3. Ecohydrology: Processes and Implications for Rangelands

Ecosystem services regulating water on rangelands include those that affect the amount, timing, and quality of blue water flows. These are to a large extent determined at the first critical juncture of the water cycle—on the soil surface, where water either infiltrates or becomes overland flow. For this reason, a great deal of research, most of it conducted at the point or plot scale, has focused on understanding the infiltration process and how it is affected by different management strategies (Pyke et al. 2002; Stavi et al. 2009).

Water that does not infiltrate, of course, becomes overland flow or surface runoff. It is at the hillslope scale that important interactions take place between vegetation patches and runoff. Surface runoff may be captured and stored by vegetation patches or other surface obstructions, a process known as runoff–runon (Ludwig et al. 2005).

An important conceptual advance in describing and clarifying the linkages between surface runoff and vegetation patches is the trigger-transfer-reserve-pulse (TTRP) framework (Fig. 3.2) (Ludwig et al. 1997, 2005). This framework was originally proposed as a way of describing runoff–runon processes observed in areas of banded vegetation (Anderson and Hodgkinson 1997; Dunkerley and Brown 1999; Valentin and d’Herbes 1999; Tongway and Ludwig 2001); it was subsequently verified for other vegetation patch types in semiarid settings (Reid et al. 1999; Wilcox et al. 2003a; Ludwig et al. 2005). The framework assumes that the redistribution of resources from source areas (bare patches) to sink areas is a fundamental process within drylands , and that this process may be disrupted if vegetation patch structure is altered by disturbances such as overgrazing or multiyear drought. These dynamics govern how runoff and runon vary with scale in semiarid settings. In regions where runoff is efficiently captured down slope by vegetation patches, unit-area runoff and erosion diminish rapidly with increasing scale. But where vegetation patch structure has been disturbed and runoff is not efficiently captured, declines in runoff with increasing scale are much smaller (Fig. 3.3). Erosion may even increase as runoff increases with increasing scale, leading to rilling and gully formation (Wilcox et al. 2003a; Moreno de las Heras et al. 2010).

Water that enters the soil may either evaporate, be transpired by plants, or drain out of the root zone and ultimately contribute to groundwater and streamflow. On rangelands, drainage is generally (but not always) a small percentage (<5 %) of the water budget (Wilcox et al. 2003b). Vegetation management that alters the amount of woody plants may affect drainage because woody plants are deeper rooted than grasses or forbs and tend to transpire more water. Therefore, woodlands and forests generally use more water than grasslands (Zhang et al. 2001). The relationship is complex, however, especially for drylands, as it is modified by other factors—including climate, soils, and topographic position (Huxman et al. 2005).

The shrub–streamflow framework (Fig. 3.4) was developed to aid in determining which landscapes are most “hydrologically sensitive” to changes in woody plant cover. A hydrologically sensitive landscape is one in which a shift in functional vegetation type (woody to nonwoody or vice versa) causes an important shift in the water balance. Hydrological sensitivity is dictated or strongly influenced by how vegetation change affects drainage; and it is also influenced by factors such as depth to water table, soil and geological characteristics, and topographic position. The shrub–streamflow framework uses these concepts to predict where hydrologically sensitive shrublands might exist (Wilcox et al. 2006).

The first criterion for hydrologic sensitivity—the presence of shallow groundwater—is likely to be stronger where the groundwater table is within a few meters of the surface, as in riparian zones or groundwater-coupled rangelands. Obviously, this condition affords more opportunity for interaction between deep-rooted vegetation and groundwater.

Seasonality of precipitation is a second criterion in determining hydrologic sensitivity. Those rangelands having the greatest potential for water to move deeply into the soil—beneath the rooting zone of herbaceous plants—will be the most hydrologically sensitive. Such deep drainage occurs in regions where winter precipitation is high. It is no coincidence that the strongest linkage between woody plants and streamflow has been observed in Mediterranean climates where precipitation is often “out of phase” with potential ET. For example, in South Africa (van Wilgen et al. 1998), Spain (Puigdefabregas and Mendizabal 1998), Australia (Walker et al. 1993), and California (Hibbert 1983), dramatic changes in drainage have been observed following vegetation changes in native shrublands. Similarly, shrublands in which soil recharge comes mainly from snowmelt may be hydrologically sensitive; a large pulse of melting snow often produces enough water to saturate or exceed the water storage capacity of the upper soil (Baker 1984; Seyfried and Wilcox 2006).

