| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Biological Systems Engineering, 200 Seitz Hall (0303), Virginia Tech, Blacksburg, VA 24061-0303
b Forestry, 228 Cheatham Hall (0324), Virginia Tech, Blacksburg, VA 24061-0303
* Corresponding author (tesswynn{at}vt.edu)
Received for publication August 5, 2003.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
= 0.01). While there were no significant differences in total RLD below 30 cm, forested sites had significantly greater concentrations of fine roots, as compared with herbaceous sites ( = 0.01). As research has shown that erosion resistance has a direct relationship with fine root density, forested vegetation may provide better protection against stream bank erosion.
Abbreviations: BSA, basal stem area • RAR, root area ratio • RLD, root length density • RVR, root volume ratio • SCV, shrub crown volume • TD, tree density
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
To provide clarity for the following discussions, the authors adopted the terminology proposed by Lawler et al. (1997). Specifically, the term "erosion" is used to describe the detachment, entrainment, and removal of individual soil particles or aggregates by hydraulic forces. The phrase "bank failure" denotes the physical collapse of all or part of the stream banks as a result of geotechnical instabilities. Bank erosion and bank failure commonly work in concert to produce "bank retreat" or the net recession of the stream bank. Two additional terms, soil "erodibility" and "critical shear stress" describe, respectively, the ease with which soil is removed from the bank face and the hydraulic shear stress at which erosion is initiated. These parameters are used in the excess stress equation and are primarily dependent on soil properties (Hanson and Simon, 2001).
Many of the benefits associated with riparian vegetation are related to the root systems. Stream bank retreat typically results from erosion of the bank toe followed by collapse of the upper bank. Roots increase the strength of bank soils, making them more resistant to soil erosion and bank failures (Wu and McKinnell, 1976; Murgatroyd and Ternan, 1983; Abernethy and Rutherfurd, 2001; Mamo and Bubenzer, 2001). It is believed the root systems of woody and herbaceous plants physically bind bank soils in place, increasing the soil critical shear stress, 
c (Gray and Leiser, 1982; Coppin and Richards, 1990; Thorne et al., 1997). Additionally, root exudates may increase soil cohesion chemically (Amarasinghe, 1992; Thorne et al., 1997). Smith (1976) found that meadow grass and scrub vegetation increased the erosion resistance of stream banks by a factor as great as 20000. The erosion rate decreased linearly with increases in the percentage of root biomass. Kamyab (1991) and Dunaway et al. (1994) also found that erosion rate was inversely proportional to root length density and root volume, respectively.
There is considerable debate in the literature regarding the relative merits of herbaceous versus woody riparian vegetation in stream bank stabilization (Lyons et al., 2000; Simon and Collison, 2001). Herbaceous vegetation has a greater density of very fine roots, as compared with woody vegetation (Tufekcioglu et al., 1999). This high root density will probably produce greater c under herbaceous vegetation; however, bank reinforcement extends only to the rooting depth (Thorne, 1990). While trees have fewer fine roots, they also have a greater rooting depth (Gregory and Gurnell, 1988). Root density at the bank toe is more critical for bank stability, since hydraulic shear stress increases with stream depth. As a consequence, undercutting of grass banks is commonly observed (Davies-Colley, 1997). Millar and Quick (1998) determined that the mean c for forested stream banks was two to three times that of grass-covered banks.
The density and distribution of roots within a stream bank play an important role in stream bank erosion and stability (Allen et al., 1999; Millar, 2000); therefore, the assessment of riparian buffers and stream restoration designs requires knowledge of the impact of roots on bank stability and soil erosion. The overall goal of this research is to evaluate the effects of woody and herbaceous riparian vegetation on stream bank erosion by measuring the erodibility and critical shear strength of vegetated stream banks and relating those parameters to root density. As a first step in this analysis, the density and distribution of roots within stream banks must be quantified for different types of riparian vegetation. In this paper the distribution and density of roots along alluvial stream banks as a function of riparian vegetation type and density are discussed.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Root length density is the total length of all roots within a unit soil volume. The RLD provides an estimate of the total number of roots and is not skewed by the presence of large roots, as compared with root mass, root volume, or root area ratio (Böhm, 1979). If vegetated soils are viewed as a fiber-reinforced composite material, the RLD represents the number of fibers in the sample. For comparison with previous studies, the RVR was also calculated (Gray and MacDonald, 1989). Root volume ratio represents the total volume of roots per unit soil volume.
