Connectivity in rivers

2.1 Disparities in rates

Direct measurements of material fluxes have typically been conducted over relatively small spatial and short temporal scales. Extrapolating from direct measurements to larger scales can be problematic, however, particularly in the context of sediment movement. Schumm (1963) noted that contemporary rates of uplift are about eight times greater than contemporary rates of denudation, as inferred from measured sediment transport in rivers. Walling (1983) introduced the concept of the sediment delivery problem, in which only a fraction of the sediment eroded within a catchment will reach the basin outlet and be represented as sediment yield (Fryirs, 2013). Each investigator invoked sediment storage mechanisms to explain the discrepancy, and sediment storage can be viewed as a limitation on sediment connectivity. Kirchner et al. (2001) compared stream sediment fluxes directly measured over periods of 10¹ years to erosion rates derived from cosmogenic ¹⁰Be over timespans of 10³ years and erosion rates from apatite fission tracks over periods of 10⁶ years. The longer term estimates of sediment yields were on average 17 times greater than measured stream sediment fluxes. Kirchner et al. attributed these discrepancies to the episodic delivery of sediment from mountainous terrain. Even relatively short, annual to decadal records of sediment flux record temporal disparities in sediment movement, with much of the sediment flux occurring during a limited portion of the time interval being measured (Farnsworth and Milliman, 2003; Kao and Milliman, 2008). In a sense, sediment fluxes are fractal in time, with episodic fluxes characterizing a range of temporal and spatial scales. Understanding the processes by which sediment movement occurs and the factors that limit these processes (i.e., limit sediment connectivity), whether the limitations result from energy available to entrain sediment particles, supply of sediment to the river, or obstructions within the river, can illuminate the causes of episodic sediment fluxes (Bracken et al., 2015; Brierley et al., 2006; Fryirs, 2013). The apparent discrepancies in sediment delivery estimates over varying time scales and associated scaling issues continue to generate controversy (e.g., Parsons et al., 2006). However, the characteristics of sediment connectivity within drainage basins over diverse spatial and temporal scales can be a useful conceptual framework for understanding disparities in rates of sediment transport and storage.

2.2 River response to disturbance

Ecologists define a disturbance as any event that disrupts an ecosystem, community, or population structure and changes resources, substrate availability, or the physical environment (White and Pickett, 1985). Geomorphologists have tended to follow this definition (e.g., Butler, 2001; Montgomery, 1999). Natural disturbances in rivers typically result from abrupt and substantial changes in water and sediment supplied to the river corridor, such as during floods or debris flows. Human-induced disturbances can also be abrupt or can involve a sustained change associated with channel engineering or flow regulation. The resistance of a natural system describes how much the disturbance changes the system: a resistant bedrock channel might undergo little change in channel geometry or substrate characteristics as a result of a large flood. The resilience of a system, also referred to as sensitivity in the geomorphic literature (e.g., Fryirs, 2017), describes how quickly the system returns to pre-disturbance characteristics. A resilient alluvial channel might undergo substantial change in cross-sectional and bedform geometry during a large flood, but return to its pre-flood configuration within two or three years.

The resistance and resilience of a river both reflect connectivity during and after the disturbance. An anastomosing channel segment within a broad, vegetated floodplain may be resistant to change during a flood because the high flow spreads across the floodplain and dissipates energy. Energy dissipation occurs through overcoming hydraulic resistance caused by relatively shallow flow depths and high boundary roughness on the floodplain. In this example, the high lateral connectivity creates resistance. The broad alluvial floodplain can also have substantial hyporheic storage, which limits the longitudinal connectivity of water and reduces the erosional energy exerted by the peak flow against downstream river segments. In this case, high vertical connectivity and lower longitudinal connectivity create resistance. A laterally confined, gravel-bed, pool-riffle channel could be resilient to a flood because the excess sediment deposited during the flood is transported downstream within a few years by lower flows confined within the channel (Lisle, 1982). Here, high longitudinal connectivity creates resilience. In these and other cases, understanding the degree of connectivity during disturbances and intervening periods, as well as the factors that create or limit connectivity, can explain how a river responds to disturbances. Intertwined with this understanding is the realization that the alteration of sources of resistance or resilience can create substantial geomorphic change and alter the trajectory of adjustments within a river corridor (e.g., Brooks and Brierley, 2004; Brooks et al., 2003).

