Woody Plant Encroachment Impacts on Groundwater Recharge : A Review

Woody plant encroachment has profound impacts on the sustainable management of water resources in water-limited ecosystems. However, our understanding of the effects of this global phenomenon on groundwater recharge at local and regional scales is limited. Here, we reviewed studies related to (i) recharge estimation methods; (ii) mechanisms by which woody plants impact groundwater recharge; (iii) impacts of woody plant on recharge across different soil and geology; (iv) hydrological repercussions of woody plant removal; and (v) research gaps and needs for groundwater studies. We identified six different methods: water balance, water table, isotopes, chloride mass balance, electrical geophysical imaging, and modeling were used to study the impact of woody encroachment on groundwater. Woody plant encroachment could alter soil infiltration rates, soil water storage, transpiration, interception, and subsurface pathways to affect groundwater recharge. The impact is highly variable, with the extent and the magnitude varying across the soil, substrate, plant cover, and topographic locations. Our review revealed mixed effects of woody plant removal on groundwater recharge. Studies of litter interception, root water uptake, soil moisture dynamics, and deep percolation along with the progression of woody plant encroachment are still limited, warranting further experimental studies focusing on groundwater recharge. Overall, information about woody plant encroachment impacts on groundwater resources across a range of scales is essential for long-range planning of water resources.


Groundwater Recharge and Estimation Methods
Groundwater, the water that flows or seeps downward and fills porous spaces in soil or geologic strata, accounts for 99% of the earth's liquid fresh water, and provides drinking water to more than two billion people and irrigation water to 38% of the global croplands [33,34].The total volume of groundwater is estimated at 22.6 million km 3 in the upper two kilometers of the continental crust [35].Table 1 shows groundwater extraction in ten major countries for different sectors.Global water use for irrigation accounts for >70% of groundwater withdrawal every year [36].Groundwater withdrawal at rates higher than natural renewal rates has caused aquifer depletion in different parts of the world [34].For example, total groundwater depletion in subhumid to arid regions worldwide increased from 126 km 3 year −1 in 1960 to 283 km 3 year −1 in 2000 [37].Total groundwater depletion in the USA is estimated at 32 km 3 per year −1 [37].Groundwater depletion reduces groundwater discharge to streams and reduces yields from water wells, increases pumping cost, impairs water quality, and negatively affects terrestrial and aquatic ecosystems [38].
Table 1.Ten countries with the highest estimated groundwater (GW) extraction (Source: National Ground Water Association [33]).

Groundwater Recharge and Estimation Methods
Groundwater, the water that flows or seeps downward and fills porous spaces in soil or geologic strata, accounts for 99% of the earth's liquid fresh water, and provides drinking water to more than two billion people and irrigation water to 38% of the global croplands [33,34].The total volume of groundwater is estimated at 22.6 million km 3 in the upper two kilometers of the continental crust [35].Table 1 shows groundwater extraction in ten major countries for different sectors.Global water use for irrigation accounts for >70% of groundwater withdrawal every year [36].Groundwater withdrawal at rates higher than natural renewal rates has caused aquifer depletion in different parts of the world [34].For example, total groundwater depletion in subhumid to arid regions worldwide increased from 126 km 3 year −1 in 1960 to 283 km 3 year −1 in 2000 [37].Total groundwater depletion in the USA is estimated at 32 km 3 per year −1 [37].Groundwater depletion reduces groundwater discharge to streams and reduces yields from water wells, increases pumping cost, impairs water quality, and negatively affects terrestrial and aquatic ecosystems [38].
Groundwater recharge or deep drainage is the downward movement of water from the surface to the saturated zone.In vadose zone studies, "recharge" is synonymously used for "net infiltration", "drainage", "percolation", and "residual flux" to connote water movement below the rooting zone [39].Direct or diffuse recharge refers to precipitation water percolating vertically through the unsaturated zone [40,41].Indirect recharge refers to concentrated or localized recharge from surface depression [40].Groundwater recharge depends on numerous factors such as climate, land use and land cover, geology, soil texture and structure, irrigation water use, among others.These factors may work independently to influence recharge or may work in concert, interacting in complex ways to affect recharge.Quite directly, increased rainfall, in general, translates into increased recharge.More complex, a change in land cover may alter the land surface-atmosphere interactions and therefore the groundwater recharge [6,31,41], while a complete conversion of grassland to woodland at a regional scale could not only reduce recharge but also feedback to influence the regional climate [42] leading to further changes in recharge potential.In arid and semiarid regions, woody plants generally develop more extensive root systems in subsoils to meet the increase in transpiration demand and consequently these woody plants decrease the potential for local recharge of groundwater.Groundwater recharge varies across and between the seasons.An isotopic analysis of precipitation and groundwater from 54 global locations revealed higher groundwater recharge ratios during winter in arid and temperate climates and during wet seasons in tropical climates [43].Different methods to quantify groundwater recharge have been widely discussed in past reviews [39,40,44], but these reviews did not emphasize quantification in landscapes undergoing woody encroachment.Groundwater recharge monitoring techniques, based on surface-water, include water balance, channel-water budget, seepage meters, baseflow discharge, heat tracers, stable isotopes, and rainfall-runoff modeling.Techniques based on unsaturated zone studies include lysimeters, zero-flux plane, Darcy's law, tracers (bromide, 3 H and dyes), chloride mass balance, geophysical imaging, and unsaturated-zone modeling.Techniques based on saturated zone studies include the water table fluctuation method, tracers, chloride mass balance, and numerical modeling [39,44].In this paper, we provide a summary of the frequently used techniques (Table 2) to quantify groundwater recharge in the context of woody plant encroachment.

