| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Dep. of Ecology, Evolution, and Marine Biology, Univ. of California, Santa Barbara, CA 93106
b Dep. of Geological Sciences, Univ. of California, Santa Barbara, CA 93106
* Corresponding author (fierer{at}lifesci.ucsb.edu)
Received for publication August 5, 2001.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTS AND DISCUSSIONCONCLUSIONREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTS AND DISCUSSIONCONCLUSIONREFERENCES |
|---|
The role of surface runoff in nutrient dynamics becomes particularly important in light of global climate change scenarios. Predictions from climate change models for continental North America include the "repackaging" of total annual rainfall into fewer, more intense rainfall events (Easterling, 1990; Houghton et al., 1990). These changes in rainfall patterns and intensities may result in greater amounts of overland flow and a concomitant increase in the quantities of C and N removed from hillslopes (Edwards and Owens, 1991). Over time, any increase in nutrient removal from soil may have long-term consequences for soil quality and ecosystem productivity (Pimentel and Kounang, 1998). Furthermore, an increase in hillslope nutrient losses may reduce the quality of regional stream waters (Crosson, 1985).
In the Central Coast region of California, the rate of nutrient transport by surface runoff may be strongly controlled by vegetation type. The hillslopes of the study area are generally vegetated by either coastal sage scrub or annual grassland communities, with little mixing of the vegetation types (Mooney, 1977). In recent years, land cover changes associated with these plant communities have become increasingly important. There has been considerable loss of sage scrub habitat due to the combined pressures of development, agriculture, and conversion to nonnative annual grasslands for grazing (Davis et al., 1994). Throughout California, the area of sage scrub habitat has been reduced to 10 to 15% of its former extent (Westman, 1981), but few studies have examined the potential impacts of this loss of sage scrub habitat on nutrient export from hillslopes.
Rainfall simulation experiments allow us to study the factors controlling the quantity and forms of nutrients lost from hillslopes via overland flow. With a rainfall simulator, we can apply rainfall with realistic raindrop sizes and velocities to small hillslope plots and quantify both the discharge from overland flow and the rates of sediment transported from the plot in the runoff. We can then estimate plot-level nutrient losses via overland flow by analyzing the nutrient contents of the runoff water and suspended sediments. By conducting a series of rainfall simulations with varied rainfall intensities on plots with different characteristics, we can begin to predict hillslope-scale nutrient losses under natural rainfall conditions.
Our primary objective was to determine the magnitude of solute and sediment C and N loss occurring by overland flow from coastal sage scrub and grassland hillslopes. We also looked at the short-term effects of cattle trampling on nutrient losses from sage scrub hillslopes. With plots of both vegetation types, we focused particular attention on hillslope sediment-bound nutrient losses, which are inherently difficult to measure (Lowrance and Williams, 1988) and often ignored in similar studies. Simulations were performed with and without raindrop impact to study the processes associated with nutrient transport by overland flow. The transfer of solutes from the soil to surface runoff were of specific interest because they are so poorly understood (Walton et al., 2000). The results from the rainfall simulations conducted on annual grassland plots were incorporated into a numerical model of sediment removal by overland flow, allowing us to scale-up from the plot level and investigate the controls on sediment-bound nutrient losses for entire hillslopes.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTS AND DISCUSSIONCONCLUSIONREFERENCES |
|---|
Rainfall simulations were conducted on a total of nine annual grassland plots and six sage scrub plots (Table 1) . Sage scrub habitats are dominated primarily by the perennials California sage (Artemisia californica Less.) and purple sage (Salvia leucophylla Greene). Annual grassland habitats are dominated by grasses such as brome grass (Bromus spp.) and Mediterranean barley (Hordeum murinum L.). The grassland plots, like most of the annual grasslands in the region, have been historically grazed by cattle. The coastal sage scrub plots were ungrazed. Hillslope angles in both sage scrub and grass plots ranged from 4 to 25 degrees (Table 1). Vegetation cover density on each plot was measured with a pin frame at 10-cm intervals.