To measure root distribution with depth along the stream banks, 10 soil cores were taken at each site using a 7-cm-diameter, 15-cm-long soil corer. The adequacy of the sample size of 10 soil cores was determined in a preliminary study, as outlined by Crépin and Johnson (1993) and Elzinga et al. (1998). Cores were taken 30 cm back from the tops of banks at locations chosen using a stratified random scheme, although samples were constrained to distances greater than 30 cm from any tree bole, due to physical limitations. While Abernethy and Rutherfurd (2001) found root area ratio decreased with distance from isolated tree boles, McGinty (1976) found less horizontal variation in root biomass in a natural hardwood forest than in a pine plantation. He attributed this to the ability of the wide variety of plant types in a natural forest to fully utilize soil niches. Samples were taken in 15-cm increments to a depth of 105 cm, where possible. The sampling depth of 105 cm was selected because most roots are located in the top 1 m of the soil profile and because the bank exposure at most of the sites was approximately 1 m (Davidson et al., 1991; Shields and Gray, 1992; Simon and Collison, 2002). At three sites, restrictive gravel layers at the same elevation as the channel bed limited sampling depth. Samples for each depth increment were combined to produce one composite core per field site. A total of 1710 corer volumes were taken to produce 171 composite samples for the 25 cores (one composite core per site). Each composited sample was thoroughly mixed using a small cement mixer (BigCat Mixer Type B; Monarch Industries, Winnipeg, MB, Canada). Two subsamples, each representing one-tenth of the total sample weight, or one soil corer volume, were taken. One subsample was used for root measurements and the second subsample was used for soil particle size analysis. Soil samples were stored in a walk-in cooler (4°C) until they could be processed.
Roots were removed from each subsample by hand, washed over a no. 30 (0.5 mm) sieve to remove all soil, and stored in a refrigerator (4°C) until analysis. Dead roots, identified based on root color, flexibility, and strength, were removed from the samples (Böhm, 1979). Root length and root volume were assessed for each of five size classes (Abernethy and Rutherfurd, 2001). These classes are very fine roots (<0.5 mm in diameter), fine roots (0.5 to 2 mm in diameter), small roots (2 to 5 mm in diameter), medium roots (5 to 10 mm in diameter), and large roots (10 to 20 mm in diameter) (Böhm, 1979). Roots larger than 20 mm in diameter were disregarded because they contribute little to bank stability and are difficult to sample with a hand corer (Coppin and Richards, 1990). Root length and root volume were measured for each diameter class using a Régent Instruments STD 1600+ scanner and WinRHIZOTM analysis software (Arsenault et al., 1995; Régent Instruments, Quebec, Canada).
Particle size analysis (PSA) was conducted on each distinct soil horizon in the stream banks. Each composite core was evaluated to determine textural changes along the core. The PSA subsamples were then combined for each increment in the soil horizon. The horizon composite was then thoroughly mixed and passed through a no. 10 sieve. Particle size analyses were conducted following methods outlined by the USDA Soil Survey to determine sand, silt, and clay fractions (USDA, 1996).
To estimate the amount and type of aboveground vegetation, groundcover, shrubs, and trees were measured using 1-, 25-, and 100-m2 nested quadrats, respectively (Hession et al., 2000). Three sets of nested quadrats were measured at each field site. The parameter measured for each vegetation type was chosen based on ease of measurement and the extent to which the measurement would probably reflect belowground biomass.