Finally, soil or geological conditions also determine hydrologic sensitivity, by affecting the potential for deep drainage. We would expect higher hydrologic sensitivity where soils are sandy (Moore et al. 2012; Dzikiti et al. 2013), are deeply cracked (Richardson et al. 1979), or are shallow and overlie fractured bedrock (Huang et al. 2006).

For many rangelands , the opportunities for deep drainage are quite limited, because of either climate or soils. In these landscapes, shifting from a grassland to a shrubland will have little effect on the overall water balance, but may nevertheless affect drainage in important ways. For example, in areas where even small amounts of drainage can be significant (such as sites where hazardous wastes are buried), the presence of deep-rooted shrubs may ensure that drainage below the root zone seldom occurs (Scanlon et al. 2005b, c; Seyfried et al. 2005).

The common perception is that rangelands are exclusively dryland environments. Even when the presence of riparian environments is recognized, these are generally seen as a minor component of the entire landscape system. However, there is a growing body of research showing that riparian environments are not only key habitats for rangeland fauna and flora, but also critical providers of ecosystem services to rangeland inhabitants (Milton 1990; Dean et al. 1999, 2002; Naiman et al. 2002; Sabo et al. 2005; Le Maitre et al. 2007; Soykan and Sabo 2009; Jones et al. 2010; Merritt and Bateman 2012).

Many of the features and key processes in groundwater-coupled systems are likewise found in riparian environments. The principal differences are that (1) riparian zones receive inputs of surface water, often from areas far upstream, that play a major role in their ecology (Boulton and Hancock 2006; Bunn et al. 2006; Nagler et al. 2008) and (2) the dynamics of riparian zones are strongly influenced by flow regimes and fluvial processes (Ward 1998), particularly sediment movement (Naiman et al. 1999; Tabacchi et al. 2000).

Riparian environments are typically located in the lowest parts of a landscape , where surface water (overland flow) and groundwater (subsurface flow) collect; thus they integrate outputs from all watershed-scale processes (Naiman et al. 2002). Their structure is long and narrow with a very large perimeter-to-area ratio—which makes them highly connected to, and thereby highly influenced by, events originating in the adjacent drylands. The headwaters section of a river typically has steep gradients, and the river bed contains rocks or boulders with little accumulation of fine sediments, whereas the middle and lower reaches are characterized by extensive alluvial deposits and wide floodplains (e.g., Nile, Platte, Euphrates, Ganges) (Vannote et al. 1980; Wiens 2002). These deposits are typically heterogeneous, with coarse sediments that can capture, store, and transmit large volumes of water interspersed vertically and horizontally with fine sediments that have a lower storage capacity and low transmissivity (Blasch et al. 2006; Morin et al. 2009). In humid-to-semiarid landscapes, rainfall is sufficient to generate runoff and groundwater that sustain river systems, and the rivers are gaining water, albeit seasonally. But in arid landscapes the rivers are often hydrologically disconnected from the adjacent dryland areas—except for losing water to the floodplain, and gaining water during rainy seasons or after very high rainfall events.

Water use by riparian communities has been intensively studied in the southwestern USA, mainly to estimate transmission losses, but also to quantify the effects of woody species such as the invasive or saltcedar (Tamarix spp.) on these losses. Evapotranspiration rates can exceed annual rainfall in these arid environments (Doody et al. 2011). Early research on saltcedar suggested that its water use was very high: up to 200 m³ ha⁻¹ day⁻¹ (Sala et al. 1996); but subsequent research has found that it is much lower and close to that of the native poplars and willows that grow in riparian systems of the southwest. Therefore, removal of these plants would have little effect on water loss if native riparian shrubs remain (Nagler et al. 2009; Shafroth et al. 2010b; Doody et al. 2011). Similar work in Australia found that although invasive Salix species in the river channel can use large quantities of water (±2000 mm year⁻¹ vs. 1500 mm year⁻¹ for open water), overall ET for this invasive species is very similar to that found for native riparian Eucalyptus forest (Doody and Benyon 2011; Doody et al. 2011).