|
Rainfall Simulations
We constructed a rainfall simulator capable of sprinkling a 6- x 2.5-m plot with realistic raindrop sizes and terminal velocities (Dunne et al., 1991; Gabet and Dunne, 2002). Water was obtained from a nearby deep well and transported to the site in a stainless steel tank. Applied rainfall intensities varied from 40 to 140 mm h-1 and simulated rain events were 20 to 60 min in duration. Surface runoff was collected in a trough on the downhill side of the plot, with discharge determined by timed volumetric sampling. A total of 49 rainfall simulations were conducted on grassland plots and 15 on sage scrub plots. These rainfall simulations were done during the dry season, July and August of 1999 and 2000. On each plot, the rainfall simulations were conducted consecutively within a one-week period. The infiltration capacities of the plots remained constant between rainfall simulations, regardless of antecedent soil moisture. Between rainfall simulations, vegetation cover was experimentally reduced on each plot by clipping vegetation and carefully removing ground litter. Two rainfall simulations were conducted on grassland plots during the wet season, February 2000, to determine if there were any seasonal effects on nutrient loss rates. No wet season rainfall simulations were conducted on sage scrub plots.
Flow Simulations
To investigate the importance of raindrop impact on the transfer of nutrients from the soil to the surface runoff as solutes, we performed nine flow simulations without rainfall on four grassland plots. Flow simulations were conducted by introducing water from the top of the plot through a perforated pipe installed at ground level. Discharges were adjusted to approximate discharges from the rainfall simulations.
Nutrient Analyses
The fluxes of C and N in sediments were determined by collecting 1-L runoff samples approximately every 3 to 5 min. These runoff samples were then filtered through a Whatman (Maidstone, UK) #1 paper filter and the sediment was weighed to quantify the sediment load suspended in each runoff sample. The sediments were analyzed for C and N content on a Fisons (Danvers, MA) NA1500 C/N analyzer. There were no measurable quantities of inorganic C in the sediment (as determined by acid digestion), so total C content was assumed to be equivalent to organic C content.
A second set of runoff samples (50 mL) was collected every 3 to 5 min for analysis of dissolved nutrients. These samples were kept on ice in the field and filtered with a Whatman #1 filter to remove particulates, and the flow-through was frozen for later analyses. The NH+4 and NO-3 concentrations in the runoff were determined using a Lachat (Milwaukee, WI) autoanalyzer. Ammonium was analyzed using the diffusion method (Lachat Method #31-107-06-5-A) and NO-3 was analyzed using Griess–Ilovsay reaction after Cd reduction (Lachat Method #12-107-04-1-B). To estimate dissolved organic C and N concentrations, the samples were digested in an autoclave using a persulfate digestion technique (A.P. Doyle, personal communication, 2000). The digested and undigested samples were then analyzed on a Lachat autoanalyzer for NO-3 and CO2-3. The differences in total dissolved C and N concentrations between the digested and undigested samples represent the concentrations of dissolved organic C and N. Generally, less than 5% of total dissolved C was inorganic C, so all of the C measured after digestion was assumed to be organic C. The concentrations of nutrients removed in runoff were obtained by subtracting the nutrient concentrations in the applied water from that in collected runoff.
Runoff and Erosion Modeling
The numerical model, described in detail in Gabet and Dunne (2002), calculates the rate of sediment detachment from raindrop impact. With the field data, the model can be applied to calculate sediment-bound nutrient loss as the product of sediment detachment rate and the percent C and percent N in the sediment.
Rain power is the time derivative of the kinetic energy of rainfall and incorporates rainfall, ground cover, and hillslope angle so that:
 | [1] |
= density of water (1000 kg m-3), i = rainfall intensity (m s-1), v = raindrop velocity (m s-1), Cv = fraction of plot area covered by vegetation, and 
= hillslope angle (degrees).