Groundcover was defined as all herbaceous vegetation and woody vegetation less than 1 m tall. Any groundcover falling within a 1-m3 volume was clipped to ground level and divided into woody vegetation, grass, and forbs (Bonham, 1989). These subsamples were then oven-dried at 60°C and weighed to determine dry biomass in kg/ha. Shrub crown volume (SCV, m3/ha) was measured by estimating the geometric shape of each shrub and then taking the appropriate measurements to calculate the volume (Bryant and Kothmann, 1979; Bonham, 1989). Tree basal stem area (BSA) was estimated by measuring the largest and smallest diameter of all trees at breast height of 1.4 m (Bonham, 1989; Davidson et al., 1991). Tree diameter was calculated as the geometric mean of the two measurements (Husch et al., 1982). Trees were distinguished from shrubs based on the stem diameter and the general size and shape of the plant. Tree density (TD, stems/ha) as well as the stand BSA (m2/ha) were calculated for each site (Van Miegroet et al., 1984). While tree crown volume may better indicate root biomass than BSA, crown volume is difficult to measure and the large measurement error inherent with this parameter would probably offset the theoretical gains in accuracy. Both trees and shrubs were identified to the genus level.
Using K-means cluster analysis, the sites were split into two categories, forested and herbaceous, based on aboveground vegetation measurements (Johnson and Wichern, 1992). The aboveground vegetation quantities, the RLD, and the RVR in each depth increment for the two buffer types were compared using the nonparametric Mann–Whitney test, which tests for differences in the sample median (Neave and Worthington, 1988). Changes in RLD over the two-month sampling period were investigated by conducting Theil–Sen nonparametric linear regressions of RLD versus sampling date for the different buffer types, root diameter classes, and depth increments (Theil, 1950a, 1950b, 1950c; Sen, 1968; Hollander and Wolfe, 1973). Additionally, plots of RLD versus sample date were visually checked to detect any nonlinear temporal trends in RLD.
Multiple linear regression analysis was conducted to determine the influence of aboveground vegetation type and density, site management, and soil texture on RLD and to develop a relationship to predict the RLD in stream banks. If a relationship can be established between RLD and bank erosivity, an equation predicting RLD could ultimately be used in the design of riparian buffers for stream bank stability. To eliminate problems with multicolinearity, Kendall's tau was calculated for each aboveground vegetation and soil parameter (Hollander and Wolfe, 1973). Stepwise multiple linear regression was then conducted and the residuals of all significant regressions were visually evaluated for normality and homoscedasticity using normality and residual plots. For simple linear regressions where the residuals appeared non-Gaussian or heteroscedasticity was present, the regression relationships were checked using the nonparametric Theil–Sen regression technique; if the least-squares regression equation was similar to that produced by the Theil–Sen method, the least-squares regression was considered valid (Hollander and Wolfe, 1973).
|
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Based on the quantities and types of aboveground vegetation present, 11 sites were classified as herbaceous and 14 were classified as forested using K-means cluster analysis. While trees and shrubs were present at some herbaceous sites, they were generally scattered and did not form a full canopy. Measurements of groundcover biomass and shrub crown volume ranged over several orders of magnitude, reflecting both natural variability and errors inherent in vegetation measurement (Table 1). Because trees are generally simple to differentiate, measurements of BSA and tree density were less variable. Differences in medians between the two buffer types for each aboveground vegetation measurement were significant at = 0.05, except for forbs and woody groundcover. This is not unexpected as even mature forests have understory vegetation. Median grass and total groundcover biomass were significantly greater for the herbaceous sites at p = 0.0001.
|
Root Length Density
Roots were found at all depths at all sites, including below the water table. Total root-length densities (i.e., all roots <20 mm in diameter) varied from 16 cm/cm3 in the top 15 cm of an intensively managed pasture to 0.04 cm/cm3 at a depth of 1 m in a clay loam soil under a forested riparian buffer. This range of root densities is similar to those found under corn and soybean on a Plano silt loam in Wisconsin [0.2–24 cm/cm3 (Mamo and Bubenzer, 2001)], but as much as two orders of magnitude less than RLDs reported for riparian meadows in the Sierra Nevada [7–750 cm/cm3 (Kamyab, 1991); 26–4650 cm/cm3 (Kleinfelder, 1992)]. For both the herbaceous and forested buffers, the majority of the roots were less than 5 mm in diameter.
