Quantifying Prairie and Forest Impacts on Soil Water Holding Capacity and Infiltration
The uses of native vegetation
The aim of this article is to encourage more use of native vegetation in our land-use practices. This encouragement takes the form of soil water retention and infiltration predictions for prairie vegetation (more than three years old) and forests (more than five years old) based on soil texture class and bulk density. These predictions are databased and describe the central tendencies of soil water properties in the upper soil column of native landscapes almost as well as established literature values do for cultivated landscapes. Native plants are the true “green” of green infrastructure and a stormwater best management practice (BMP) of largely untapped potential. The impact of native plants with extensive root systems on soil water storage and infiltration is significant. This article summarizes recent findings on the magnitude of these differences and describes how and why these differences occur.
For decades, the application of hydrologic science has mostly been in the service of altering the landscape. While natural or naturalized landscapes have not been ignored per se, their hydrology was more like a starting point for the landscapes we aimed to create. By devising our specialized landscapes, we make them easier to control, understand, and accept. Natural landscapes appear to defy quantification and seem haphazardly characterized by variety, chaos, and tightly linked webs of feedback loops.
There is a growing body of field and laboratory testing of soil water properties for natural and naturalized landscapes. New datasets on soil water properties are being compiled by researchers looking at the performance of crop optimization practices, forest and prairie carbon sequestration, domestic and international soil conservation, and flood control programs, as well as green infrastructure and low-impact development. However, the conversion of these data into tools that modelers, designers, regulators, and technology promoters can apply or reference on a daily basis is still in its infancy.
These “new” naturalized landscapes are, in a sense, rolling back the clock to a time before we tilled, drained, irrigated, and paved the land. The work we tend to rely on to specify soil water properties is almost entirely based on the properties of land already impacted by human uses. For example, the Green-Ampt equation parameters used frequently for SWMM modeling and cited in the Handbook of Hydrology (Maidment 1993) were originally published by Rawls et al. in 1982. Rawls analyzed the properties of 1,323 surface soils from 32 states in the US Department of Agriculture (USDA) database, and almost all the soil samples were from agricultural land uses. Rawls and later Saxton and Rawls (2006) specifically excluded from their analyses soils that either had low bulk densities or high organic matter contents, the very soil properties that tend to characterize native or naturalized landscapes.
Using the Saxton-Rawls transformations on the kind of data they excluded, we are connecting new data with established analyses to help quantify how native or naturalized landscapes function hydrologically. In particular, we want to help develop tools to estimate soil water properties from readily available physical parameters for natural and naturalized landscapes just as we have for cultivated landscapes.
Soil is the thin skin of the Earth and supports the biotic foundation of most of the higher life forms on this planet. The soil environment is marked by water, air, and nutrient scarcity and patchiness. Even when these resources are in abundance, they can be hard for organisms to access, because the soil matrix can tightly bind up water and nutrients. Roots of plants and associated microbes have coevolved with soils to play a significant role in soil formation processes through a wide range of physical, chemical, and biological processes (Hinsinger et al. 2009).
Soils are the largest reservoir of biodiversity on Earth. A single gram of soil may contain 107 to 1012 bacteria, 104 protozoa, 104 nematodes, and 5 to 25 kilometers of fungal hyphae, but given their microscopic size, the actual area they cover is 10-5 to 10-6% of the total soil surface area (Hisinger et al. 2009). Soil is like a desert where life is discretely distributed and found almost exclusively in the rhizosphere.
The rhizosphere is defined as the volume of soil around living roots, either in direct contact with the roots or within several millimeters of the root surface (Figure 1). The biological activity in this zone is in large part fueled by plant exudates, the chemical byproducts of photosynthesis that either “leak” or are “pushed” or sucked out of plant roots. Plant exudates are carbohydrates (including sugars) and proteins that fuel the growth of hungry bacteria and fungi, which in turn are fed upon by nematodes and protozoa. Studies have shown that plants can control the numbers and diversity of fungi and bacteria in the rhizosphere by the exudates they produce (Lowenfels and Lewis 2006).
The inorganic component of soil is a complex porous media created by weathering, including wind, rain, snow, sun, heat, cold, abrasion, acids produced by fungi, and ammonia and nitric acids produced by bacteria. Mosses and lichens attach to rocks and produce acids and chelating agents that dissolve rock in order to utilize its minerals. This slow dissolution creates fissures, water gets into the fissures, and freeze-thaw cycles widen the cracks. Large plants colonize the fissures and their roots widen them and further break apart the rock. As William Bryant Logan says, the Earth is “a monument to the different weathering rates of its constituent minerals” (Logan 1996).