Water Balance Method
Water balance method is a residual approach of recharge estimation.In this method, recharge is estimated as the difference between the input of precipitation and output by evapotranspiration and runoff after accounting for the change in soil water in the unsaturated zone.
where, G = groundwater recharge, P = precipitation, ET = evapotranspiration, Q = runoff, and delta S = change in soil water storage.Further, where I = water intercepted by the canopy, E= evaporation from soil, and T = water transpired by plants.Weltz and Blackburn [45] evaluated different components of water balance in the Eastern Rio Grande Plain of Texas to understand the impact of vegetation manipulation on water yield.Results showed that replacing shrub by grasses had a marginal impact on water yield and minimal increase in percolation.Wilcox et al. [46] measured long-term water balance in a semiarid mesquite rangeland undergoing redberry juniper invasion.Results showed that 99% of precipitation was lost to evapotranspiration, and relatively low percolation occurred below one-meter depth of soil.Huxman et al. [9] used the water balance model to discern the impacts of woody plants on different water balance components in water-limited ecosystems.They reported that woody encroachment could potentially increase water loss by transpiration that would otherwise be available for recharge.In drylands, ET is frequently >90% of P, and as such, there is little rainwater left available for recharge.Therefore, it is imperative to be able to adequately measure the recharge to the system.While rainfall and runoff can be fairly accurately quantified, errors in the estimates of evapotranspiration and water storage change in the unsaturated zone will translate into error in the recharge estimate [40].

Water Table Method
The water table method has been used since the 1920's to estimate groundwater recharge [47].Measurements of the rise and fall of water table can be used to estimate the recharge or discharge of groundwater for a site [48].The underlying assumption is that the water level rise is driven by rainfall recharging the aquifer.Therefore, a rise in the water level can be inferred to be recharge arriving at water table [40,47].The fluctuation in the water table is affected by land use and land cover changes, which modify the atmospheric water fluxes and the spatial and temporal heterogeneity of soil moisture.For additional details about the method, see Healy and Cook [47].
Oliveira et al. [49,50] monitored groundwater recharge in Brazilian cerrado using water table method.Water table fluctuation measured through a pressure sensor in a 42-m deep well from December 2011 to July 2014 in an undisturbed cerrado showed no evidence of net groundwater table changes.Jobbagy and Jackson [51] reported that a Eucalyptus plantation reduced groundwater recharge, and decreased water table by 38 cm compared to adjacent grassland during a two-year period in the Argentine Pampas.Acharya et al. [30] monitored water levels of perched aquifers under grassland and grassland encroached by Juniperus virginiana (Eastern redcedar) in the South-Central Great Plains, USA.They found that water level was higher under grassland than under Eastern redcedar woodland; this observation indicates that woody plants are likely to decrease water table in a perched aquifer.These results also suggest that woody plants may reduce or have no effect on groundwater recharge.Indeed, the influence of vegetation on groundwater recharge depends on the thickness of the vadose zone in rangelands with shallow water tables [51,52], site and climate, among others.Piezometric data and water table method are best applied to areas with shallow water tables [47].However, it may be difficult to discern land use effects [53], and account for the steady recharge rate [47].Additionally, the water table fluctuation method requires close monitoring of groundwater wells and the volume of water yielded from soil or rock through gravity drainage [40].Sometimes, air trapped between wetting font and water table, "Lisse effect", may cause a rise in water table, and be mistaken for recharge.The water table method is inadequate in systems where activities like large-scale pumping are ongoing [54].

Isotopes
The commonly used isotopes in groundwater studies include 3 H, 2 H, 18 O, 14 C, 13 C, 36 Cl, and 15 N.The stable isotopes of O and H are increasingly being used to evaluate sources of plant water, plant water use, the mechanisms and pattern of groundwater recharge, and the water flow paths in lands experiencing woody encroachment [55][56][57][58].Comparison of the isotopic composition of plant xylem water with soil water from different depths helps to define soil water use.McDonnell [59] indicated potential use of isotopes to disentangle water sources from two worlds: (i) Water that is used by trees with no contribution to runoff/streamflow and (ii) mobile water contributing to infiltration, runoff, and recharge.
Examining the isotopic compositions of water from different sources provides important information on groundwater recharge.Precipitation naturally differs in isotopic composition resulting in distinctive isotopic signatures of groundwater [60].The isotopic values of groundwater can be co-related to the water table.For example, a decrease in δ 18 O value may indicate a drop in the water table.Isotopic studies on plant water use indicate differential water use pattern and effect on recharge among plant species.Rossatto et al. [56] used δ 18 O and δ 2 H in a neotropical savanna and found that woody plants use water from deeper soil layers than do grasses.However, in another study, no difference was observed in depth of water source used during late summer/early fall among co-occurring woody plants Prosopis glandulosa and Zanthoxylum fagara, although rooting depth differed [61].Evaristo and McDonnell [58] conducted a meta-analysis on plant xylem water stable isotope to understand plant water use, and indicated that plants use 37% of groundwater globally.Barbeta and Penuelas [57] compiled and analyzed 35 isotopic studies to understand groundwater use and reported that groundwater is most likely used by plants in areas with lower precipitation and during dry seasons.
Use of isotopes is relatively advantageous than water balance method because they often rely on long-term water fluxes compared with a water balance method.The isotopic method avoids frequent field visit, but the analyses required by the method are rather costly [62].Overall, isotopic analysis of plant and soil water could help better understand water extraction strategies of occurring plant species and the implication for the downward flux of water into the deep soil, but hydraulic redistribution of deep soil moisture and their potential effect on isotope composition, if any, must be accounted for [61].