There was wide variability in RLD for both riparian buffer types, outliers were common, and the distributions of RLD were typically positively skewed (Fig. 1) . The largest range in RLD was for very fine roots under herbaceous vegetation at a depth of 0 to 15 cm (1.37–10.68 cm/cm3). Root length density under forest cover was less variable. For both buffer types, the variability generally decreased with increasing depth and root diameter. There were very few medium or large roots in either buffer type at any depth; RLD ranged from 0.00 cm/cm3 to a maximum of 0.05 cm/cm3 for medium roots under forested vegetation.
|
= 0.05). The statistically significant regressions indicate that errors due to seasonal changes in RLD over the course of the sampling period are at most 116%. Considering that total RLD in the study varied over three orders of magnitude and that only 5% of the tested relationships indicated any seasonal differences, it appears that errors due to sampling date are relatively insignificant. The location on the stream banks chosen for core sampling may have also influenced the study results. Soil cores were taken 30 cm from the edge of the top of the bank to minimize disturbance of the bank face for future measurements of soil erodibility. To determine if the RLD at 30 cm from the edge of the top of the bank was different from that at the bank face, two horizontal cores were taken at two forested and two herbaceous sites (eight cores total). Samples were taken in 15-cm increments to a distance of 105 cm into the bank, at depths of 30 and 100 cm from the top of the bank. Results of these investigations indicated there was a difference in the RLD between the bank face and 30 cm from the top of the bank for a site with dense vegetation growing on the bank face. Similar bank face conditions were present at 6 of the 25 sites. For these sites, the RLD at depths greater than 30 cm may be underestimated by as much as an order of magnitude. For this reason, the results of this study may be applicable only for nearly vertical stream banks with little vegetation on the bank face. Further work is needed to determine root density on densely vegetated, gently sloping banks.
As would be expected, roots were concentrated in the top of the bank profile (Fig. 2) . More than 55% of the RLD in the forested stream banks was located in the top 30 cm, compared with 75% for the herbaceous vegetation. Roots tend to concentrate in the upper soil horizons because these horizons have higher nutrient and oxygen concentrations and lower bulk densities (McGinty, 1976; Gray and Leiser, 1982; Coppin and Richards, 1990).
|
0.0074. At depths greater than 30 cm, the distribution of roots was fairly uniform for both buffer types and there were no significant differences in total root densities ( = 0.05; Fig. 1). In their study, Davidson et al. (1991) indicated that roots grew deeper in sandy soils as compared with clay soils. Shields and Gray (1992) also cited the low moisture content of the sandy levee soils as a reason for the deep root growth they observed. Similar conditions could have resulted in the deep herbaceous root growth observed in this study. During the summer of 2002, the mid-Atlantic U.S. states experienced a major drought. Precipitation was 80% of normal for the summer and 89% of normal for the previous three-year period (Virginia Department of Environmental Quality, 2002). Very dry soils and unusually low water table levels were noted during the field sampling. Combined with the friable loamy soils, these conditions could have promoted deep root growth in the study stream banks.
Greater differences between the buffer types can be seen by evaluating the distribution of roots by diameter class (Fig. 2). The herbaceous sites were dominated by very fine roots (diameter < 0.5 mm) at all depths and they had significantly greater very fine RLD than forested sites to a depth of 60 cm (p < 0.02). Below 60 cm there was no significant difference in very fine RLD between the vegetation types. For the remaining diameter classes, the forested RLDs were generally greater below 30 cm ( = 0.05; Table 2).
|
Root Volume Ratio
As stated previously, the root volume ratio was calculated for comparison with other research that evaluated root area ratio (RAR) or root biomass (Davidson et al., 1991; Abernethy and Rutherfurd, 2001; Simon and Collison, 2002). Median RVR for roots <20 mm in diameter ranged from 0.0009 at 90 to 105 cm to 0.0215 at 1 to 15 cm for herbaceous cover and from 0.0037 at 90 to 105 cm to 0.0179 at 15 to 30 cm for forested cover (Fig. 3)
. While similar RAR values were presented for woody vegetation by some researchers (Greenway et al., 1984; Gray and MacDonald, 1989; Shields and Gray, 1992), they are an order of magnitude greater than those measured by several other researchers (Wu, 1976; Dunaway et al., 1994; Abernethy and Rutherfurd, 2001; Simon and Collison, 2002). These differences in root measurements may be the result of an imperfect correlation between RVR and RAR and/or variations due to climate, soils, and vegetation type, age, and density.