In the floodplain of the perennial San Pedro River in Arizona, Prosopis woodlands have replaced native grasslands, increasing ET from 407 to 639 mm year⁻¹ (Scott et al. 2006). Evapotranspiration from Prosopis woodlands in floodplains linked to perennial rivers ranges from about 350 to 750 mm year⁻¹ (Scott et al. 2004, 2008), which suggests that other perennial river systems (such as those in South Africa where native tree species such as Acacia karroo are sparse or absent) could be similarly affected by invasion of non-native species.

The Working for Water program in South Africa, a national initiative for removal of invasive plants, emphasizes clearing to increase river flows (van Wilgen et al. 1998). Extensive invasions by Acacia mearnsii, Eucalyptus species, willows, and poplars have taken place along perennial rivers in the arid grasslands and savannas of the interior, where the native riparian species are mainly shrubs or small trees. If the difference between the annual ET from stands of these species and that from native species is as much as indicated by some studies (Dye and Jarmain 2004), or by data for willows and eucalypts from Doody and Benyon (2011) and Doody et al. (2011), removal could lead to an increase in river flows that would be substantial and very important for downstream water users and ecosystems. However, there may be other cases in which streamflows could be significantly reduced, such as invasions of species that are high water users along ephemeral streams (Doody et al. 2011; Hultine and Bush 2011).

The distinct species composition, structure, and dynamics of riparian environments generate a suite of ecosystem services very different from that of dryland environments. This makes them a key resource area, particularly in developing countries where they are less likely to have undergone extensive transformation by agriculture and other activities (Tockner and Stanford 2002; Kgathi et al. 2005; Brauman et al. 2007).

Recent work has also documented the extent to which large fauna can alter riparian processes. For example, in riparian ecosystems, large-scale earthworks are created mainly by the activities of fauna, particularly large mammals, which shape floodplains at a range of scales, from the microtopographical to that of river channels (Naiman and Rogers 1997; Moore 2006). Ecosystem modifications brought about by beavers, through the construction of dams, have been well studied; but much less is known about the ecological roles played by large mammals. In wetlands like the Okavango, large mammals (elephant, buffalo, hippopotamus) open up flow paths for water through reeds, changing water circulation patterns. Similarly, warthogs carve out feeding patches that form temporary pools during the wet season, creating habitats for many other species to complete their life cycles.

Riparian vegetation provides important feedbacks to the river system: it captures and stabilizes sediments, shapes river channels, and determines and regulates biotic processes (and, thus, water quality) (Tabacchi et al. 2000; Naiman et al. 2002). By these means, the vegetation creates its own habitat as well as that for animal species, and ensures its replacement through succession. Further, riparian vegetation serves as a buffer, shielding the aquatic ecosystems from the effects of land-use practices in adjacent environments—by filtering sediments, nutrients, and other pollutants (Naiman et al. 1999; Tabacchi et al. 2000; Brauman et al. 2007; Corenblit et al. 2009).

In all rangelands the recharge, transport, and quality of groundwater depend on the nature of deep drainage and solute leaching. Where groundwater tables are shallow, reciprocal interactions between vegetation and groundwater are often observed (Le Maitre et al. 1999). This two-way exchange of water and solutes increases primary and secondary production, particularly under dry climatic conditions; at the same time; however, it renders water, soil, and vegetation resources more vulnerable to land management.

Groundwater-coupled rangelands—those in which shallow water tables are found, and the potential for a strong coupling between vegetation and groundwater exists—are increasingly recognized as important, and yet are poorly understood. These ecosystems have been categorized as “groundwater dependent”; yet the degree to which they are dependent varies greatly in time and space (Boulton and Hancock 2006; Eamus and Froend 2006). For this reason, we prefer the term “groundwater coupled” to describe the broad array of rangelands characterized by shallow water tables. We do know that vegetation has a major role in regulating groundwater resources in these systems, and significant strides have been made recently in understanding these ecohydrological interactions.