Water always moves down gradient from higher to lower potential energy. In soils, water can move downward, driven by the force of gravity through pore spaces, or sideways and upwards, driven by capillary forces and adsorption—the attraction of water to the soil or to itself. Mass flow of water can also be driven against gravity by the decrease in hydrostatic pressure in the upper parts of the plant due to diffusion of water out of stomata in the leaves. This creates a gradient that helps pull water from the soil into the roots via osmosis.
Flow through soil pores follows a torturous pathway, with numerous constrictions or “necks” and occasional “dead ends.” It is difficult to describe flow on the microscopic level, and even more daunting considering that the rhizosphere is an actively growing and changing living space. We typically resort to a description of flow as the average of all microscopic flow paths over the total volume of soil. This has led to Darcy’s Law, in which groundwater flow is held to be proportional to the energy gradient, and the proportionality constant, K, is called hydraulic conductivity.
|Photo: Courtesy of Bill Pan, Washington State University
Figure 1. High-resolution image of canola root and root hairs
|Photo: Feeney et al.
Figure 2. Three-dimensional map of sandy loam bulk soil porosity (white/blue areas) of a) bulk soil (in brown), b) bulk soil with fungal activity, and c) soil around the rooting zone of Lolium perenne after 30 days of plant growth. Images created using X-ray microtomography of intact soil cores.
Hydraulic conductivity is considered a function both of the porous medium—that is, total porosity, distribution of pore sizes, and pore connectivity/tortuosity—as well as the fluid’s properties, including density and viscosity. The porous medium’s contribution to hydraulic conductivity is called the intrinsic permeability and can be estimated as a function of the material grain size. However, in the rhizosphere, the physical architecture of the soil is as much dictated by the structure of inorganic soil particle packing as it is by biological processes.
Plant root exudates and the organic compounds released and transformed by microbial activity include mucilages, amino acids, lignins, polysaccharides, proteins, cutins, chitins, melanins, suberins, and paraffinic macromolecules. These compounds, particularly the “sticky” ones—those with slower decomposition rates and fungal hyphae—act like glues binding individual mineral and organic particles together, helping create soil aggregates.
Root growth itself directly affects soil structure through changes in packing and density as well as creation of new porosity and connectivity of porosity. Using a combination of high-resolution X-rays and 3D-geostatistical analysis, Feeney et al. (2006) showed for the first time in three dimensions large and significant increases in porosity caused by root and microbial activity (Figure 2). Feeney, as well as other researchers, have shown not only that is there an increase in porosity due to biological activity, but also that the correlation of the pore space increases—that is, the pores change from a random distribution to a more ordered, correlated structure. This leads to an increase in local diffusion rates and higher resource allocation to the microsites where the microbial populations reside.
Well-vegetated soil is permeated by biological structures—live and dead roots, worm holes, insect burrows, mycorrhizal fungi, soil aggregates, and so on. This biologically related porosity contributes to the growth of macropore flow paths that can dominate the movement of water through soils. Some researchers think of soil in terms of two domains. The first domain is described by Darcy flow: a regular medium where flow is assumed to move through at an average rate. The second domain is macropores, each with their own specific flow and velocity fields.
Soil Water Retention
It has been shown that the pore size distribution and soil structure can be inferred from the soil water retention curve, also known as the soil moisture characteristic curve (Hillel 1982). The soil water retention curve is derived experimentally by applying varying degrees of suction (or vacuum) to a saturated soil column and determining the relative amount of water retained by the soil versus water released for each suction value.
The force with which the soil “holds onto” water is measured in terms of matric (soil matrix) potential, or matric suction. It can be thought of as the degree of suction needed to “pull” water from the soil and break the surface tension between the water and soil surfaces. By convention, this suction is typically expressed as a positive pressure. A matric potential of zero means the soil is completely saturated and water is freely available. As a soil dries out, it takes more suction to break the attraction between the soil and water. At varying thresholds for different soils and plants, the suction needed to break this bond exceeds the plant’s suction capacity. This is called the permanent wilting point (PWP), the point beyond which a plant cannot pull enough water from the soil to sustain growth.
At relatively low matric suctions, for example, between 0 and 1 bar (100 KPa), most of the water removed is from the large macropores. Water is retained in smaller pores mostly due to the capillary effect (which is a combination of surface tension in the water and its adhesion to pore walls) and mostly as a function of soil structure. At higher suctions, as smaller and smaller pores empty out, the water retained is due more to water adsorption and is influenced less by soil structure and more by texture and specific soil surface properties. Hillel (1982) notes that there is a strong correlation between a suction of 15 bar (1,500 KPa) (often taken to be the lower limit of moisture availability to plants, the PWP) and the surface area of a soil. At this suction there is only enough water for a thin coating over soil particle surfaces, but this is enough to sustain microbial life in the rhizosphere.