|
= 0.05; Table 2). These results differ from those reported by Shields and Gray (1992), who found similar RARs for herbaceous and woody vegetation below 20 cm in sand levees in California. This difference may be the result of drier conditions and greater rooting depths in the sand levees, as compared with the loamy soils and wetter climate of the Appalachian Mountains. Summing down the bank face, the median total RVR for the woody stream banks was twice the RVR for the banks with herbaceous cover. Similar results were noted for root biomass by Davidson et al. (1991).
|
Unlike RLD, the distribution of RVR across the diameter classes was dominated by the fine and small-diameter roots. The RVR under both vegetation types was dominated by fine roots (0.5 mm < diameter < 2.0 mm), representing 33 and 53% of the total median RVR for the forested and herbaceous sites, respectively. Similar results were reported by Simon and Collison (2002) for herbaceous vegetation, although roots greater than 5 mm in diameter accounted for much of the woody RAR in their study. This may be because roots with diameters less than 1 mm are difficult to detect and darker woody roots that are similar in color to the upper soil horizons may be missed with the profile wall method. Comparing RVR for the two buffer types in each diameter class with depth, the herbaceous buffers had significantly greater very fine root volume (diameter < 0.5 mm) in the upper 60 cm, while the forested buffers had significantly greater volumes of fine and small roots below 30 cm (0.5 mm < diameter < 5 mm, = 0.05).
Regression Analysis
Analysis of Kendall's tau revealed that several aboveground vegetation parameters were correlated. Basal stem area was positively correlated with other measures of woody vegetation, including shrub crown volume ( = 0.46, approximate p = 0.0014) and tree density ( = 0.65, approximate p = 0.0000). In contrast, a dominance of trees was inversely related to measures of groundcover. The BSA was negatively correlated with both grass biomass and total groundcover biomass ( = –0.59, approximate p = 0.0000; = –0.51, approximate p = 0.0004, respectively). Similar, stronger correlations were found between grass and total groundcover biomass and tree density ( = –0.70, approximate p = 0.0000; = –0.64, approximate p = 0.0000, respectively). Similarly, SCV is positively correlated to TD ( = 0.43, approximate p = 0.0024) and negatively correlated to overall groundcover ( = –0.52, approximate p = 0.0003). These results are unsurprising since eastern U.S. forested riparian buffers are often composed of a tree canopy with a shrub and sapling understory. Additionally, grass and other groundcovers would not be present in large quantities under the shade of a mature forest (Coppin and Richards, 1990). There was also some positive correlation between total groundcover and forbs ( = 0.49, approximate p = 0.0007) and total groundcover and grass ( = 0.62, approximate p = 0.0000). This positive correlation shows the general dominance of grasses and forbs over woody groundcover at the sites.
Stepwise linear regressions were conducted for both RLD and RVR at the seven depth increments for the five diameter classes. Statistically significant regressions are presented in Tables 4 and 5 for RLD and RVR, respectively; the slopes were statistically significant at = 0.05. Coefficients of determination (r2) ranged from 0.132 for total RVR at depths of 45 to 60 cm to 0.579 for fine RVR at depths of 75 to 90 cm. McGinty (1976) evaluated the impact of 17 soil and vegetation parameters on root biomass in the top 30 cm of soil under a hardwood forest and found that no single parameter could attribute for more than 16% of the total variability in root biomass. While these results indicate aboveground vegetation types and densities are significant in explaining root density, other factors, such as soil bulk density, moisture content, and nutrients, may play a major role in the growth and distribution of roots in stream banks (Gray and Leiser, 1982; Coppin and Richards, 1990; Abernethy and Rutherfurd, 2000). McGinty (1976) believed the local soil and climatic conditions around each sample had the greatest influence on root biomass (diameter < 25 mm). These results suggest it may not be possible to predict root density based on aboveground vegetation type and density alone.