Sediment detachment rate can be expressed as a product of rain power and a dimensionless attenuation function that accounts for the dampening of raindrop impact by water on the surface (Gabet and Dunne, 2002):
 | [2] |
= sediment detachment rate (g m-2 s-1), 
and ß = empirically determined constants, A() = attenuation function, h = water depth (mm), and d = raindrop diameter (mm).
Values for and ß were determined with the results from the rainfall simulations and Eq. [2] was found to accurately predict rates of sediment detachment (Gabet and Dunne, 2002). This formulation for calculating detachment rates is coupled to a flow-routing algorithm to calculate flow depths for the attenuation function and can predict sediment loss for entire lengths of hillslopes (Gabet and Dunne, 2002).
Statistical Analyses
The concentrations of dissolved nutrients did not change during steady state runoff so the average values of the collected runoff for each simulation were used for the analyses. All statistical analyses were conducted using Systat 10 for Windows 2000 (SPSS, 2000). Loss rates of nutrients with different experimental treatments or conditions were compared using two sample t tests. The assumptions of equal variance, as required for the t tests, were validated using F tests.
|
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTS AND DISCUSSIONCONCLUSIONREFERENCES |
|---|
Despite applied rainfall intensities of 140 mm h-1, no overland flow was generated on undisturbed sage scrub plots. Even after the removal of all vegetation and surface litter from the plots, rainfall simulations failed to yield any surface runoff. This indicates that the infiltration capacities of the soils in the sage are sufficiently high to absorb natural rainfall, preventing surface runoff and associated nutrient losses. Although no wet season rainfall simulations were conducted on sage scrub plots, there is no evidence for surface runoff during even the most intense winter rainstorms.
Careful inspection of the mineral soil under coastal sage scrub vegetation reveals a laterally extensive and continuous 1- to 2-cm-thick biotic crust, composed primarily of fine roots, moss, and lichens. We hypothesize that the high porosity of this biotic crust layer is responsible for the high infiltration rates in coastal sage scrub soils. Biotic crusts of similar morphology are common in other semiarid sagebrush habitats (Johansen, 1993). West (1990) and Eldridge (1998) have shown that highly porous biotic crusts can dramatically reduce surface runoff volumes by increasing infiltration rates.
The conversion of sage to grasslands for cattle grazing is common in the region, so the effect of trampling on the biotic crust is critical to our understanding of how grazing may affect surface runoff generation and associated nutrient loss. To investigate this, we conducted rainfall simulations on two plots before and after disturbance to the surface soil by trampling. Due to the difficulties inherent in persuading cattle to walk around inside a small plot, we mimicked cattle trampling by affixing cow hooves to the feet of a 90-kg human (carrying an additional 30 kg of weight) who then walked within the confines of the plot for approximately 15 min. We were only able to simulate a maximum trampling pressure of 120 kg hoof-1, whereas, with an average cow weight of 600 kg, the maximum pressure would be 200 kg hoof-1.
The effect of the trampling was considerable. Infiltration capacities decreased from more than 140 mm h-1 to a maximum of 60 mm h-1. This dramatic change was most likely due to destruction of the biotic crust and compaction of the top layer of soil. This is supported by an increase in bulk densities of soil samples taken from the top 1 to 2 cm of the soil before and after trampling. Pretrampling samples had bulk densities of 0.62 g cm-3 (SE = 0.03) and posttrampling samples had bulk densities of 1.16 g cm-3 (SE = 0.23). A similar decrease in soil infiltration capacities after disturbance to biotic crusts has been observed elsewhere (Eldridge, 1998; Belnap, 1995). However, other studies have found no discernible effect of biotic crust cover on soil hydraulic properties (Eldridge et al., 1997; Williams et al., 1999). Regional differences in soil characteristics and crust composition make it difficult to develop a general theory relating biotic crust cover to soil hydrological properties.