Chloride Mass Balance
Chloride is a conservative anion and, as such, it neither leaches nor gets absorbed by soil particles and sediments [62][63][64][65].Chloride mass balance (CMB) is based on several assumptions as documented by Gaye and Edmunds [63] and Sibanda et al. [64].Some examples of these assumptions include: (a) Net change in storage of Cl − in the unsaturated zone is zero during integration time; (b) Cl − mass flux over time is constant; (c) there is no unaccounted Cl − input from external or groundwater sources such as weathering and dissolution of minerals; and (d) the flux of Cl − due to wet and dry deposition at the surface is equivalent to the flux beneath the root zone.
Using CMB, recharge is determined by the following equation: P is the average annual rainfall (mm year −1 ), Cl p is the average Cl input from all sources (mg L −1 ), Cl s is the average Cl concentration of pore water below root zone (mg/L), and G is the average annual recharge rate (mm year −1 ) [65,66].Cl s is calculated by dividing soil Cl − concentration (in mg/L) by gravimetric water content (in mm 3 /mm 3 ) [66].
CMB provides point-scale recharge estimates.Chloride mass balance approach was used by Acharya et al. [31], Wine et al. [67], and Moore et al. [68], among others to understand the impact of woody plant encroachment on deep drainage and recharge.Woodlands, in general, have more Cl − content in the pore water compared with grassland, indicating lower recharge [31,66].
CMB method is a low-cost and straightforward approach to provide reliable estimates of groundwater recharge.However, sampling many profiles and deeper soil samples are often labor-intensive and time-consuming.Additionally, it requires consistent long-term land use [53].CMB can underestimate recharge process under a system dominated by macropores and in heavy textured soil [65].Use of chloride is challenging in the latter case due largely to Cl − exclusions (also called anion exclusions), which could increase chloride concentration in the pore water [65].Despite associated challenges and limitations, CMB is a widely used method in arid and semiarid regions because these regions have higher Cl − concentration as evaporation of rainfall allows residual chloride to accumulate in the unsaturated zone [69].

Modeling Approach
Hydrological models (e.g., SWAT, HYDRUS, and RHESSys) are also used to estimate the effects of woody plant encroachment on deep drainage and recharge [31,67,70,71].In this review, we discuss these three models used in ecohydrological studies in the context of woody plant encroachment.

Soil and Water Assessment Tool (SWAT)
Soil and Water Assessment Tool (SWAT) (USDA Agricultural Research Service at the Blackland Research & Extension Center, Temple, TX, USA) is a frequently used hydrological model to estimate recharge of aquifers in hundreds of watersheds globally.SWAT is a spatially explicit semi-distributed model with an easy to use ArcGIS interface developed by Dr. Jeff Arnold in the 1990's [72].SWAT model divides the watershed into sub-basins, hydrological response units (HRU), and elevation bands to represent snow, soil, shallow aquifer, and deep aquifer storage, and the model simulates runoff, infiltration, and ET at temporal scales.Model inputs include daily maximum and minimum temperature, precipitation, wind speed, solar radiation, relative humidity, soil type and classes, land cover, and land use and management, among others.SWAT also provides a mechanism to partition the recharge between the shallow and deep aquifer and baseflow contribution to streams.In addition, SWAT accounts for the removal of water from the aquifer by deep-rooted plants and irrigation pumps.For additional details, see Gassman et al. [73].
SWAT model is relatively simple and robust with a wide user community.It integrates upland and channel processes, and required input datasets are usually readily available.However, many databases used with hydrological models lack plant-specific biophysical and hydrological parameters.A recent study calibrated these parameters for eastern redcedar woodland using SWAT model and paired experimental watersheds [70] and found that Eastern redcedar, a prolific woody plant in South-Central Great Plains, reduced both soil moisture and surface runoff.SWAT model is relatively flexible to integrate with other groundwater models such as MODFLOW (United States Geological Survey, Reston, VA, USA) for a more in-depth investigation of the interaction between the groundwater and surface water (https://swat.tamu.edu/).However, challenges exist in exploiting the model in complex terrain and extreme meteorological settings [74].

HYDRUS
HYDRUS (PC-PROGRESS, Prague, Czech Republic) is another model, which can be used to understand recharge and root zone soil moisture distribution [75].This is a finite element model that numerically solves Richards' equation to simulate water movement in variably saturated soils [76].Model input includes maximum and minimum temperature, humidity, radiation, wind, soil moisture, plant rooting depth, albedo, leaf area index, and soil hydraulic properties (e.g., saturated water content, residual water content, saturated hydraulic conductivity, bubbling pressure, and pore connectivity parameter), among others.HYDRUS uses the van Genuchten-Mualem hydraulic model and/or single porosity model to determine soil water retention and hydraulic conductivity.Feddes et al. [77] and S-shaped function [78] are used in the model to assess root water uptake.
HYDRUS-based studies indicate reduced groundwater recharge with woody plant encroachment.Wine et al. [67] used HYDRUS 1D model and reported that woody plant encroachment reduces deep drainage of water due largely to change in plant rooting depth, growing season duration, water stress, and surface compaction.In another HYDRUS based study, lower groundwater recharge rates were observed for eastern redcedar compared to grasslands [31].Scott et al. [75] estimated soil moisture recharge at different depth intervals under shrub and grass cover in semiarid rangelands.
While HYDRUS serves as an important tool to simulate drainage, there is a large uncertainty in plant rooting depth and its distribution at spatiotemporal scale.2.5.3.The Regional Hydro-Ecological Simulation System (RHESSys) The Regional Hydro-Ecological Simulation System (RHESSys) (Source code: GitHub Inc., San Francisco, CA, USA) model is a physically based dynamic model that simulates water, carbon, and nitrogen fluxes at spatial-temporal scale.The model is increasingly used to understand climate and land-use change impact on hydrology and ecosystem processes [79][80][81].For example, Tague and Peng [82] applied the RHESSys model in the California Sierras to determine snowmelt recharge.
Christensen et al. [79] used the model to access the sensitivity of transpiration rates to climatic variability in Upper Merced River watershed, California.
RHESSys model is structurally composed of different, process-based submodels-climate interpolation model (MT-CLIM, University of Montana, Missoula, MT, USA), vegetation and soil model (BIOME-BGC, University of Montana, Missoula, MT, USA), and vertical and lateral hydrologic flux model-and describes soil water in three different layers: Upper surface, saturated zone, and unsaturated zone [79,81,83,84].Six hydrologic parameters are calibrated in the model: (i) Saturated hydraulic conductivity at the surface; (ii) the decay of hydraulic conductivity with depth; (iii) pore size index; (iv) air entry soil water potential; (v) the fraction of recharge entering into deeper groundwater storage by bypassing the shallow subsurface flow system; and (vi) the drainage rate of deeper groundwater store.
Model inputs include temperature, precipitation, radiation, humidity, saturation vapor pressure, vegetation types, leaf area index, canopy height, soil types, stream-flow, atmospheric CO 2 , and nitrogen deposition, among others.The model is less complicated compared to other process-based models (e.g., MIKE-SHE, DHI, Hørsholm, Denmark).However, there are substantial uncertainties [85].Landscape heterogeneity, the complex spatial hierarchy of landscape, multiple parameters, and complex interface programs may cause difficulty in parameterization and execution of the RHESSys model [83].
Currently, studies on impacts of woody plants on deep drainage and groundwater recharge are few.This may be because measuring the entire rooting zone soil moisture is largely constrained by the installation of soil moisture probes in the deep soil profile and experimental costs are high.The recent increase in the use of geophysical techniques such as electrical resistivity imaging provides a non-intrusive approach to detect deep moisture profiles and drainage, as discussed next.