The importance of pore size distribution to water retention can be demonstrated by examining the difference in porosity and the water retention characteristics of sand and clay. The greater the clay content of a given soil, the greater the water retention at any particular suction. In a sandy soil, most of the pores are relatively large, and once these large pores are emptied very little water remains in the rest of the soil (Figure 3a). Clay typically has a higher total porosity and water content than sand (also shown in Figure 3a), but due to its very fine, poorly connected porosity, it requires more energy (suction) to pull water out of clays than out of sands.
Soil structure also affects the water retention curve, particularly in the low suction range. For instance, soil compaction decreases total porosity and decreases the volume of larger interaggregate pores. Compaction decreases both the saturated water content, the total amount of water a soil can hold, and the initial decrease of water at low suction values (Figure 3b). Compaction tends to crush or squeeze preferential flow paths, the structures that provide the soil’s highest water-conducting pathways.
Developing a soil water retention curve is a time-consuming test procedure. Pedotransfer functions (PTFs) are functional relationships that can be used to “transfer” or convert readily available soil properties, such as soil texture and organic matter content, into hard-to-measure properties such as soil water retention and infiltration. The Green-Ampt equation parameters of Rawls et al (1982) were derived from a PTF that is based on multiple regressions of clay, silt, sand, and organic matter fractions. Later, Saxton and Rawls (2006) refined these relationships and created a freeware soil water characteristics program for their Soil-Plant-Air Water (SPAW) field and pond hydrology model. SPAW provides bulk density, the soil water retention curve, and hydraulic conductivity from user-specified sand, silt, clay, and organic matter contents, along with minor corrections for compaction and salinity. Using their soil water retention relationships, they derived an expression for saturated hydraulic conductivity (Ksat) as a power function of moisture held at low suctions (Equation 1).
|Top: Figure 3a. Below: Figure 3b. Effect of a) texture
and b) compaction on soil water retention (hypothetical curves, adapted from Hillel 1982)
In SPAW, all the variables of this equation are derived from regressions of the particle size distribution and organic matter content. This equation depends primarily on the change in soil moisture between saturation and 3.3 bar (33 KPa). This is the difference in soil water content between saturation and field capacity, where field capacity is the water content of a soil after all water that can freely drain via gravity has drained away. The soil volumes first drained tend to be the macropores, interaggregate spaces, and preferential flow paths.
A literature search was conducted to find studies that compiled sand, silt, clay, organic carbon/organic matter contents, bulk density, saturated hydraulic conductivity (Ksat), and, where possible, soil water retention curves. We tried to find studies where the field testing was done as paired samples for different land uses for the same soil. Data were limited to the topsoil horizons (usually just the A-horizon). In most studies these parameters were evaluated at depths between 0 and 10 centimeters or 0 and 30 centimeters. Several dozen studies were compiled for use in this investigation. For the sake of space, most of the studies are not cited in the references at the end of this article. The entire list is available from the author upon request.
The first task of the data analysis was to compare the results of the Saxton-Rawls pedotransfer function with the data from the collected studies. The land uses covered in the studies included conventional agricultural fields, both tilled and active grazing land, suburban lawn, conservation reserve program (CRP) areas including grass (typically switchgrass) and agrofrestry buffers, and some examples of prairie remnants and second growth forest.
The result of the comparisons between the Saxton-Rawls predictions versus data for bulk density and Ksat are shown in Figures 4 and 5. With a few exceptions, Figure 4 clearly shows a strong relationship between predicted and observed bulk densities for tilled land and pasture. But the Saxton-Rawls pedotransfer function tends to underpredict bulk densities for lawns and to overpredict for CRP and natural/naturalized landscapes.
Figure 5 shows that the Saxton-Rawls pedotransfer function tends to predict the average Ksat for agricultural and suburban landscapes. There is spread around the average, but that spread tends to be distributed fairly well around the average. Field and watershed-scale modeling relies on a certain degree of homogenization and abstraction to deal with variation. Working with central tendencies to derive predictions that aim for descriptive averages has long been an acceptable standard in hydrologic modeling and analysis.
The Saxton-Rawls function tends to underpredict Ksat for CRP, prairie, and forest. The differences in Ksat between land uses are very strongly correlated with bulk density. Relatively small differences in bulk density, on the order of 10% to 25%, for the same soil tend to correlate with significant differences in Ksat. These differences result in, on average, an increase in hydraulic conductivity for natural/naturalized land that is roughly four to 10 times higher than for cultivated land and the Saxton-Rawls predictions. In some cases, particularly for prairie remnants, the measured infiltration rates are nearly 20 to 40 times higher than the measured rate for adjacent cultivated land and the Saxton-Rawls predicted rates.