Infiltration Capacities on Grassland Hillslopes
Infiltration capacities on the grassland plots were much lower than for sage scrub plots, generally ranging from 30 to 50 mm h-1 during the summer dry seasons when the rainfall simulations were conducted for this study. Infiltration capacities were measured on a few grassland plots during the winter and were much lower, ranging from 5 to 10 mm h-1. This substantial seasonal difference is probably due to the swelling of the smectitic clays during the wet winter months. Since approximately 10% of the recorded 1-h rainfall intensities are greater than 5 mm h-1 (Figueroa Mountain Ranger Station, National Oceanic and Atmospheric Administration [NOAA], 5 km from site), the generation of overland flow on these hillslopes during the winter is infrequent, but not rare.
Rainfall Intensity and Loss Rates of Dissolved Nutrients
On grassland plots there was no systematic variation in rates of dissolved nutrient losses with respect to hillslope vegetation cover or slope (r2 < 0.2 in all cases, for cover and slope versus NH+4, NO-3, organic N, and organic C losses). We expected a relationship between rainfall intensity and rates of dissolved nutrient losses since studies have shown that raindrop impact is important in mixing surface runoff with pore water (Ahuja, 1990; Ahuja and Lehman, 1983). Surprisingly, only NO-3 loss rates (mg N m-2 s-1) were influenced by rainfall intensity (Fig. 1) and they can be estimated for different rainfall intensities (i, mm h-1) with:
 | [3] |
|
|
Vegetation Type and Dissolved Nutrient Losses
The average loss rates for dissolved N and C for the two trampled sage plots and the nine grassland plots are shown in Fig. 3
. The trampled sage plots lost significantly more organic C (P = 0.02), but substantially less NO-3 (P = 0.001) than the grassland plots. Rainfall simulations were conducted immediately after trampling, so the measured organic C losses may represent an initial flush of organic C into runoff, not steady-state conditions that would develop after a season of rainstorms. Any long-term organic C removal by runoff from coastal sage scrub plots is likely to be of lower magnitude. Compared with the grassland plots, the plots under coastal sage scrub lost relatively little N as NO-3, possibly because nitrification rates in sage scrub soils are lower than in grassland soils (Dornelles and Schimel, 2000).
|
Sediment-Bound Nutrients
Loss rates of sediment during flow simulations were several orders of magnitude less than during rainfall simulations, indicating that raindrop impact is essential for soil particle detachment (Gabet and Dunne, 2002). For this reason, we only present sediment-bound nutrient loss data from the rainfall simulations.
There was little variability between rainfall simulations with respect to the percentages of C and N in the sediment collected, so values were averaged for all rainfall simulations according to vegetation type. The percentages of C and N in the trampled sage plots were 13.4 (SE = 1.25) and 0.9 (SE = 0.08), respectively. The percentages for the grassland plots were lower, 6.4 (SE = 0.07) and 0.7 (SE = 0.01) for C and N, respectively. The higher percentages of C and N in sediment from the trampled sage scrub plots are probably due to the large amounts of biotic crust and other organic detritus broken up by the trampling. As mentioned above, this probably represents an initial flush of organic matter and subsequent surface runoff events would transport sediment with lower organic matter concentrations.
The data presented above give us an estimate of the proportional losses of C and N forms from plots; by incorporating sediment loss results from the plot experiments into a numerical model we can predict rates of sediment-bound nutrient losses for an entire hillslope. The sediment loss data are only useful when incorporated into a model since the applied rainfall intensities were generally higher than natural rainfall intensities, resulting in unnaturally high sediment detachment rates. The model was only applied to annual grassland plots because overland flow does not appear to occur on undisturbed hillslopes vegetated by coastal sage scrub. The processes that determine the concentrations of dissolved load in the runoff are poorly understood, therefore, no equivalent model was used to predict rates of dissolved losses.