Geophysical Imaging
Geophysical methods are increasingly used to understand catchment hydrology and evaluate vegetation impacts on groundwater recharge [30, [86][87][88].Some geophysical methods include electrical resistivity imaging, electromagnetic induction, ground penetrating radar, and vertical electrical soundings.Electrical resistivity imaging (ERI) is a widely used geophysical technique in which surface electrodes are used for the acquisition of apparent resistivity data [89][90][91][92].Such resistivity data is collected from the same location at different time intervals [93], and when no subsurface chemical reactions exist, any change in resistivity generally indicates a change in water content [86,94].Resistivity ρ α (Ω-m) refers to the ratio between measured potential difference (∆V) and induced electric current (I) into the ground.Soil electrical conductivity σ (S m −1 ) is defined as the reciprocal of resistivity.
Subsurface electrical resistivity is affected by different factors such as solid constituents (grain size and distribution), soil moisture, porosity (void distribution and form), soil cavities and fractures, the degree of water saturation, concentration of dissolved salts, and soil temperature [92].There is a strong correlation between electrical resistivity and volumetric soil moisture content.With increasing water content, the electrical conduction increases and resistivity decreases.Details on principles, methods, and application of electrical resistivity survey are available from earlier reviews [86,92,94].
Electrical resistivity imaging has been applied to grassland, forest, and woodland biomes to monitor spatiotemporal changes in soil moisture and sap flow.Acharya et al. [30] used temporal electrical resistivity imaging to monitor deep water content in four contrasting vegetation types (tallgrass prairie, prairie heavily encroached by juniper, juniper woodland, and oak forest) in the South-Central Great Plains, USA.Results showed increased spatial-temporal variability in root zone conductivity under juniper-encroached catchment compared with grassland catchment, and two-layered moisture migration profiles under these two catchments (Figure 2).Recently, ERI was used to understand soil and xylem moisture content in ponderosa pine trees [95].Overall, geophysical methods can be used to characterize subsurface anisotropy, understand groundwater flow, and evaluate groundwater recharge.These data can be integrated with point hydrogeological data to provide a detailed view of subsurface flow conditions.After examining the different methods used to estimate recharge, we see that the choice of the method varies with factors related to climate and scales of time and space [39].For example, saturated zone techniques are more common in humid regions, while unsaturated zone techniques are commonly practiced in arid and semiarid regions [39].Our review also indicates that isotopic and electrical geophysical methods are increasingly being used to investigate the impact woody plants encroachment on soil and groundwater.Point measurements fail to provide adequate information on inter-relationships between vegetation and subsurface properties and processes.In the future, isotopic and electrical methods could improve our understanding of the interconnections and feedbacks associated with soil moisture, plant water usage, subsurface properties, and water storage and migration.While each method is associated with different degrees of uncertainty, a combination of different methods could complement each other [31].After examining the different methods used to estimate recharge, we see that the choice of the method varies with factors related to climate and scales of time and space [39].For example, saturated zone techniques are more common in humid regions, while unsaturated zone techniques are commonly practiced in arid and semiarid regions [39].Our review also indicates that isotopic and electrical geophysical methods are increasingly being used to investigate the impact woody plants encroachment on soil and groundwater.Point measurements fail to provide adequate information on inter-relationships between vegetation and subsurface properties and processes.In the future, isotopic and electrical methods could improve our understanding of the interconnections and feedbacks associated with soil moisture, plant water usage, subsurface properties, and water storage and migration.While each method is associated with different degrees of uncertainty, a combination of different methods could complement each other [31].

Mechanisms by Which Woody Plant Encroachment Affects Deep Drainage and Recharge
With woody encroachment, the vertical movement of water in deep soil matrix is interfered by the change in rooting pattern or depth, which results from the change in plant functional type.An improved understanding of vegetation impact on ecohydrological properties and water balance components is necessary to ensure sustainable ecosystem functions and services and to develop land-based mitigation and adaptation strategies under global environmental change.For example, understanding the impact of woody plant encroachment on streamflow and groundwater recharge is essential to proper land management to promote increased water yield in water-limited ecosystems [46,103,104].