2.3 Crossing boundaries: River corridor science

Investigations of rivers have been mainly reductionist for nearly a century, partly out of a desire to develop a quantitative, mechanistic understanding of rivers and partly because a reductionist approach can facilitate isolating and manipulating individual variables. During this period, individual disciplines, such as stream ecology, fluvial geomorphology, aquatic biogeochemistry, and engineering hydrology, have emphasized specific aspects of rivers, sometimes treating these aspects in isolation. Explicitly focusing on connectivity can facilitate a more holistic understanding of rivers, as reflected in recent calls to think about riverscapes (Fausch et al., 2002) or river corridors (Harvey and Gooseff, 2015) rather than just river channels. Hyporheic zones, reciprocal terrestrial-aquatic subsidies, inundation hydrology, and segment-scale sediment budgets illustrate the centrality of connectivity in understanding river corridors.

Hyporheic exchange flows exemplify vertical connectivity within rivers and the relatively recent explosion of investigations devoted to this form of connectivity highlight the importance of vertical connectivity for water quality and river ecosystem sustainability (e.g., Acuña and Tockner, 2009; Gooseff, 2010; Tonina and Buffington, 2009). Reciprocal terrestrial-aquatic subsidies describe diverse exchanges among a channel, riparian zone, and upland environment. Examples include riparian plant litter, large wood, and insects that form an important part of river habitat and trophic systems (Livers and Wohl, 2016; Nakano and Murakami, 2001) and aquatic insects that emerge from the river and enter riparian trophic systems as food for riparian predators (Baxter et al., 2005). Hynes (1975) recognized these forms of lateral connectivity in a paper that is a classic of stream ecology and the recent use of isotopic tracers has further illuminated the fundamental importance of these exchanges within river corridors (England and Rosemond, 2004).

Vertical connectivity and lateral connectivity in other forms beyond main channel-floodplain connectivity can create differences in water quality, sediment dynamics, and flood hydrograph. Mertes (2000) introduced the phrase inundation hydrology, for example, to emphasize that the inundation of floodplains can occur before overbank flow on the main channel as a result of water inputs from other sources, such as a rising water table, overland flow, and overbank flows from tributaries.

Segment-scale sediment budgets from some floodplain rivers indicate that at least half of the sediment yield from a river segment can come from remobilization of floodplain sediment through lateral channel migration, rather than from transport of sediment originating upstream from the segment (Aalto et al., 2008; Dunne et al., 1998; Ham, 2005). In this example, lateral connectivity strongly influences longitudinal sediment flux.

One implication of these diverse examples is that what seems to be the most obvious form of connectivity in any given system – downstream flow of water in a channel, consumption of aquatic insects by fish, overbank flow during a flood, or longitudinally continuous sediment transport within a channel – may not be the only or even the most important aspect of connectivity within the river.

2.4 Retention in rivers

Retention here refers to storage of material over varying time scales, from a few minutes to tens of thousands of years or more, depending on the example under consideration. Although there are scenarios in which retention within a river is not desirable, such as retention of pollutants adsorbed to fine sediment, in many cases retention is a benefit. Retention fundamentally reflects some limit on longitudinal connectivity and fluxes of water, sediment, solutes, and organic matter. Consequently, explicit focus on the processes underlying connectivity can facilitate understanding and enhancing retention in rivers. River corridors that retain water within hyporheic zones or floodplains can attenuate flood peak flow and sustain base flow during dry periods (Lininger and Latrubesse, 2016; Wegener et al., in press). The retention of nitrates or dissolved organic carbon within hyporheic zones, riparian forests, or floodplain wetlands can improve water quality if these portions of the river corridor contain microbial communities capable of processing the nitrates and carbon (Zarnetske et al., 2011). River corridors that retain particulate organic matter can support greater biomass and biodiversity (Beckman and Wohl, 2014; Jones and Smock, 1991). Because of these benefits, much effort is now devoted to restoring retention within river corridors that has been lost through river engineering, such as channelization or levee construction (Palmer et al., 2005; Wohl et al., 2015b).