Many rangeland landscapes host, at their lowest topographic points, shallow groundwater zones that are sustained by local or distant recharge sources (Tóth 1999). In dry rangelands, where evapotranspiration recycles essentially all precipitation inputs back to the atmosphere, local recharge is negligible (Scanlon et al. 2006) and such shallow aquifers are rare. Regions where they do occur are characterized by sandy or rocky soils (such as sand dunes, fractured rock outcrops), highly seasonal and intense precipitation regimes, and zones of extensive lateral flow and intense runon. In such regions, at least some deep drainage into the saturated zone will eventually take place (Scanlon and Goldsmith 1997; Athavale et al. 1998; Seyfried et al. 2005; Small 2005; Gates et al. 2008). Recharge from more distant sources is particularly significant in arid regions located downstream of water-yielding mountains . For example, shallow water tables, wetlands, and lakes fed by mountain snowmelt are found at topographic lows within sand-dune rangelands such as the Great Sand Dunes of Colorado (Wurster et al. 2003), the Bahrain Jaram and Taklamakan deserts in China (Thomas et al. 2000; Chen et al. 2004; Gates et al. 2008), and the Monte desert in Argentina (Jobbágy et al. 2011).

In more humid rangelands , local groundwater recharge is more widespread and frequent, and shallow water tables are commonly found at topographic lows and along riparian corridors (Jobbágy and Jackson 2007). Finally, shallow water tables are a widespread feature in very flat, sedimentary rangelands (Fan et al. 2013) (Fig. 3.5); some conspicuous examples are the steppes of Western Siberia, the Northern Caspian plains in Asia, the grasslands of the Llanos and Pampas in South America, and the Miombo systems of sub-Saharan Africa (von der Heyden 2004).

The cycling of water on rangelands is obviously driven by many factors, some of which have been discussed in the previous section. A fundamental factor is the process of evapotranspiration (ET), which on most rangelands accounts for more than 95 % of the water budget (Wilcox et al. 2003b). Evapotranspiration is the sum total of interception—water captured by vegetation or litter and subsequently evaporated, transpiration, and evaporation from the soil or surface of water bodies. Recently, ecohydrologists have recognized the importance of better understanding the dynamics of ET, and in particular have placed more emphasis on accurately partitioning ET into soil evaporation and transpiration (Newman et al. 2006). Soil evaporation, from an ecohydrological perspective, is not a productive use of water because it does not contribute to plant productivity and carbon sequestration, food, fiber, or fuel production (D’Odorico et al. 2013a). This insight indicates that the main focus of ecohydrology should be to develop methods for better partitioning of the green water resources (i.e., decrease soil evaporation and increase transpiration) in semiarid and subhumid landscapes (Falkenmark and Rockstrom 2004). The same insight is motivating ecohydrologists to better understand and quantify ET .

Evapotranspiration can be partitioned into three components: (1) water that is intercepted by foliage and then evaporates back to the air; (2) water that is intercepted by litter on the soil surface, infiltrates into that litter and into the soil, and then evaporates; and (3) water that infiltrates into soil, is absorbed by plants, and later transpired back to the atmosphere . An additional process, previously not taken into account, is the potential for plants to absorb foliar-intercepted rainfall (Breshears et al. 2008); this process can be important during protracted periods of water stress, allowing plants to take advantage of rainfall events that are just large enough to be intercepted, but not large enough to infiltrate soil (Loik et al. 2004; Owens et al. 2006). This process has not been fully investigated, and the degree to which it may affect multiple species of plants is not yet known.