Water Use and Evapotranspiration
Woody plant encroachment affects individual component of water balance differently and the effect varies with climate, vegetation, and edaphic factors [9,105].A substantial shift from T to more ET is likely to occur in semiarid and subhumid regions as the ratio of bare ground to vegetated ground decreases with encroachment [9].An Oklahoma study looking at the effect of vegetation on evapotranspiration in tallgrass prairie, compared the one-year ET of a denuded tall grass prairie plot with an undisturbed prairie plot and found that ET was lower in the non-vegetated site (547 mm year −1 ) compared to 728 mm year −1 in the actively growing prairie [106].In the South-Central USA with arid and subhumid climate, ET is mostly confined by the amount of soil moisture available for latent heat to transform liquid water into water vapor.There is a tight coupling of vegetation and water fluxes, and the conversion of herbaceous vegetation to woody life forms is associated with a substantial increase in ET [9,107].In Western USA, Scott et al. [108] reported a higher mean annual ET of 449 mm year −1 under mesquite woodland, compared with 279 mm year −1 under shrubland and 259 mm year −1 under grassland in gravelly sandy loam soil of Arizona, USA.
Drier climates may show the non-linear response of ET to increase in woody plant cover in part due to the nonlinear relationship between rainfall amount and canopy interception loss [109], but also because trees are deep-rooted and can access deep water.This response was explained by Walter in 1939.He purposed a two-layer hypothesis for tree-grass niche separation in dry savannas, which assumed that trees could access water from deeper layers, whereas grasses access water from topsoil [110].Woody plants have higher air turbulence in the canopy boundary layer, which could increase potential ET.Additionally, many woody plants, such as the Eastern redcedar, are evergreen and consume water throughout the year.Assuming no change in soil water storage for a given water year, an increase in ET translates into a reduced potential for groundwater recharge.Overall, the literature reviewed herein suggests that ET increase and groundwater recharge decrease with woody plant encroachment.

Infiltration
Change in plant functional type, due to woody encroachment, alters transpiration and interception [32,109] and soil physical properties such as organic matter and soil aggregates, which in turn leads to changes in soil infiltration rate, soil water storage [104], and subsurface flow pathways [111]; all of which affect streamflow and groundwater recharge [70,103].Eldridge et al. [112] measured lower steady state infiltration and sorptivity under shrubland compared to grassland and observed that shrub interspaces, not the increase in total shrub cover, were the major driver of lower infiltration following encroachment.The infiltration rates under shrubs (e.g., 48.2 mm h −1 ) were greater than at the interspaces (e.g., 17.0 mm h −1 ).Wine et al. [113] observed lower sorptivity and hydraulic conductivity under juniper trees compared with grasses under dry condition.Recorded low infiltration rates are usually associated with increasing cattle trampling and compaction for the interspace and development of hydrophobic litter and soil.However, infiltration capacity may be greatly enhanced once the woody canopy is closed and cattle grazing is reduced or excluded.Magliano et al. [114] observed higher infiltration rate (139.9 vs. 71.6 mm h −1 ) under woodlands compared to pastures in dry Chaco rangelands, Argentina.Zou et al. [104] observed three-fold higher steady state infiltration rates under juniper canopy compared to grassland.Such effects may occur due to "island of fertility", extensive root distribution, root activities, earthworm diversity, and macropores beneath the canopy.Increase in soil infiltration capacity assists conservation of water in the ecosystem by reducing overland flow from high-intensity storm events and may improve soil moisture recharge to deep soil layer.Qiao et al. [111] reported that, compared to grasses, woody plants reduced surface flow and subsurface flow and shifted runoff from saturation excess overland flow to infiltration excess overland flow.Overall, the impacts of woody plants on infiltration are mixed.However, increased infiltration in the soils underneath the canopy and around the canopy coverage area is likely.Yet, it remains fundamentally unexplored how climate, vegetation types, vegetation canopy cover, and grazing regime interact to affect soil infiltration capacity.

Plant Traits
Strong coupling between groundwater, vegetation, and atmospheric processes has long been recognized in the semiarid and sub-humid ecosystems where woodland encroachment has been wildly reported [9,115].Plants and their traits, such as leaf area, rooting structure, and phenology, determine the degree of precipitation loss to canopy interception and the depth and seasonality of root water uptake [10].A major effect of woody plant encroachment is to change the albedo and therefore the partitioning of energy input in grassland.Grasslands usually have a low canopy surface roughness and consequently high albedo.In contrast, encroaching and infilling of shrubs and trees into the prairie is characterized by a mosaic structure, and tends to generate higher surface roughness.Encroaching trees produce a shading effect both on adjacent trees and grasses, and this shading effect is thought to be an important controlling factor of energy dynamics in these systems.Additionally, plant phenology will likely play a proactive role in energy and soil moisture dynamics among different vegetation types.For example, drier soils were reported under juniper in Nebraska due to a more extended growing season and higher interception compared to grasslands [116].Further, access and extraction of groundwater are enhanced by woody encroachments.The deeper and more extensive rooting architecture of woody plants allows them to have a more robust water uptake strategy to explore both shallow and deeper soil moisture sources to meet the atmospheric demand [56].Woody encroachments also form preferential flow paths via continuous root channels and enhanced organic matter and biological activity [117], thus increasing downward water fluxes.While more emphasis has been placed on woody plant encroachment into grassland or savanna type vegetation, woody understory encroachment into woodlands or forests can alter the physiological state of the trees and modify overstory functioning [13], and therefore groundwater recharge.