|
|
Implications for Stream Bank Stability
The results of this study have implications for the management of riparian areas for stream bank stabilization. As discussed in the introduction, soil erodibility is strongly affected by root density. In a laboratory study of meadow soil erodibility, Kamyab (1991) that found soil erodibility was most significantly influenced by the fine RLD. While herbaceous buffers had a much greater total RLD than forested buffers, the roots were largely composed of very fine roots concentrated in the upper 30 cm of the stream bank. Within a stream channel, hydraulic shear stress is a function of depth; the greatest shear stresses are applied to the bank toe. At depths greater than 30 cm, the forested sites had significantly greater fine and small RLD than herbaceous sites (p < 0.03). Additionally, overall root volume for diameters less than 20 mm was significantly greater below a 15-cm depth for the forested sites. This indicates that while forested stream banks have a lower overall root length density, the density and volume of fine and small roots is higher where the greatest hydraulic stresses are applied. Thus, for nearly vertical banks (those without significant vegetation growth on the bank face), woody vegetation may provide better protection against scour of the bank toe. Additionally, considering that previous research showed that root tensile strength (N/m2) decreases with increasing root diameter, and that bank failure typically occurs low in the bank profile, these results could indicate that woody vegetation also provides greater geotechnical reinforcement of stream banks (Greenway et al., 1984; Abernethy and Rutherfurd, 2001; Simon and Collison, 2002). For gently sloped banks, fine root density may be higher if there is herbaceous vegetation growing on the bank face, although fine RLD does not appear significantly greater in the top 30 cm at herbaceous sites, as compared with forested sites (p = 0.07).
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
At depths greater than 30 cm, the forested sites had significantly greater ( = 0.01) fine and small RLD than the herbaceous sites. This finding has significant implications for the use of vegetation in stream bank stabilization. Research by Kamyab (1991) indicated that soil erosion was strongly influenced by the quantity of fine roots present. Considering that the greatest hydraulic shear stress is applied at the toe of the stream bank, forested vegetation may provide better protection against scour of the stream bank toe for nearly vertical stream banks. Because bank failure caused by increases in bank height or bank angle is initiated by scour of the bank toe, woody vegetation may be the best choice for riparian vegetation where stream bank erosion and stability is a management concern.
Results of the regression analysis indicate the quantity of grass in the riparian buffer strongly influences the density of very fine roots in the top 30 cm of the stream bank. At depths greater than 30 cm, the quantity of roots with diameters greater than 0.5 mm is significantly influenced by the amount of woody vegetation present in the riparian buffer. These findings reinforce the conclusion that woody vegetation produces significantly greater root length and volume in stream banks below depths of 30 cm.
Further study on the effects of root density on stream bank erodibility is necessary. Given the large variability in root density found in this and other studies, future studies should include a large number of root samples taken at the bank face. Results based on a few samples or limited areas may be misleading and may not be applicable beyond the localized environment (McGinty, 1976). The results of this study also illustrate the differences resulting from measurements based on root length versus root volume or root area. Comparing Fig. 2 and 3, it is evident that root volume measurements are biased by larger roots. For studies where the density of roots in the soil is important, RLD will more accurately represent the overall fiber content than RVR.
In addition to research on bank erosion, more research is required to determine the total effect of vegetation on stream bank stability and stream morphology. The effects of vegetation on riparian hillslope hydrology, stream hydraulics, and soil freeze–thaw cycling need to be evaluated. An ongoing study by the authors will address the effects of vegetation on freeze–thaw cycling, soil desiccation, and stream bank erodibility.
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Related articles in JEQ:
This article has been cited by other articles:
|
|
 |
|
  R. W. McDowell and R. J. Wilcock Sources of Sediment and Phosphorus in Stream Flow of a Highly Productive Dairy Farmed Catchment J. Environ. Qual., March 1, 2007; 36(2): 540 - 548. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| The SCI Journals | Agronomy Journal | Crop Science | |||
| Journal of Natural Resources and Life Sciences Education |
Vadose Zone Journal | ||||
| Soil Science Society of America Journal | Journal of Plant Registrations | The Plant Genome | |||