2.5 Nonlinear behavior

Nonlinear behavior occurs when outputs or responses are not proportional to inputs or stimuli (Boeing, 2016). Nonlinear behavior includes system changes that occur without external forcing; patterns that emerge spontaneously; asynchronous and contrasting responses within a system; and the presence of thresholds and feedbacks (Favis-Mortlock, 2013; Phillips, 2003). Spatial and temporal variations in connectivity are responsible for at least some of these forms of nonlinear behavior, as illustrated by intrinsic thresholds in channel incision and emergence (Harrison et al., 2006) in the formation of anastomosing planform.

Field studies indicate that ephemeral channels can switch from aggrading to incising without a change in external variables, such as water and sediment inputs or relative base level (Schumm and Hadley, 1957). The change in channel behavior results from episodic longitudinal disconnectivity in water and sediment fluxes as surface flow infiltrates into the alluvial bed in dryland rivers, depositing sediment on the bed and creating locally oversteepened reaches at which headcuts subsequently form. In the case of anastomosing planform, channel-spanning logjams can facilitate overbank flow that leads to the formation of multiple, subparallel channels (Davies and Gibling, 2011; Wohl, 2011) where variations in river corridor geometry create longitudinal disconnectivity in the transport of large wood (Wohl and Cadol, 2011). Bank erosion within the multiple channels recruits additional wood for logjam formation and large logjams create erosionally resistant units that persist within the floodplain as channels migrate laterally (Collins et al., 2012). The simple interactions involved in wood recruitment and logjam formation result in an emergent, persistent channel planform. In these and other examples, temporal and spatial variations in connectivity create instabilities and feedbacks that manifest as nonlinear behavior.

2.6 River management

Connectivity is strongly influenced by spatial and temporal heterogeneity in river corridors (Burchsted et al., 2010; Wohl, 2016). The cumulative effect of diverse forms of river engineering and anthropogenic watershed alteration has been to reduce the spatial heterogeneity of rivers (Peipoch et al., 2015) and to reduce the temporal heterogeneity associated with water and sediment fluxes (Poff et al., 2007). This has altered connectivity, although the details vary. Infrastructure, such as levees and dams, reduces lateral and longitudinal connectivity, respectively, whereas channelization can increase longitudinal connectivity but decrease lateral and vertical connectivity.

As river management increasingly seeks to restore lost functions, such as retention and habitat abundance and diversity, processes of and controls on connectivity become critical (Jansson et al., 2007). Traditional design and operation of bridges (USBR and ERDC, 2016), culverts (Blakely et al., 2006; Park et al., 2008), levees (Galat et al., 1998), and dams (Olden and Naiman, 2010) are being modified to restore reduced connectivity and to facilitate the dispersal of organisms and fluxes of water and sediment that create and maintain habitat. Strategies are being developed to limit connectivity where connectivity has been artificially increased by altered flow regime (Hassan and Egozi, 2001), water transfers have allowed dispersal of invasive organisms (Waters et al., 2002), or riparian vegetation has been removed (Amoros and Bornette, 2002). Fundamentally, connectivity can provide a conceptual framework for understanding how human activities have modified river corridors, as well as how to mitigate the negative consequences of these modifications (Croke et al., 2017; Kondolf et al., 2006).

4.1 Relevant scales of connectivity

With increasing emphasis on connectivity, there is the danger that river management, in particular, may overly simplify connectivity under the assumption that increased connectivity is always good. Natural river corridors include numerous forms of at least temporary or local disconnectivity that enhance channel stability, attenuate hydrologic extremes, retain material moving down the river network, and enhance the dispersal of organisms (e.g., Burchsted et al., 2010; Poole, 2002). The key point is to identify natural or desired levels of connectivity within any particular river, as well as the mechanisms that create and maintain those levels of connectivity (Fryirs and Brierley, 2009). This is not necessarily straightforward. Water (Poff et al., 1997), sediment (Wohl et al., 2015a), and nutrient regimes (Dodds, 2007) vitally influence the sustainability of river ecosystems. In the same river, constraints on connectivity during large flows may limit sediment flux, whereas constraints on longitudinal hydrologic and nutrient connectivity during base flow or droughts may limit those fluxes. Because a river corridor is a multidimensional, multivariable, nonlinear system, identifying all relevant aspects of connectivity presents a significant challenge.

