1 Root biomass is not the same as water uptake

Vertical root distributions are widely thought to determine how much water a species can absorb, which species can coexist, and how much water reaches the atmosphere and aquifer in water-limited systems [1,2,3]. For example, Walter’s two-layer hypothesis suggests that shallow roots provide grasses pre-emptive and superior access to water as it enters the soil while deep roots allow woody plants to avoid competition with shallow-rooted grasses [4, 5]. This hypothesis has been popular for over 100 years, but a growing body of evidence suggests that by focusing on vertical root distributions and ignoring variation in root activity (i.e., the rate of water uptake per root length), root plasticity (i.e., the rate of change in active root distributions), and soil water flow, the two-layer hypothesis has oversimplified and potentially misguided our understanding of water uptake (Text box) [6,7,8,9].

Vertical root distributions Root biomass as a function of soil depth is most often measured by sieving and weighing or by image analyses in rhizotrons. These biomass measurements provide a direct measure of root biomass but an indirect measure of water uptake. In contrast, natural abundance stable isotope techniques and tracer techniques provide more direct measures of vertical water uptake, but more indirect measures of root biomass.

Fine roots It has long been recognized that large, suberized roots do not actively absorb water from the soil and so measurements of non-suberized roots or roots < 2mm diameter are used to infer the distribution of active roots.

Active roots Roots that actively absorb water or nutrients into cell membranes. Thin roots, roots with low water potentials, and roots with more or more active membrane transport proteins (e.g., aquaporins) will be more active (i.e., transport water more quickly into root cells).

Root activity The rate of resource (i.e., water) uptake. Active roots must be located in ‘wet’ soils to absorb water. Soil water will generally flow into roots when the water potentials in the roots are lower than the water potentials in the soil.

Root plasticity The rate of change in active root distributions. Species can change their active root distributions by growing new roots, thin roots, highly active roots, or by changing root-fungal interactions. Some species can make these changes quickly over large areas (plastic root responses) while other species maintain consistent vertical active root distributions (static root responses).

A central problem for the two-layer hypothesis, and our understanding of water uptake in general, is that root biomass is not the same as water uptake [6, 10, 11]. Some roots can absorb water more quickly than others (i.e., are more active) [12, 13]. Root hairs, root thickness, aquaporin abundance and activity and mycorrhizal colonization all affect the rate of water uptake [14,15,16]. Further, some species can quickly change their active root profiles in response to water availability (i.e., root plasticity), while other species have a more static rooting strategy [7, 17, 18]. Water uptake, therefore, is a function not just of root location, but also of root activity and root plasticity [19,20,21].

In addition to root location, activity, and plasticity, it is also important to know when and where water is available in the soil because even active roots cannot absorb water from dry soils [6, 22]. Fortunately, computational models based on physical processes are quite good at predicting water flow through soils and into plant roots [22, 23]. For example, soil water flow models can predict exactly how shallow water will be when precipitation events are small, soil texture is fine, evapotranspiration is fast, and roots are shallow [18, 22, 23]. These models are used extensively by hydrologists and environmental engineers to simulate runoff, aquifer recharge, salinity, and pollution plume dynamics [24,25,26,27].

Validation of soil water flow model predictions has revealed that root plasticity is important at the ecosystem level. For example, models that assume that plant roots ‘follow’ water as it percolates through the soil outperform models that assume that roots maintain a static distribution in the soil [18, 28]. It is likely that this root plasticity varies phylogenetically or as a suite of plant traits [8, 29]. Unfortunately, little is known about variation in root plasticity among species [20, 30, 31].

To summarize, active roots must be present in wet soils to absorb water. This seemingly simple concept has been difficult to demonstrate because it is difficult to measure water uptake across the root systems of different species in the field [2, 9, 32]. Addressing this knowledge gap can be expected to improve predictions of plant growth, coexistence, and response to climate change [31, 33, 34]. In short, it should not be surprising that the amount of water a root system can absorb could predict plant abundance in water-limited systems, but until recently, this process has been difficult to measure and understand [2, 9, 32].

2 An emerging experimental approach

Tracer-informed ecohydrologic modeling (i.e., tracer/modeling) approaches are providing new insight into root location, activity, and plasticity [9, 35,36,37]. Some of the most detailed of these approaches injects hydrologic tracers to different soil depths in different replicate plots [38,39,40]. By measuring tracer concentration in plant tissues or transpired water, it is possible to describe the distribution of active roots of different species [41, 42]. Injections can be repeated over time or space to describe root plasticity [17]. Though more difficult to perform than natural abundance stable isotope techniques, tracer techniques provide more detailed, deeper, and robust descriptions of species water uptake [32, 43, 44]. The level of detail provided by tracer experiments may be necessary because small differences in root water uptake can have biologically important effects on plant growth [9, 21, 45].

Both a strength and weakness of tracer approaches is that they measure water absorbed from a certain soil depth regardless of how that water was absorbed. Plants can manipulate water uptake by producing more roots, finer roots, roots with mycorrhizal symbionts, roots with more or more active aquaporins, by producing roots that outcompete roots of adjacent species, or by hydraulically redistributing water. Tracer uptake describes the amount of water absorbed from different depths, but does not distinguish these different pathways.