The rate at which soil evaporation takes place depends on several variables, including soil texture, soil temperature, and near-surface wind; these in turn are affected by basic properties of rangeland structure, such as the amount and type of woody-plant canopy cover. Recently, considerable work has focused on ways to identify the linkages between vegetation characteristics, soil evaporation, and microclimates for a diverse set of rangeland vegetation types —including mesquite, piñon-juniper, ponderosa pine, eucalypt, and saguaro cactus (Breshears and Ludwig 2010; Royer et al. 2010; Villegas et al. 2010a, b; Zou et al. 2010; Royer et al. 2012). Other recent work has focused on understanding how changes in woody plant cover may affect the ratio of transpiration to ET (Wang et al. 2010b, 2012a).

Understanding spatial dynamics and scale relationships has been a formidable challenge in both ecology and hydrology and, by extension, ecohydrology (Wood et al. 1990; Sivapalan and Kalma 1995; Sposito 1998; Grayson and Bloschl 2000; Western et al. 2001). Nevertheless, important advances have been made (Newman et al. 2006; Asbjornsen et al. 2011). For example, comparative studies across spatial scales have revealed the nonlinear nature of runoff and erosion with changing scales and how disturbance alters these relationships (Fig. 3.3) (Bergkamp 1998a; Puigdefabregas et al. 1999; Wilcox et al. 2003a; Favreau et al. 2009; Moreno de las Heras et al. 2010). Nonlinear responses in runoff and erosion are the result of redistribution across the landscape as well as alterations in runoff generation mechanisms with changing scale (Seyfried and Wilcox 1995).

Similarly significant strides have been made in quantifying the spatial variability of infiltration at the hillslope scale (Berndtsson and Larson 1987; Seyfried 1991; Pierson et al. 1994, 2001; Bhark and Small 2003; Daryanto et al. 2013). Infiltration capacities are generally higher under shrub canopies than in intercanopy areas, and these differences markedly influence patterns of soil moisture (Breshears and Barnes 1994, 1999). In addition, runon from intercanopy patches often contributes additional water to the shrub patches (Ludwig et al. 2005). Correspondingly, a number of conceptual advances have enhanced our understanding of the spatial variability of vegetation patterns on rangelands and how these are regulated by rainfall and runoff (HilleRisLambers et al. 2001; Rietkerk et al. 2002, 2004; Thompson et al. 2011).

Faunal activities also play an important role in structuring dryland landscapes as well. One feature of many arid landscapes is the formation of mounds, generally regularly dispersed, that range in diameter from a few meters to tens of meters. Known as mima mounds in the western USA, they are called heuweltjies in South Africa, where they cover from 14 to 25 % of the landscape (Lovegrove and Siegfried 1986, 1989; Whitford and Kay 1999). Their regular distribution is probably the result of competition among fauna for resources (Lovegrove and Siegfried 1986; Laurie 2002). Most authors agree that these enigmatic features are initiated by animals, whether mammals or invertebrates. One theory regarding the heuweltjies is that they developed over buried termite nests (Milton and Dean 1990; Moore and Picker 1991), but a recent paper argues that they are relicts of shrub-clump-controlled erosion processes (Cramer et al. 2012). Whatever their origin, the accumulation of transported organic matter, softer soil, and food remains they contain increases their fertility (Midgley and Musil 1990) and infiltration rates (Dean 1992), supports a distinctive suite of plant species (Knight et al. 1989), and attracts faunal activity—digging by termite-eating mammals , burrowing by rodents and/or nesting ostrich (Lovegrove and Siegfried 1986, 1989; Milton and Dean 1990), and foraging by game and domestic livestock (Armstrong and Siegfried 1990; Kunz et al. 2012). The movement of water across and between the vegetation mosaic and the heuweltjies has not been studied to determine whether these mounds contribute to groundwater recharge; but their higher infiltration rates (Dean 1992) suggest that their ecohydrological function may be analogous to that of the vegetation patches; that is, they may capture and filter runoff and act as foci for deep infiltration and recharge of groundwater.

The origins of the mima mounds in North America are no less controversial, but in this case small mammals (gophers) appear to be the primary drivers for the accumulation of materials (Whitford and Kay 1999; Jackson et al. 2003; Horwath and Johnson 2006; Johnson and Horwath-Burnham 2012). Whether or not that proves to be the sole explanation, these features also accumulate materials and alter the ecohydrology of the landscape. These important soil modifications justify the need for further research into the ecohydrological consequences of soil (Westbrook et al. 2013).