4.2 Quantifying and predicting connectivity

Many aspects of connectivity are intuitively obvious once they are identified: of course, tributary fans represent sediment storage sites that limit lateral and longitudinal sediment connectivity (e.g., Fryirs et al., 2007) and, of course, aquatic insects emerging from the stream are preyed on by riparian birds, creating lateral connectivity between streams and riparian zones (e.g., Baxter et al., 2005). The challenge lies in quantifying such relationships and being able to use quantitative understanding to predict how the river corridor will react to disturbances or management (e.g., Shafroth et al., 2010). How will quantities, grain size distribution, mode of transport, duration, frequency, and travel length of sediment increase if an alluvial fan is altered by construction? How will riparian predators respond if an introduced species of fish alters the emergence of aquatic insects (Baxter et al., 2004)? Developing this level of knowledge and predictive ability is likely to remain a major challenge for many years, yet is crucial to managing river corridors to achieve the desired conditions.

4.3 Managing connectivity

Given that restoring natural or desired levels of connectivity is increasingly a goal of river management, the quantitative predictive ability discussed above is likely to be critically important. In this context, natural levels of connectivity may not be as relevant as threshold levels of connectivity to restore desired effects. Bernhardt and Palmer (2011), for example, question the fuzzy logic of restoring limited sections of river in order to reverse catchment-scale degradation, but just how much does lateral connectivity between uplands and channels need to be limited by restored riparian vegetation, for example, in order to achieve targeted levels of nitrate retention and uptake? Bernhardt and Palmer (2011) also note the importance of considering upland-scale management that influences fluxes to the river corridor, particularly in urban settings where increased impervious areas (Walsh and Kunapo, 2009) and engineering, such as stormwater drainage and channelization, have created hyper-connectivity within channel networks (Walsh et al., 2016). The inclusion of upland land management reflects the recognition that connectivity outside of the river corridor (e.g., uplands-river corridor, atmosphere-river corridor, ground water-river corridor) can strongly influence river corridor functioning.

Effectively balancing human consumptive and other demands on rivers against restoration of connectivity requires a relatively high level of understanding of what management is trying to achieve (Blue and Brierley, 2016), how a particular area targeted for restoration is situated within a larger river catchment (Brierley and Fryirs, 2009, 2016; Fryirs et al., 2009), and what types and magnitudes of connectivity are critical to river function. The latter point includes distinguishing beneficial and detrimental connectivity. Beneficial connectivity can be defined as connectivity that creates a desired process or characteristic within a river corridor, whereas detrimental connectivity creates an undesirable process or characteristic. The designation of a particular form of connectivity as beneficial or detrimental may well be inconsistent between stakeholders. An engineer concerned primarily with reducing overbank flooding will view increased channel conveyance and longitudinal connectivity as beneficial, for example, whereas an ecologist concerned primarily with preventing the upstream migration of an introduced aquatic species may consider increased longitudinal connectivity in the same river corridor to be detrimental.

The development of predictive numerical models relevant to the processes or features (e.g., habitat) influenced by connectivity is necessary for effective, integrated river management in the future. As we move toward this goal, more qualitative, observational-based connectivity analyses have already made a difference in how we perceive and manage river corridors, as illustrated by river restoration projects explicitly designed to restore some aspect of connectivity (e.g., Tockner et al., 1998; Woolsey et al., 2007). Continuing improvements in our ability to quantify and numerically model connectivity will likely improve our ability to predict river responses to specific disturbances or management activities. This is exemplified by investigations of sediment connectivity in the Colorado River within the Grand Canyon, where more than a decade of hydrologic and sediment measurements have been used to progressively refine sediment budgets that inform experimental flow releases designed to restore river ecosystem functions (Melis, 2011). Although the substantial effort, time, and cost involved in this intensive investigation of sediment connectivity will not be feasible in many river-management contexts, the conceptual framework of connectivity can provide relevant insights for most aspects of river management.