 loci) of 151 key species and time-calibrated with 10 well-known fossils (Jetz et al., 2012). A total of 10,000 trees were sub-sampled for our target species using a pseudo-posterior distribution¹ to obtain 100 trees, which were used to prepare a consensus tree used in further analysis. Packages “ape” (Paradis et al., 2004) and “phytools” (Revell, 2012) were used to construct a phylogenetic tree of the species recorded in the field.

We obtained data on the trophic level, foraging niche, and morphometry from Supplementary Dataset 1 in Pigot et al. (2020). Birds were classified into four trophic guilds (herbivore, carnivore, omnivore, and scavenger) and 10 foraging niches (e.g., aquatic dive, aquatic ground, aquatic perch, ground) (Pigot et al., 2020). For morphological trait data, we used the principal component scores reported for the overall body traits (comprising data on body mass, beak, wing, tail, and tarsus). For overall body traits, we used principal component scores of the first eight axes, which explained 99.9% of the variation among 9,963 bird species. Please see Pigot et al. (2020) for additional details on the functional trait data.

Global patterns in species richness of river specialist birds highlight the role of net primary productivity (NPP) and temperature in shaping species richness patterns (Buckton and Ormerod, 2002). We downloaded net primary productivity data from the Terra Net primary productivity database from MODIS.² We obtained temperature seasonality (bio4) and annual precipitation (bio12) data from WorldClim (Fick and Hijmans, 2017). The recently developed human modification raster layer from Kennedy et al. (2019) and the forest cover data from Global Forest Watch³ and its extracted values using a 500 m buffer along the surveyed river reaches were used in this study. We kept the buffer radius short as the life history parameters of the focal bird species are strictly tied to the riparian habitat (Buckton and Ormerod, 2008). We recorded elevation and river width at survey locations during fieldwork.

We recorded 3,897 individuals of 39 species of birds belonging to 15 families during combined field surveys. We encountered 2,996 individual birds consisting of 36 bird species in the E Himalaya and 901 individual birds from 17 species in the west (Table 1). Muscicapidae (old world flycatchers) and Alcedinidae (kingfishers) were the most represented families in both the east and the west. Plumbeous Redstart (Phoenicurus fuliginosus) was the most encountered species in W Himalaya and the second most encountered species in E Himalaya. Ibisbill (Ibidoryncha struthersii) was the least encountered bird in W Himalaya, whereas Black-backed Forktail (Enicurus immaculatus) and White-throated Dipper (Cinclus cinclus) were the least encountered species in E Himalaya (for species list see Table 1). We recorded four species of conservation concern, including the Critically Endangered White-bellied Heron (Ardea insignis), Endangered Black-bellied Tern (Sterna acuticauda), and the Near Threatened River Lapwing (Vanellus duvaucelii) and Great Thick-knee (Esacus recurvirostris). There was some evidence phylogenetic clustering for the W Himalaya sites as their median values were beyond −1.96 units from the null expectation (Figure 2).

We did not find differences in taxonomic and MNTD metrics for the phylogenetic and functional components (p > 0.05). However, we found greater clustering in phylogenetic (W = 2,229,371; P < 0.001) and functional (W = 2,230,439; P < 0.001) MPD for the W Himalaya than that for the E Himalaya (Figure 2). Beta diversity was highest for the taxonomic component compared to that for phylogenetic and functional components (Figure 3). MPD was higher than MNTD for both the phylogenetic and functional components of beta diversity (Figure 3).

The relative contribution of turnover and nestedness to the overall beta diversity varied across the different components and also between the E and W Himalaya. In the E Himalaya, there was a greater contribution of turnover to overall beta diversity as compared to that by nestedness for the taxonomic (W = 10,556,567, P < 0.001), phylogenetic (W = 7,699,831, P < 0.001), and functional (W = 8,994,152, P < 0.001) components (Figure 3). In the W Himalaya, there was greater contribution of turnover to overall beta diversity only for the taxonomic component (W = 581,821, P < 0.001). The overall turnover (β-total) was similar across E and W Himalaya for the taxonomic component (P > 0.05). However, the overall turnover was higher for the E Himalaya than that for W Himalaya for phylogenetic (W = 1,698,385; P = 0.023) and functional (W = 1,756,820; P < 0.001) components (Figure 4).

The GDM explained 64–90% of the variation in different beta diversity measures, thus explaining beta diversity patterns in river bird communities in the Indian Himalaya (Figure 5). Overall, GDM explained maximum variance for the functional components of beta diversity followed by the taxonomic and phylogenetic components; comparatively, a higher proportion of variance was explained for the W Himalayan communities than that for the east (Figure 5). Maximum variance was explained for β-FD_MPD (93%) followed by β-FD_MNTD (82%) and β-TD (81.8%) in the west.