Tracer injections describe water uptake at a point in time. The amount of water that flows into different root systems between tracer injections can be estimated using soil water flow models [46, 47]. These models simulate how water moves over land, through the soil profile, or to the atmosphere via evaporation or root uptake. These processes are driven by precipitation, available energy for evaporation, resistance to evaporation (stomatal and aerodynamic), matric and gravitational forces on water in soil and the presence of plant roots [27, 48, 49]. These models typically use energy-balance approaches such as the Penman–Monteith or Hargreaves equations to simulate heat and water vapor transfers and they solve Richard’s equation to explain water flow through soils [23, 27, 48, 50]. These models can be used to simulate water flow into roots between injection dates, or for different environmental conditions (e.g., across the landscape or over time; Fig. 1) [51]. Root water uptake can then be converted to biomass production to estimate the growth of different plant species as a function of their root location, activity, and plasticity and their water use efficiency [22, 52, 53].

Fig. 1
figure 1

Water matching predicts grass and tree cover across a range of conditions in savannas of South Africa [51]. The response surfaces show the amount of water that tree or grass root systems were estimated to absorb as a function of soil water availability and mean annual precipitation. Tracer experiments were used to describe active root distributions in four sites with different soil (clay or sand) and climate [~ 350 mm or 700 mm mean annual precipitation (MAP)] conditions. Water uptake by active tree and grass root systems between sampling events was estimated using the soil water flow model Hydrus 1D. The mean depth of active tree roots (~ 30 cm) was slightly deeper than the mean depth of active grass roots (~ 20 cm). Slightly deeper roots were estimated to absorb more water than shallow roots under most conditions. There were two somewhat counterintuitive hydrologic ‘windows’ where shallow grass roots were expected to absorb more water than deeper tree roots. In dry conditions (MAP < 350 mm) with deep percolation, shallow roots had a water uptake advantage that predicted the observed increase in grass abundance (lower left corner). In wet conditions (MAP > 500 mm) with shallow water percolation (< 250 mm) grasses had an advantage (right edge). Water uptake predictions made using this tracer/modeling approach predicted tree and grass abundance on the landscape (see inserts) [51].

Full size image

In the handful of sites where it has been tested, the tracer/modeling approach has predicted species water uptake and abundance on the landscape. In a shrub-steppe site in North America, the tracer/modeling approach predicted landscape abundances (R2 = 0.90) of five species that represented 84% of plant cover [54]. In that site, larger plants demonstrated deeper and more dynamic rooting patterns that absorbed more water, and these plants were more abundant on the landscape. In sub-tropical savannas of South Africa, the tracer/modeling approach similarly found that deep roots extracted more water than shallow roots across a range of abiotic conditions [51]. The extent of this ‘deep root advantage’ predicted tree cover on the landscape (Fig. 1). In both studies, larger plants were found to have deeper and more dynamic rooting distributions, for example absorbing the most water at 20 cm one month and at 60 cm the next month [51, 54]. The fact that plant landscape abundance could be predicted from estimates of water uptake alone suggests that water uptake is a primary determinant of plant growth upon which other factors such as fire, herbivory, life history, water use efficiency, etc. act. An important aspect of the tracer/modeling approach is that model estimates of soil water content can be validated using observed soil moisture.

2.1 Water matching: a new framework for understanding water uptake

Here a framework for understanding and measuring water uptake that incorporates vertical root distributions, root activity, root plasticity, and soil water flow is proposed. In this framework, total transpiration demand is the ultimate cause of root strategies in water-limited ecosystems [55]. Niche partitioning resulting from these different strategies is the proximate cause of species growth and coexistence [56]. Species with large transpiration demands must access large soil water pools [21, 57]. Deep roots provide access to a wide range of soil depths and, therefore, a potentially large soil water pool (Fig. 2). Further, because they are large, deep roots are likely to have stored carbon reserves that allow the production of ‘flushes’ of fine roots that forage for water pulses [58,59,60]. These flushes can occur vertically or horizontally [61, 62]. These deep and dynamic root systems come with large construction and maintenance costs [63]. For example, larger plants are likely to be taller with large water potential gradients across long, costly transpiration systems. These systems are more likely to realize large repair and replacement costs due to cavitation [58,59,60,61,62]Large transpiration demand, therefore, induces a large investment/large return strategy [8, 64, 65]. This strategy can be described as deep and dynamic.

Fig. 2
figure 2

Larger and dominant plants have deeper roots than smaller plants and sub-dominant plants. The mean depth of tracer uptake by 28 tree, grass, and forb species (points) sampled in savannas of Kruger Park, South Africa versus plant height (TRY database) [51]. The positive slopes of the curves indicate that taller plants absorb deeper water than shorter plants. The higher grey curve indicates that dominant plants absorb deeper water than sub-dominant plants (blue points and curve)

Full size image

Plants with small transpiration demands tend to be small and produce herbaceous tissues that can be produced quickly from relatively small amounts of water [4, 34]. These plants can rapidly cover the ground and reproduce using smaller or more ephemeral pools of water [34, 66]. Shallow roots allow these species to intercept water as it enters the soil [4, 66]. This interception strategy often does not provide the most water (Fig. 2), but it provides enough water to allow growth and reproduction of small, fast-growing plants [51].