Canopy and Litter Interception
Canopy and litter interception affect local recharge of groundwater by modifying the net precipitation and thus the net water input to the soil.Studies on interception loss to woody plants include [32,109,[118][119][120][121], among others.One such study found that woody plants have higher canopy storage capacity for rainfall compared to grassland [109].Under the hot and dry climate in the South-Central Great Plains, over 35% of the bulk rainfall can be intercepted by the closed canopy of junipers on an annual basis [71,119].However, 5-10% of bulk rainfall may be redistributed to the base of the juniper trees from stemflow and thus contributes to net water input [107,118].
The litter layer plays an interactive role in water flux at the soil and atmosphere boundary.A thick layer of litter will intercept a part of throughfall as a net loss to precipitation input.However, the litter layer also serves as a cover on soil, slowing or preventing soil evaporation [122] and this effect can be significant in an ecosystem with substantial bare soil and intercanopy patches [122].The limited number of studies on litter interception makes drawing conclusions difficult.Therefore the quantitative role that litter plays in the water budget and soil moisture replenishment in many ecosystems remains largely unknown.Further work and development of practical and reliable approaches to quantify the water absorption and release in litter under field conditions will help in our understanding and estimation of litter interception [32].

Soil Water Repellency
Soil water repellency, also referred to as hydrophobicity, is caused by non-polar coatings of hydrophobic organic compounds on soil and by hydrophobic particulate organic matter [123].While water repellency was reported back in 1917, studies were limited, and research gained momentum only after the 1970's.Two approaches commonly used now to test repellency include water drop penetration time and molarity of ethanol droplet.The occurrence of repellency is largely associated with litter layer, secretion of hydrophobic compounds by plants, organic matter and humus, fungal hyphae, and microbial activities [123].Soil water repellency affects local recharge by re-routing the flow paths.Robinson et al. [124] observed soil water repellency under pinyon and juniper due largely to litter layer and fungal mats.This repellency occurred to depths of 20 cm under juniper trees and concentrated water underneath the canopy due to the presence of bypass flow.Wine et al. [113] also observed higher soil water repellency under the Eastern redcedar than under grasses.Soil water repellency may reduce deep percolation and local recharge due to a decrease in infiltration rates [125].Conversely, increased repellency can enhance the transport of water into deeper profiles due to the creation of uneven wetting fonts and preferential flow paths [123].Increased repellency can also prevent litter interception and therefore increase surface evaporation [124].Overall, studies indicate the mixed effect of soil water repellency on groundwater recharge.

Sandstone Bedrock
The substrate plays essential roles in routing percolated water.Sandstone bedrock, with a typical porosity ranging from 5 to 30% and effective porosity from 0.5 to 10%, is generally permeable and may act as a reservoir to hold deep percolated water in place or turn that water into lateral flow or interflow [30,31].Vertical and lateral propagation of soil moisture with the same sandstone bedrock differs between grasslands and sites heavily encroached by the Eastern redcedar [30].In grasslands, a high electrical conductivity layer forms after extended precipitation and water migrates laterally at soil bedrock interfaces, even though the sandstone is permeable (Figure 2).In contrast, the high electrical conductivity layer rarely forms along the soil-bedrock interfaces under woody plant encroachment.Woody plants attenuate the precipitation pulse in the rooting zone and slow or prevent soil water from leaking out of the rooting zone [107].Additionally, woody plants are capable of accessing and extracting water held in sandstone bedrock as the soil dries out [126].Therefore, the difference in plant functional types interacts with substrates to alter the subsurface flow and recharge potential.This difference may not prevail if the soil is shallow, dry and the soil moisture and root system interaction are reduced, such as in karst ecosystems [127].

Karst Ecosystems
Karst landscapes are highly fractured and often contain caves, sinkholes, and shallow, rocky soils.Recharge in karst ecosystems often occurs via conduit flow.For example, two geological formations are apparent in the Edwards Plateau: The Edwards formation (higher elevation, consists of thin, rocky soil underlain by fractured limestone bedrock) and the Glen Rose formation (lower elevation, consists of 0.1-1 m deep soil underlain by marl or dolostone) [128].Marl is relatively permeable with higher water holding capacity whereas the dolostones are hard and less permeable.Karst ecosystems are unique because cracks and fissures allow the preferential flow of water, while highly weathered marls in rock layers could retain substantial water.
Trees can access deep water through deeper roots and affect ecosystem function and services.This has long been observed in the drylands of Southwestern North America.While such processes generally hold true, woody plants, however, show very limited access to deep water in karst ecosystems [129].As discussed before, impacts of woody plants on hydrological processes can vary with climate, land use and land cover, and edaphic factors [9,105].Wilcox et al. [127] reported that change in shrub cover had little to no effect on streamflow in semi-arid karst rangelands and that recharge in that setting depends on discrete fractures.Moreover, the groundwater bodies are poorly connected in Karst region [127].Ochoa et al. [96] studied the effects of juniper woodlands in a semiarid rangeland in moderately shallow to deep soils, with fractured rock substrates.They observed that lower topographic relief and subsurface geology increased transient water storage and water release through shallow aquifers.Groundwater wells over fractured rock in their study site showed rapid rise and decline in groundwater levels with woody plant removal.Three major conclusions can be drawn from earlier studies: (i) Woody plant removal may increase baseflow or recharge in areas where springs were present or from areas where trees may access deep water; (ii) in karst systems, plants reduce recharge by reducing epikarst (subcutaneous zone between soil and bedrock) water storage, not by tapping into perched water tables [128]; and (iii) land degradation, not woody plant encroachment, reduces recharge in Edwards Plateau region of Texas [130].