Results of spatial GDM showed that the habitat and environmental variables, rather than geographical distance, influenced the observed beta diversity patterns in river bird communities in both the east and the west (Figure 5). River width explained a bulk of the variation in beta diversity for most measures across all three components (Figure 5). The relationship between river width was non-linear and beta diversity increased exponentially with increasing river width, particularly in the E Himalaya and reached an asymptote in the W Himalaya (Supplementary Figures 1–6). While forest cover and elevation explained variation in beta diversity patterns in the east, rainfall was an important variable influencing beta diversity patterns in the west (Figure 5). Temperature seasonality, net primary productivity, and human modification showed limited roles in driving the beta diversity patterns (Figure 5). The relationships between the different predictors and beta diversity measures were non-linear and varied across the different measures and the E and W Himalaya regions (Supplementary Figures 1–6).

A significant proportion of the variation in beta diversity patterns in riverine bird communities was best explained by habitat features such as river width, elevation, forest cover, and climatic factors such as precipitation. River width was a consistent driver of all measures of beta diversity (Figure 5). In the E Himalaya, beta diversity for all three measures increased with river width, with a sharp rise in dissimilarity across river stretches wider than 150 m (Supplementary Figures 1–3). Larger river width is associated with river stretches at lower elevations with a higher discharge; they are also characterized by variable flow rates offered lotic to lentic habitats, which provide niches to a diverse array of riverine birds preferring slow- or medium-flowing river sections. The bank substrates are also variable (rocky or sandy), particularly in the lower sections of the rivers. Bouldery and pebbly bank substrates offer important microhabitats for birds such as dippers and redstarts, while the lower elevation sandy banks are preferred by wagtails, pratincoles, and lapwings (Sinha et al., 2019a). Since several riverine bird species show a preference for specific micro-habitats, there is likely to be higher species turnover as rivers progress from narrow headstreams to wider river stretches. Narrower river stretches support prey-base for a small group of specialist birds (e.g., redstarts and forktails) that predominantly have an insectivorous diet (Buckton and Ormerod, 2008; Sinha et al., 2019a). In comparison, wider river stretches offer greater river depths associated with larger prey such as fishes, amphibians, and mollusks, thus attracting other waterbirds such as kingfishers, cormorants, mergansers, gulls, and sandpipers to visit these river stretches opportunistically (for species list see Table 1).

Interestingly, the relative influence of other drivers varied across the E and W Himalaya. Forest cover and elevation explained significant variation for the E Himalayan birds. Along with the river channel and flood plain, the riparian forests are also recognized as an integral part of the riverscape (Weins, 2002). Distributions of many riverine bird species are positively or negatively associated with tree cover (Sullivan et al., 2007; Vaughan et al., 2007; Sullivan and Vierling, 2012; Sinha et al., 2019a). Here, beta diversity increased with increasing forest cover, indicating a higher homogenization in bird composition in river stretches with lower forest cover (Supplementary Figures 1–3). Forests in the Himalaya are being lost at an alarming rate (Srivastava et al., 2002; Gaur et al., 2019; Sheth et al., 2020), which could potentially result in lower species turnover in riverine bird communities with cascading effects on overall diversity patterns in the long-term.

Elevation is an important driver of riverine bird diversity (Manel et al., 2000). Interestingly, at elevations above 300 m in the E Himalaya, the beta diversity patterns were consistent and showed a relatively lower species turnover at elevations below 300 m. In the E Himalaya, the lowermost elevations are characterized by higher anthropogenic disturbances, resulting threats such as lower forest cover, and boulder and sand mining (Srivastava et al., 2002; Menzies et al., 2021), which likely contribute to the lower species turnover in the lowland elevations. Anthropogenic activities negatively impact specialist bird species resulting in the persistence of generalist species that are able to withstand habitat changes (Sinha et al., 2019a) likely resulting in lower turnover in the more modified, low elevation riparian habitats. Additionally, lower elevation avian communities were dominated by large-body sized species (e.g., cormorants, mergansers, pratincoles, waders) which prefer wider river stretches and habitat features, like sand bars and river islands. These could be uniform across sites in the lower elevation which may have resulted in lower turnover. These aspects need to be explored in greater detail in future.

