Background
Downstream hydraulic geometry can predict several aspects of stream properties including mean velocity and depth (Leopold & Maddock, 1953; Lee & Julien, 2006), however, channel width is most widely studied due to the ease of making remote or simple field measurements (e.g., Fisher et al., 2013; Dunne & Jerolmack, 2020). Longitudinal increases in channel width show a consistent scaling relationship with bankfull discharge, in that the exponent (b) commonly approximates 0.5 (Equation (1)); whereas, the coefficient (a) reflects local variations in streambank erosivity, controlled by sediment texture, vegetation type, flow regime, etc. (Anderson et al., 2004). Although downstream hydraulic geometry relationships are commonly well-developed in alluvial channels, strong downstream hydraulic geometry relationships are evident even in regions with external non-alluvial controls, such as in bedrock channels with local variations due to variations in bedrock erosivity (Montgomery & Gran, 2001), and in those with discontinuous colluvial input (Wohl & Wilcox, 2005). However, in cases where the substrate resisting forces exceed hydraulic driving forces, downstream hydraulic geometry relationships are not as well developed, such as in mountain streams with very coarse sediment relative to hydraulic forces (where the stream power to sediment size ratio (Ω/D84) < 10,000 kg/s3) (Wohl, 2004). Lakes can also contribute to overall poor DHG relationships, with higher width to depth ratios directly downstream of lakes, as a result of lakes trapping sediment (Arp et al., 2007). In cases where bankfull discharge data are unavailable, drainage area can be used a proxy for bankfull discharge (Equation (2) & (3)) (Soar & Thorne, 2001; Faustini et al., 2009). Because there may be additional factors to drainage area controlling bankfull discharge, DHG relationships using drainage area tend to show significantly more scatter, yet is still a powerful tool since only a digital elevation model is required as input data (Faustini et al., 2009). However, although drainage area-based DHG relationships are poor in some regions, in a review of regression coefficients for width vs. drainage area relationships (Eq. 3), all R2 values exceed 0.24 and most >0.6; exponent-values (beta), which should be less than b (commonly ~0.5) since the relationship between drainage area and discharge (Eq. 2) where the exponent-value (y) is commonly slightly less than 1 (Faustini et al., 2009), mostly ranged from 0.3-0.45.
\(w=aQ^{b}\),  (1)
where w is bankfull width, a is a regional coefficient, Q is bankfull discharge, and b is a regional exponent (commonly ~0.5).
\(Q=xA^{y}\), (2)
where A is drainage area, and x and y are regional coefficients and exponents, and y is commonly slightly < 1.
\(w=\alpha A^{\beta}\), (3)
where α and β are empirical parameters and β is commonly <0.5.
Moreover, adjustment of geomorphic parameters assume an alluvial channel where self-adjustment of channel form based on the current flow regime is possible (Leopold & Maddock, 1953, Phillips & Jerolmack, 2016). Therefore, semi-alluvial channels, which contain cohesive or coarse sediment deposited by non-fluvial geomorphic processes (Polvi et al., 2014; Pike et al., 2018; Polvi, 2021) could also contribute to poor-fitting relationships between geomorphic and ecological longitudinal patterns rather than fitting the established patterns from highly connected stream networks with alluvial channels.
Geomorphic and ecological longitudinal trends are intertwined with one another, depending on connectivity of flows and sediment in addition to ecological meta-communities, through migration of organisms and dispersal of seeds and propagules. Because hydrochory is dependent on longitudinal connectivity for dispersion, the organization of passively dispersing organisms (e.g., riparian vegetation), provides a natural test of functional connectivity. Through hydrochory, species accumulate downstream, which leads to higher species richness or densities downstream (e.g., Nilsson et al., 1989, Dunn et al., 2011, Kuglerová et al., 2015). In anthropogenically-fragmented catchments (e.g., due to dams), hydrochoric seed dispersal is interrupted, thus altering the biotic communities, causing proximal reaches on either side of a dam to form dissimilar riparian vegetation species compositions (Jansson et al., 2005).  Although anthropogenic factors can create barriers in connectivity (e.g. dams) (Nilsson, 2005; Nilsson et al., 2010) and stream restoration has focused on increasing connectivity, many geomorphic and ecological engineering processes serve to create natural forms of longitudinal disconnectivity. These spatially and sometimes temporally intermittent barriers or buffers to flow, sediment and propagule fluxes, have received much less attention in the hydrochory literature and how they influence metacommunity organization. Once prevalent beaver dams and log jams cause widespread ‘leaky’ barriers (Wohl & Beckman, 2014), serving to trap sediment, carbon, and attenuate flows. Similarly lakes, which are widespread in northern latitudes (Messager et al., 2016), particularly where Pleistocene glaciation has eroded bedrock and deposited moraine dams, can have substantial effects on stream topology and form and function of rivers where they are connected to stream networks (Gardner et al., 2019), such as the connectivity of sediment (Arp et al., 2007), seed dispersal (Su et al., 2019a, b), and diversity of invertebrates (Green et al., n.d.) .
We focused our study on two catchments in boreal northern Sweden that are heavily influenced by past continental glaciation, creating a stream network with multiple instream lakes and coarse till deposits (Nilsson et al., 2002; Polvi et al., 2014; Su et al., 2019a, b). The stream networks can be divided into three process domains, defined by Montgomery (1999) as zones with distinct geomorphic processes that structure ecological disturbances and thus organize biotic communities: rapids, slow-flowing reaches, and lakes (Nilsson et al., 2002; Su et al., 2019a, b) (Figure 1a). Rapids are steep (S0: 0.1-5%) gravel- to boulder-bed channels with coarse glacial legacy sediment (cobbles and boulders) and bedforms that do not conform to alluvial bedform-channel slope relationships, sensu Montgomery & Buffington (1997); slow-flowing reaches (S0 < 0.1%) are straight or meandering channels flowing through peat or fine sediment with wetland vegetation riparian zones; lakes have inlets and outlets and are lined with either fine sediment or coarse till. Within our study catchments, we examined the longitudinal distribution of process domains along nearly the entire mainstem channel, evaluated how channel width changes downstream and examined how riparian vegetation communities change along the channel in each new process domain. Our aim was to determine whether these glaciated, boreal stream networks fit established patterns of geomorphic and ecological longitudinal changes. We test the functional geomorphic and ecological connectivity of these catchments with high-resolution spatial data of riparian vegetation communities and channel width.