Recharge as Affected by Soil Texture and Depth
The impact of woody plants on deep drainage and recharge across different locations, climate, and soil are summarized in Table 3. Soil texture and soil depth affect pore size distribution in soil, water-holding capacity, and water available for transpiration, all of which affect groundwater recharge.Fine textured (clayey) soils have higher water holding capacity than coarse-textured (sandy) soils due to more pore space.Clayey soils have more micropores and thus higher plant available water than coarse-textured soil.However, water may move slowly through clayey soils due to small grain size and greater surface area, which renders lower specific yield and drainage.Indeed, coarse-textured soils with higher hydraulic conductivity have higher groundwater recharge, whereas fine-textured soils may have negligible recharge [51].Kennett-Smith et al. [131] reported that groundwater recharge decreases as clay content increases from 0 to 25% in the top 2 m soil profile.Increasing depth of layers with low permeability will reduce the downward flux of water.Kim and Jackson [66] reported 50% more recharge in sandy soils compared to clayey soils.
Cracking clay soils have significant water holding capacity.In 1979, Richardson and coworkers studied the effect of brush control on Houston clay soil in Blackland Prairie [132].These soils are defined as fine montmorillonitic cracking clays.Specifically, they retain water during dry spells, but swell when wet, leading to relatively low infiltration rates when soils are under wet conditions.Honey mesquite was observed to extract water from a deeper depth (120-150 cm) compared with prairie grasses, and brush control of the mesquite increased runoff by 10%.While recharge was not estimated in their study, after the brush removal, rainwater recharged soil moisture down to 150 cm and this water eventually drained to deeper soils.
Shallow soils have roots of grasses and woody plants occupying the entire soil profile, which causes the soil to dry out rapidly.Consequently, there may not be a measurable difference in plant water use among plant functional types.In deep soils, woody plants can, however, access deep water that is beyond the reach of shallow roots (https://ok.gov/conservation/documents/).Deep clay or silty soils may have very limited potential for recharge [54,133].Schwinning [128] showed that recharge in deep soil is largely affected by the plant's ability to use shallow water.Deep soils provide higher buffering capacity for trees against rainfall variations, but in shallow soils, such climatic variations could concurrently affect both herbaceous and woody plants [128].In relatively deep soil such as in the South-Central Great Plains, soil depth under different plant functional types affect variations in soil moisture periodicities, with variation decreasing with increasing soil depth.High frequency periodicities in shallow soil occur less often under woody plant encroachment compared to grassland [107].Consequently, the impact of woody plants on groundwater recharge varies with soil type, texture, and depth [52,68,128,132,134].These findings suggest that soil texture and depth variably affect hydraulic conductivity, plant water extraction, and groundwater recharge.An improved understanding of their interactive effects on groundwater recharge is critical in forming and assessing land management strategies to curtail the proliferation of woody plant encroachment.

Does Woody Plant Removal Increase Groundwater Recharge?
Encroaching woody plants increase transpiration and interception rates leading to loss of available water in the system [137].For example, Saleh et al. [138] reported that brush control might reduce ET by up to 25%.Two important determinants of groundwater recharge as a response to woody plant removal include shallow effective rooting depth and plant-available water.Seyfried and Wilcox [139] reported that the effective rooting depth decreased from 200 to 140 cm post-fire treatment in semiarid rangelands of Southwest Idaho, which increased the deep drainage of water by six cm [139].
Table 3 shows that woody plant removal could have a mixed effect on groundwater recharge.For example, in Verde Valley area, Arizona, Wyatt et al. [140] observed a 2.8% increase in recharge when basal area reduction was used as a forest restoration treatment, but such effects decreased with regrowth of underbrush and a subsequent increase in ET.In the Carrizo-Wilcox aquifer area of Texas, woody plant removal increased deep moisture and therefore recharge by a modest amount [68].The effect of woody plant removal may vary with species and site.For example, the removal of juniper is likely to increase recharge in areas with deep soil and subsurface flow paths [103].Contrastingly, woody plant removal has a very little effect on recharge occurring via conduit flow in the Edwards Plateau region, Texas [100].Ilstedt et al. [101] proposed and tested an optimum plant cover theory in seasonally dry tropics, and reported that recharge increases with intermediate plant cover.Consequently, the effect of woody plant removal on groundwater recharge depends on the (i) amount of rainfall; (ii) woody plant cover; (iii) permeable or hardy parent material; and (iv) interception characteristics [103].More studies are needed to fully evaluate the effects of woody plant encroachment and brush management on groundwater recharge.

Climate Change, Woody Encroachment, and Groundwater Recharge
Anthropogenic emission of greenhouse gases and land use changes has caused the surface temperature to rise, with significant effects on natural, managed and human systems.Global mean surface temperature has increased by 1.0 • C since the 1850's, and is projected to reach 1.5 • C between 2030 and 2052 [141].Atmospheric CO 2 concentration has increased from pre-industrial levels of 280 ppm to 405 ppm and is projected to exceed 600 ppm by the end of this century.Climate change is likely to increase warm temperature extremes, frequency, intensity, and amount of rainfall, and the occurrence of extreme events like floods and droughts.Alterations in rainfall and temperature are likely to affect available water, flow regimes, and hydrologic budgets.Additionally, they could modify woody biomass, fire, herbivory, and vegetation structure and functions, all of which affect groundwater recharge.
Effects of climate change are highly uncertain, and plant response to climate change may vary with sites, vegetation, disturbance regime, and management.While total rainfall may decrease, rainfall intensity is expected to increase in drylands [142].Reduced rainfall in general decreases groundwater recharge.Deeper infiltration may occur under intense storms; thereby increasing water availability for woody plants [142].Climate-induced droughts could increase tree die-off via xylem cavitation, carbon starvation, and hydraulic failure [143,144].Additionally, temperature shifts may trigger biotic stress (e.g., bark beetles), and alter ecosystem productivity and water budgets [145].
An elevated CO 2 may benefit plants with C 3 photosynthetic pathways (e.g., eastern redcedar); however, effects are likely to be constrained by resources availability, stresses, and ecological interactions [24,146].To cite an example, the rise in CO 2 levels may increase water use efficiency, extend a plants growing season and shift vegetation physiognomy from herbaceous to woody dominance, with potential ramifications on recharge.The CO 2 enrichment hypothesis and woody plants encroachment have been discussed in several studies [1,24].Plant response to climate change will vary with alteration in water availability and changing water allocation [147].Indeed, with changing climate, a comprehensive understanding of the interrelationships between groundwater, surface water, vegetation and climate at multiple scales is necessary to manage water resources and to adapt to climatic changes.