We did three sets of modeling experiments to: (i) investigate the effects of land-use strategies on nutrient transport, (ii) investigate the effect of climate on nutrient transport, and (iii) estimate the annual loss of sediment-bound C and N. Grazing or other land-use changes directly affect sediment and associated nutrient loss by controlling the density of vegetation cover (Cv in Eq. [1]). A series of model simulations was performed with a range of values for Cv, while holding all other parameters constant. Sediment-bound nutrient losses are also sensitive to rainfall intensity (i in Eq. [1]) so, to predict loss rates of nutrients under various rainfall intensities, a second series of model simulations was performed by varying rainfall intensity with all other parameters held constant. Finally, the model was run to simulate an entire year of rainstorms based on regional precipitation records (Figueroa Mountain Ranger Station). For all model runs, we assume that the percentages of C and N in the sediment are constant. Parameter values for the runs are listed in Table 2 .
|
|
|
Seasonal Considerations
In the study region, the majority of large rainstorms occur in the winter when soils are generally moist from previous rainstorms. The majority of the data used in this study were obtained from rainfall simulations conducted during the summer, on soils that had received no antecedent rainfall for at least two months. To determine if winter rainfall simulations would yield different estimates for hillslope nutrient losses compared with summer simulations, we conducted a smaller number of rainfall simulations in the winter on grassland plots identical to those described above.
The percentages of C and N in suspended sediment and the loss rates of sediment were not statistically different between the summer and winter rainfall simulations (P = 0.6 and 0.8 for C and N, respectively). Therefore, our estimates of sediment-bound nutrient losses are applicable to winter rainstorm events. In contrast, there was a seasonal effect on dissolved nutrient losses. The NH+4, NO-3, dissolved organic N, and dissolved organic C loss rates were approximately 50% lower (P < 0.05 in all cases) in the winter compared with the summer (data not shown). High dissolved organic C and N concentrations may be associated with the first rainfall events of the season (Fisher and Grimm, 1985), which would explain the large dissolved organic nutrient losses observed with the summer rainfall simulations. Extractable NH+4 and NO-3 levels in the grassland soils are generally 75% lower in the winter than in the summer (Fierer, unpublished data, 2001), a change reflected in the seasonal differences in the dissolved losses of NH+4 and NO-3 in runoff.
Therefore, by conducting the rainfall simulations during the dry season, we may have overestimated dissolved N and C losses during winter storms. However, our estimates of losses of dissolved nutrients by overland flow are applicable to the first rains of the season or summer thunderstorms. In winter conditions, sediment-bound nutrient losses would probably constitute a larger proportion of total nutrient losses. The dominance of sediment-bound nutrient losses over solute losses has been noted in other studies (Barisas et al., 1978; Lowrance and Williams, 1988).
|
CONCLUSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTS AND DISCUSSIONCONCLUSIONREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTS AND DISCUSSIONCONCLUSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  X.-Q. Liang, H. Li, Y.-X. Chen, M.-M. He, G.-M. Tian, and Z.-J. Zhang Nitrogen loss through lateral seepage in near-trench paddy fields. J. Environ. Qual., March 1, 2008; 37(2): 712 - 717. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Yihe Lu, Bojie Fu, Liding Chen, Guohua Liu, and Wei Wei Nutrient transport associated with water erosion: progress and prospect Progress in Physical Geography, December 1, 2007; 31(6): 607 - 620. [Abstract] [PDF]
|
|
|
|
|
|
Y. A. Wood, T. Meixner, P. J. Shouse, and E. B. Allen Altered Ecohydrologic Response Drives Native Shrub Loss under Conditions of Elevated Nitrogen Deposition J. Environ. Qual., January 3, 2006; 35(1): 76 - 92. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| The SCI Journals | Agronomy Journal | Crop Science | |||
| Vadose Zone Journal | Journal of Plant Registrations | ||||
| Journal of Natural Resources and Life Sciences Education |
Soil Science Society of America Journal |