Only a few studies of suburban lawn hydrology were found. These studies showed a wide variety of infiltration rates. Most of the lawns were as compacted as agricultural fields, some even more so, and this condition must be partly responsible for the lower infiltration rates. We would also contend that the intensive suburban lawn maintenance regime is also partly to blame. The more grass is cut, the more the shoots will utilize available carbohydrates at the expense of the roots. Keeping grass short keeps the roots short and at greater risk of succumbing to disease, pests, or climate extremes (DiPaola and Beard 1980). The typical maintenance regime of keeping grass on life support includes inorganic fertilizers and pesticides that can significantly reduce biological diversity in the rhizosphere. This regime results in a short rhizosphere (only a few inches deep), poor soil structure, and even greater susceptibility to compaction from mowing and use.
|Figure 4. Comparison of SPAW-predicted and data bulk densities for various land uses
|Figure 5. Data versus SPAW-predicted hydraulic conductivity for various land uses
In the course of hunting for soil water data, we found a study by Teepe et al. (2003) that specifically set out to rectify the German practice of using agricultural data to specify soil water properties for forest soils. In Germany, the Soil Survey Manual for forest soils is borrowed from the German Soil Survey (AG Brodenkunde 1994), which is primarily based on agricultural soils.
At the Institute of Soil Science and Forest Nutrition at the University of Gottingen, Teepe et al. developed water retention curves for 1,500 forest soil samples and added the results of 350 curves from the literature. From this data set, they developed water retention curves for German forest soils and related them to soil texture, bulk density, and carbon content. They developed their own pedotransfer functions and found that the best relationships were derived by segregating the data by both soil texture class and bulk density. Just as with US studies, they found that the bulk density and saturated water content were significantly lower and higher, respectively, for forest soils than for agricultural soils.
Teepe et al. developed soil moisture retention curves for each German soil texture class and divided up each class into four bulk density groups: < 1 gram per cubic centimeter (g/cc), between 1 g/cc and 1.25 g/cc, between 1.25 g/cc and 1.45 g/cc, and > 1.45 g/cc. With these divisions, they were able to achieve relationships that described most of the variation for each texture class–bulk density division. For all soils, the Teepe curves showed overall higher total porosity and a larger difference in water content between saturation and field capacity for forest soils than for cultivated soils.
We used the Teepe soil water retention curves by texture class–bulk density division to recalculate Ksat with the Rawls’ equation (Equation 1) for the same natural/naturalized soil data we used to test the original Saxton-Rawls function. The results are shown in Figure 6. The estimated values using the Teepe water retention curves match the range, variation (Saxton-Rawls r2=0.24; Teepe r2=0.43), and averages of the data better than the uncorrected Saxton-Rawls function. Because our Teepe estimates are developed by soil texture–bulk density divisions, and given that most of the soils in the collected studies are silt loams, the Teepe estimates fall mostly onto three large areas on the figure.
|Figure 6. Data versus SPAW and Teepe-corrected predictions of hydraulic conductivity of prairie, forest, and CRP
The clearest physical difference between cultivated soils and natural/naturalized soils is the overall porosity and the fraction of macropores. The parameter that highlights this difference is saturated water content (qs). Saturated water content is essentially equivalent to total soil porosity. The Teepe curves and other studies of CRP, prairie, and agroforestry tend to show a qs 10% to 40% higher than for cultivated soils. Higher porosity and bulk density are functionally related and the key to higher Ksat.
This difference in hydraulic conductivity is attributed not only to the increase in overall porosity, but also to the overall change in soil structure, including more aggregation and a more correlated pore structure. In a sense, the “self-design” feedback loop of the soil-plant-microbe system grows structure that more efficiently “delivers” resources to active growth zones; growth continues, accelerates, and feeds back to create a more effective pore space and so on.
One problem with the Teepe data is that the names and thresholds of various texture classes in the German soil classification system are slightly different than in the USDA system. We have tried as much as possible to use the USDA name—for example, “clay loam”—to refer to the corresponding clay loam in the German classification system. However, the entire raw dataset should be reanalyzed to develop a one-to-one correspondence between actual sand/silt/clay contents and USDA classes. For now, we have developed the best correspondence we could to summarize the soil hydraulic properties for the soil texture classes defined in the Teepe paper. We have summarized the results of using the Teepe soil water retention curves with the Saxton-Rawls (SPAW) equation (Equation 1) in Table 1. The parameters summarized in the table include saturated water content, qs, the Saxton-Rawls fitting parameters B and li, and Ksat. For reference, we have also included representative ranges of porosity and the average soil texture class Ksat originally estimated by Rawls (1982) from USDA data.