As noted by Vivoni (2012), our understanding of the role of scale and spatial variability in ecohydrological processes on rangelands will certainly increase in the future as remote-sensing and computational capabilities continue to progress.

Ecological thresholds and feedback loops are intimately related (Runyan et al. 2012; D’Odorico et al. 2013a). Threshold behavior occurs when a relatively small change in external drivers causes a disproportionally large response. A classic example of an ecological threshold is the transition between two stable states—such as the transition from a grassland or savanna to woodland or highly eroded state (D’Odorico et al. 2013a). The shift or change in state is induced and maintained by positive feedbacks that destabilize the system (Chapter 6, this volume). Examples of positive feedbacks are those between vegetation cover and (1) erosion, (2) soil moisture, and (3) climate (Runyan et al. 2012; D’Odorico et al. 2013a). The desertification feedback loop presented in D’Odorico et al. (2013a) (Fig. 3.6) illustrates these: A decrease in vegetation cover triggers the loss of water, nutrients, and soil that may as changes in albedo and evapotranspiration. All of these changes in turn create an environment that is less conducive to vegetation growth. In the last decade in particular, a considerable amount of work has been done that helps us better understand feedback loops and their important role in ecohydrological interactions (D’Odorico et al. 2007, 2012, 2013a, b; Stavi et al. 2009; Runyan et al. 2012; Turnbull et al. 2012).

Hydrological connectivity refers to the water-mediated transfer of matter, energy, and organisms within or between elements of the hydrologic cycle (Pringle 2001). We now recognize that hydrological connectivity is essential for ecological integrity—and, more important, that activities by humans that disrupt this connectivity (dams, interbasin water transfers, etc.) can have dramatic negative consequences (Pringle 2003). “Connectivity” can be more broadly understood as the transfer of energy, matter, and organisms by not only water but also other vectors—such as wind and animals (Peters et al. 2006; Okin et al. 2009). One of the major benefits of studying connectivity in physical processes is that it identifies cross-scale interactions. For example, how do various different stomata in individual grass leaves , when under stress (e.g., from grazing or drought), function to modify water fluxes at the landscape scale? Answering such questions, on the basis of information from smaller scales, will significantly improve our ability to make predictions at larger scales (Peters et al. 2004). Hydrological connectivity has proved useful in explaining ecohydrological patterns on at the landscape scale as previously indicated (Wainwright et al. 2011). However, quantifying connectivity among different scales is still a major challenge, owing largely to a lack of a conceptual framework and modeling approaches applicable at multiple scales (Miller et al. 2012). Analogical models, which simulate the behaviors of complex physical systems using laws and theorems known to control components of those systems, may be able to fill some of these gaps. Recently, Wang et al. (2012b) developed a conceptual framework that uses electrical circuit analogies and Thévenin’s theorem to upscale ecohydrological and biogeochemical processes from point scales to watershed scales. This conceptual work, by providing a means of representing concomitant processes at both small and large spatial scales, may prove useful for multi-scale rangeland management efforts.

A number of important conceptual advances have improved our understanding of hydrological connectivity and flows—longitudinal, lateral, and vertical—within river systems as well as between river systems and landscapes, and the importance of this connectivity for river ecosystem structure, functioning, and maintenance of ecosystem services (Naiman et al. 1999; Ward et al. 2001; Wiens 2002; Caylor et al. 2004; Boulton and Hancock 2006). Combined with hydrogeomorphology, connectivity processes play a vital role in the structuring of river systems and the ecosystem services they provide (Thorp et al. 2006, 2010) (Fig. 3.8). The implication, for those involved in land management and in water resource management—two traditionally separate policy and legislative domains—is important: the two are actually inseparable (Postel and Thompson 2005). In fact, rivers are complex social–ecological systems, and if we are to ensure continued delivery of the numerous essential ecosystem services they provide, including their traditional use as water conduits , we must advance our knowledge of not only the scientific but also the social and economic aspects of managing them (Chapter 8, this volume).