Climatic factors such as precipitation and temperature are related to the variation in beta diversity patterns (Naka et al., 2020; Wayman et al., 2021). We found that precipitation was an important factor in explaining the variation in beta diversity patterns in the W Himalaya. The W Himalaya is more seasonal and drier than the E Himalaya (Price et al., 2011). There was a lower species turnover in drier habitats as compared to wetter habitats (precipitation lower than ∼1,400 mm). E Himalaya is consistently wetter across the entire elevation gradient; therefore, precipitation does not play a significant role in influencing the beta diversity patterns (Supplementary Figures 1–3). Similar to forest birds, the W Himalayan riverine bird assemblage is a nested subset of the E Himalaya assemblage, which is more species-rich (Buckton and Ormerod, 2002; Price et al., 2011; Srinivasan et al., 2014). This indicates that species that were able to pass through the environmental filter imposed by a drier and variable climate in the W Himalaya were able to successfully colonize, despite the harsh climate. However, this pattern is also likely to be a consequence of past climatic influences. During the Pleistocene glaciation, the W Himalayan region was under ice, while evergreen forests persisted in the E Himalayan region (Owen et al., 2002; Srinivasan et al., 2014). Thus, the past and present seasonality and variability in climate may have led to a greater clustering in riverine communities in the W Himalaya than that in the E Himalaya.

We found a greater contribution of the turnover component in taxonomic and functional beta diversity, especially in the E Himalaya. Higher turnover is a consequence of biotic interactions (e.g., competition), and greater nestedness is a consequence of local or regional extinctions, particularly in unfavorable and variable environments (Schemske et al., 2009; Soininen et al., 2018). As discussed earlier, the E Himalaya is more species-rich than the W Himalaya (Buckton and Ormerod, 2002); therefore, it is more likely that biotic interactions have an influence on turnover of species communities in the E Himalaya. Moreover, past (Pleistocene glaciation) and present (variable precipitation) climatic conditions likely cause similar influence of nestedness and turnover in the W Himalaya.

There was a relatively higher turnover in taxonomic diversity than that in phylogenetic and functional diversity, which indicates that species are replaced by their close relatives with similar traits across sites. Himalayan rivers are characterized by high species richness consisting of multiple species within a lineage of riverine birds (e.g., three species of redstarts, five species of forktails, five species of kingfishers) that differ in their microhabitat requirements (Buckton and Ormerod, 2008; Sinha et al., 2019a). Riverine ecosystems are extremely dynamic because of high natural disturbances offering diverse microhabitats (Ward et al., 2002). Changing microhabitats and the associated replacement of species by their close relatives play a functionally similar role that likely results in the pattern observed in this study.

We found that the different beta diversity measures were related to similar predictors within a region (i.e., E or W Himalaya). However, the relative contribution of the different predictors in explaining the variation in beta diversity differed across the different measures (Figure 5), which has been reported elsewhere (Wayman et al., 2021). Interestingly, precipitation explained a large proportion of variation in taxonomic component of beta diversity in the W Himalaya (Figure 5), but a lower proportion of variation in phylogenetic and functional component. This result could be a likely outcome of closely related species (which are likely to be functionally similar) replacing each other along a gradient of precipitation. Moreover, the proportion of variation explained by forest cover across the different components was similar; this is a likely outcome of certain groups disappearing along the forest cover gradient and needs to be examined in greater detail in the future.

Geographical distance is an important driver of beta diversity patterns in birds (Wayman et al., 2021). However, we did not find a significant influence of geographical distance on beta diversity patterns in the E or the W Himalaya. The influence of distance on beta diversity patterns is explained by dispersal limitation (Myers et al., 2013; Weinstein et al., 2014). Birds are unlikely to be dispersal limited at the scale at which the analysis has been carried out in the E and the W Himalaya. Although we found limited evidence of the influence of habitat modification on beta diversity patterns in the W but not in the E Himalaya, past studies have documented birds being sensitive to anthropogenic influences (Sinha et al., 2019a; Abreu et al., 2020; Menzies et al., 2021). Fine-scale information on human disturbance may be able to provide additional insights on human impacts.

This study was conducted primarily in the non-breeding season. Given that some riverine species exhibit altitudinal migration, beta diversity patterns and the influence of underlying drivers may vary. It will be important to conduct a similar study during the breeding season. However, conducting studies during the breeding season might be difficult due to inclement weather and associated flooding of the river. Nevertheless, the non-breeding season is also associated with a greater diversity of birds in the region as riverine stretches are being used by several other bird species that are dependent on riparian resources; hence, the non-breeding season is an important time for evaluating community organization. Given the high rates of human-driven modification of the Himalayan rivers, a systematic Himalaya wide survey across all drainages is required to determine the role of anthropogenic impacts on riverine systems.

Another potential shortcoming of the study could be variable sampling effort, particularly in the Eastern Himalaya. However, we used sampling length as a predictor in the variance partitioning analysis and demonstrate that very little variation in the beta diversity in E Himalaya is explained by the sampling length. Additionally, analysis of turnover in taxonomic beta diversity using a novel method (Zou and Axmacher, 2020) that helps control for variable sampling effort did not reveal any differences in beta diversity patterns (Supplementary Figure 7). Given this it is likely that variable sampling length will unlikely influence the outcome of this study.