These strategies can be self-reinforcing. For deep-rooted species, shallow roots may result in a low return on investment due to competition from dense root mats of shallow-rooted species [67]. For shallow-rooted species, deep roots will come at the cost of aboveground growth and competitive ability with other shallow-rooted plants.

These competing strategies suggest that root activity and plasticity are especially important for larger and woodier species. This perspective explains that root biomass measurements may have misled our understanding of root function because some roots absorb water more quickly than others [11, 14]. It also suggests that measures of root activity and plasticity will be necessary to understand water uptake over time and space, particularly for woody species [7, 18, 19].

Large transpiration demands for an individual plant can be caused by several factors [48, 68]. Large size, leaf area, stomatal density and poor stomatal control will increase water demand [22, 69]. Taller plants will have less aerodynamic resistance and greater heat loading, both increasing water demand [68,69,70].

Site conditions will determine whether shallow or deep roots have a water uptake advantage [71, 72]. Shallow-rooted species benefit from conditions preventing deep percolation, such as frequent small precipitation events, clay soils, and precipitation during high evapotranspiration demand (i.e., summer) [73,74,75]. In contrast, deep-rooted species benefit from conditions allowing deep percolation, such as large infrequent precipitation events, sandy soils, and winter precipitation. In other words, plants with root systems that best ‘match’ water availability in a system will be able to absorb the most water and grow more than other species (Fig. 2).

Interestingly, there is a feedback between root distributions and availability that can affect the shallow or deep root advantage [51]. For example, dense, shallow roots will result in shallow percolation and water availability which will benefit shallow-rooted species. Deep roots will allow deeper percolation and water availability, which will benefit deeper-rooted species. These positive ecohydrologic feedbacks should act as a mechanism of positive frequency dependence that results in patchy vegetation distributions – a common, but poorly-understood vegetation pattern [76, 77].

The water matching hypothesis differs from the long-standing two-layer hypothesis in several ways. The water matching hypothesis predicts that in contrast to shallow-rooted plants, deep-rooted plants (1) typically absorb more water, (2) will dominate in arid and semiarid conditions (Fig. 3), and (3) rely more on variation in root activity and plasticity. Further, the water matching hypothesis explains that pre-emption allows grass coexistence in arid areas and that an ecohydrologic ‘window’ allows grass dominance where water availability is shallow. In contrast, it is difficult to understand how woody plants would co-exist with grasses if grasses are more water use efficient and have pre-emptive and superior access to soil water, as suggested by the two-layer hypothesis [4].

Fig. 3
figure 3

The shrub to grass cover ratio reveals greater shrub cover in more arid conditions in the western US. Greater shrub cover in arid conditions in the south and west of the US is consistent with the ‘water matching’ hypothesis and inconsistent with the two-layer hypothesis. Red values indicate greater shrub than grass cover. Blue values indicate greater grass than shrub cover. Analysis of data from the Bureau of Land Management’s Assessment, Inventory and Monitoring Program (https://www.blm.gov.aim)

Full size image

The water matching hyhpothesis is consistent with many existing hypotheses but provides important advantages [78]. The ‘two-pools’ model suggests deep roots provide maintenance water [79]. A complementary idea is that small resource pulses benefit smaller, fast-growing plants [80, 81]. Classic ecohydrologic theory suggests that water flow will vary with soil texture [5, 82] and precipitation event size [75, 83]. Several studies have described ‘optimal rooting’ strategies as essentially ‘as shallow as possible’ [84,85,86]. Others still have emphasized the role of phenology [87, 88]. The water matching hypothesis integrates these ecohydrologic concepts but provides (1) a new explanation for how and why root water uptake varies among different plant species and functional groups, and (2) provides a means of measuring these differences.

3 Concluding remarks

Many factors determine plant growth and landscape abundance. Fire, herbivory, disease, intrinsic growth rates, life history traits, plant soil feedbacks, human distrubance etc. can all affect plant growth. However, it is reasonable to expect that the ability of a plant to absorb water in water-limited ecosystems will play a foundational role on which these other factors will act. Despite its potential importance, water uptake by different species remains difficult to measure in the field. Better estimates of root water uptake over time can be expected to improve predictions of plant growth, coexistence, and response to climate change [3, 17]. The water matching hypothesis provides a framework for understanding how different plants absorb water and tracer/modeling approaches provide a means of testing this understanding. Where species abundances vary from predictions from the water matching hypothesis, it is likely that other factors such as herbivory, water use efficiency, and human disturbance override the effects of water uptake on plant abundance. Continued research in this direction can be expected to transform our understanding of water uptake from qualitative to quantitative and allow a better understanding of the various factors that determine plant growth and coexistence under both current and anticipated climate conditions.