This article began as an exercise to see if it was possible to develop simple hydrologic modeling tools like the SWMM Green-Ampt parameter table for modeling natural or naturalized areas. We think this work has shown that with a reasonable characterization of soil texture class, land use, and bulk density, it is possible to predict average water retention characteristics and hydraulic conductivity for the upper soil column of these landscapes. The quality of the predictions for prairie/forest land using the Teepe curves and Saxton-Rawls functions are on the same order of the Saxton-Rawls predictions for cultivated land.
Frankly, we used the Teepe data because it provided a convenient bridge between cultivated land and native/naturalized land. The higher predictive value of the Teepe curves over the Saxton-Rawls function for all native/naturalized land uses was not necessarily expected. While the quality of the regression suggests that there are some strong similarities between forest and prairie soils, it will be helpful to segregate data from these land uses to develop stronger relationships between their associated soils and soil water properties.
In addition, the Teepe correction is based on individual relationships for discrete increments of soil texture classes and bulk density. The Saxton-Rawls relationships are based on continuous functions for all texture classes and bulk densities. It would be more convenient if a continuous function like the Saxton-Rawls equation could be found to characterize the soil water properties of native/naturalized landscapes, rather than incremental, discontinuous functions.
One other aspect of this work that needs more attention is the impact of compaction on porosity and infiltration. Clearly, land will be more porous and hold more water if foot and vehicle traffic stay off of it. However, the difference in compaction between land uses cannot explain all the variation between water retention and infiltration characteristics of the same soil. For instance, in a recent US Geological Survey (USGS) study of paired rain gardens in Madison, WI, two planted with prairie plants and two planted with grass, the prairie plant rain gardens significantly outperformed the grassed rain gardens in terms of water use and infiltration. In fact, as the rain garden vegetation matured over a period of five years, the impact of the prairie plants on garden soil water properties continued to improve, while the grassed garden performance pretty much peaked after two years. After five growing seasons, the roots of the prairie plants had reached nearly 5 feet in depth, and water moisture data showed that the plants were actively affecting soil water content 2.5 feet below the garden surface (Selbig and Balster 2010). All other conditions were held constant during the growing period, so any differences between gardens can very likely be attributed to the differences in vegetation.
This USGS study highlights another avenue of extension for this kind of research. For the sake of time, we kept our focus on the upper soil horizon. Moving down through the soil column, plants’ influence on soil water properties does diminish. But with deep-rooting species, there can be impacts at depths of a meter or more below the surface. As we extend this research we hope to correlate land use or ecosystem type with impacts on physical soil properties at various depths throughout the column.
We also cannot stress enough the need for practitioners to continue to collect more soil water data on natural and naturalized landscapes. In particular, it would be most useful to the development of future predictive tools if the data collected included the percentages of sand, silt, clay, and organic matter, as well as bulk density, saturated hydraulic conductivity, and soil water retention curves. As we begin to segregate land uses more finely, it will also be helpful to characterize the vegetation by species, including the cover percentages, the health of the vegetation, and the duration during which the ground has been covered in a particular vegetation type.
Part of the motivation of this article is to demonstrate that by fostering biological growth, by choosing plants with deep and extensive root systems, and by allowing the soil-plant-microbe ecosystems to function naturally, macropores and soil aggregation can penetrate deeper and deeper into the soil column. This process significantly increases the soil porosity, water storage, and infiltration commonly found in our cultivated landscapes. These soil-root-microbe aggregations lead to more robust growth and a more resilient ecosystem better capable of handling variations in climate, disease, and pests than cultivated landscapes. Beyond that, native ecosystems do not require intensive maintenance regimes, including water, nutrient, and pesticide applications to maintain them in a permanent state of stunted, vulnerable, ecological “stasis.”
We hope this article helps demonstrate that our technical understanding of native landscape hydrology is becoming less of an impediment and more an incentive for using native plants. We believe public and political nostalgia for antiquated notions of status and aesthetics is now the biggest roadblock to reaping the profound benefits of reclaiming native landscapes, the planet’s original BMP.
Author's Bio: Scott Dierks, P.E., is a senior water resource engineer with Cardno JFNew in Ann Arbor, MI.
HTMLOutput: Object reference not set to an instance of an object.
HTMLOutput: Object reference not set to an instance of an object.HTMLOutput: Object reference not set to an instance of an object.