Summary and the Way Forward in Woody Plant Encroachment Impact on Groundwater Recharge
Groundwater depletion is a global issue.High rates of groundwater withdrawal are likely to affect natural and managed ecosystems and alter climatic feedback and stream-flow response to rainfall [44,54,148].Biome-level shifts from grassland to woodland have important ecohydrological repercussions.As such, understanding the impact of vegetation and vegetation transition is pivotal to manage land cover for water supply.This review indicates that groundwater recharge is sensitive to vegetation transition and the impacts of woody plant encroachment on water budget are complex and still under-studied.Isotopes and geophysical methods were identified as powerful tools to understand deep drainage and recharge.Hydrological impact of woody plant encroachment is different between sandstone bedrocks and karst regions, and between deep soils and shallow soils.Experimental studies indicated a mixed effect of woody plant removal on groundwater recharge.However, positive effects are likely in deep soil and areas dominated by subsurface flow paths.A few areas that deserve close attention include: i.
It is important to quantify recharge under different woody plant species and environment at spatial and temporal scales.ii.
Litter interception of rainfall under woody plant encroachment is poorly studied.While litter interception affects deep drainage and groundwater recharge, only a few studies have quantified the interception storage capacity and interception loss.iii.
Root depth, size, shape, spread, and water uptake have not been extensively researched within the context of woody encroachment.Information on woody root systems is important to predict ecosystem functions (e.g., hydraulic lift, drainage, and water balance) and biosphere-atmosphere interactions [149].Use of stable isotopes can provide valuable information on rooting depths, plant water uptake, and the hydrologic linkage of transpiration and groundwater and/or surface and groundwater, among others.iv.
Very few studies have tested the effect of woody plant removal on groundwater recharge.Effects of brush control/woody plant removal vary with site and plant characteristics, and therefore removal should be focused in areas where positive effects are likely.Recently, the "alternative stable state theory" and "pyric herbivory" theories have been discussed to understand the mechanisms of such woody plant encroachment and to inform management solutions [22].An alternative stable state theory largely predicts ecosystem state transitions in savannas based on resilience and adaptability, whereas pyric herbivore theory emphasizes fire and grazing interactions to manage and restore grassland biomes.The effects of woody plant control on bypass flow, regional scale water quality and quantity, and regional climate also needs to be studied [135].v.
Vegetation mapping is envisioned as a proxy for groundwater recharge [10]; yet broader understanding and development of interrelationships between vegetation, hydraulic factors, and recharge continues to be an enigma.vi.
We reviewed major techniques to estimate recharge based on unsaturated and saturated zone data.While different methods can be used to complement recharge estimates, it is highly important to identify a cost-effective approach.vii.
Global climate change is likely to alter rainfall and temperature regimes, increase frequency and intensity of extreme events, and shift plant functional types, which could modify interception, infiltration, evapotranspiration, subsurface flow, groundwater recharge, and climatic feedbacks.The effects of climate change are largely uncertain and further research is warranted.
Overall, woody plant encroachment is likely to alter groundwater recharge, and impacts vary in magnitude with time and space.Having a thorough understanding and documentation of the impacts of woody plant encroachment in different settings of climate, vegetation, soil, geological formation, and time periods can help in the sustainable management of water resources in water-limited regions.

Figure 1 .
Figure 1.Number of studies related to the woody plant encroachment impacts on groundwater recharge retrieved from the Web of Science database (1900-2018) using different key words combinations.WP = woody plant; WPE = woody plant encroachment; WPI = woody plant invasion; WE = woody expansion; SE = shrub encroachment; WPR = woody plant removal; BM = brush management; GR = groundwater recharge; R = Recharge; DR = drainage; and DP = deep percolation.

Figure 1 .
Figure 1.Number of studies related to the woody plant encroachment impacts on groundwater recharge retrieved from the Web of Science database (1900-2018) using different key words combinations.WP = woody plant; WPE = woody plant encroachment; WPI = woody plant invasion; WE = woody expansion; SE = shrub encroachment; WPR = woody plant removal; BM = brush management; GR = groundwater recharge; R = Recharge; DR = drainage; and DP = deep percolation.

Figure 2 .
Figure 2. Time-lapse electrical resistivity (ER) image under grassland and grassland encroached by juniper (Modified from [30]).Images are transient images (background image minus subsequent ER dataset) showing % change in conductivity.A positive change indicates higher electrical conductivity (higher water content).Background data were taken in June 2014.Conductivity decreased at the lateral distance of 14 to 42 m in the top soil mantle (A-C), but increased by 5 to 50% below 1 m depth under grassland; indicating higher potential for deep drainage and recharge.Percent change in conductivity was relatively low under juniper encroachment (D-F) compared with grassland.For details, see Acharya et al. [30].

Figure 2 .
Figure 2. Time-lapse electrical resistivity (ER) image under grassland and grassland encroached by juniper (Modified from [30]).Images are transient images (background image minus subsequent ER dataset) showing % change in conductivity.A positive change indicates higher electrical conductivity (higher water content).Background data were taken in June 2014.Conductivity decreased at the lateral distance of 14 to 42 m in the top soil mantle (A-C), but increased by 5 to 50% below 1 m depth under grassland; indicating higher potential for deep drainage and recharge.Percent change in conductivity was relatively low under juniper encroachment (D-F) compared with grassland.For details, see Acharya et al. [30].

Table 2 .
Studies using different recharge estimation methods in the context of woody plant encroachment.

Table 2 .
Studies using different recharge estimation methods in the context of woody plant encroachment.

Table 3 .
Studies since 2012 reporting the impacts of woody plants on deep drainage and recharge across different location, climate and soil.