Forest and grass riparian buffers have been shown to be effective best management practices for controlling nonpoint source pollution. However, little research has been conducted on giant cane [Arundinaria gigantea (Walt. Muhl.)], a formerly common bamboo species, native to the lower midwestern and southeastern United States, and its ability to reduce nutrient loads to streams. From May 2002 through May 2003, orthophosphate or dissolved reactive phosphate (DRP) concentrations in ground water were measured at successive distances from the field edge through 12 m of riparian buffers of both giant cane and mixed hardwood forest along three streams draining agricultural land in the Cache River watershed in southern Illinois. Giant cane and mixed hardwood forest did not differ in their DRP sequestration abilities. Ground water DRP concentrations were significantly reduced (14 percent) in the first 1.5 m of the buffers, and there was an overall 28 percent reduction in DRP concentration by 12 m from the field edge. The relatively low DRP reductions compared to other studies could be attributed to high DRP input levels, narrow (12 m) buffer lengths, and/or mature (28 to 48 year old) riparian vegetation.

(KEY TERMS: agriculture; forest; nonpoint source pollution; riparian ecology; water quality; wells.)


Nonpoint source nutrient pollution and subsequent eutrophication is a leading cause of water quality impairment in the United States (USEPA, 1996). Since phosphorus is the most common limiting nutrient for primary productivity in freshwater ecosystems (Carpenter et al., 1998), minimizing its transport to streams is a critical goal for watershed management entities responsible for maintaining or improving surface water quality (Sharpley et al., 2000). The majority of phosphorus is transported to streams and lakes as particulate phosphorus in surface runoff, which must be converted to a dissolved inorganic form (orthophosphate or dissolved reactive phosphate, DRP) before it is available to primary producers (Vought et al, 1994; Reddy et al., 1999). However, recent investigations have discovered significant subsurface transport of dissolved phosphorus in ground water in intensive agricultural areas (Novak et al., 2000; Burkart et al., 2004).

Within the past two decades, protecting and/or restoring riparian buffers has become a common best management practice to combat nonpoint source nutrient pollution (Welsch, 1991). Extensive research has been conducted on the abilities of vegetated riparian buffers to sequester nitrogen, phosphorus, and sediment in surface runoff and nitrogen in ground water (Hill, 1996; Dosskey, 2001; Wigington et al., 2003; Puckett et al., 2004). Much less is known about the ability of riparian buffers to abate DRP in ground water (Osborne and Kovacic, 1993; Takatert et al., 1999; Novak et al., 2002). The two primary mechanisms responsible for DRP attenuation in riparian areas are soil sorption and assimilation by vegetation and microbes (Lyons et al., 1998).

Riparian areas in southern Illinois serve as habitat for a once widespread native species of bamboo, giant cane [Arundinaria gigantea (Walt.) Muhl.]. Giant cane has erect culms bearing evergreen foliage that can grow to heights of 6 to 10 meters (Platt and Brantley, 1997). Prior to European settlement, giant cane was a prevalent riparian and wetland species in the lower midwestern and southeastern United States (Thwaites, 1966; Platt and Brantley, 1997). In southern Illinois, giant cane occurs only in sporadic patches, primarily along riparian corridors (Schoonover and Williard, 2003). Giant cane communities are critical habitat for neotropical migratory songbirds such as Swainson's warbler (Limnothlypsis swainsonii), six species of butterflies, and several species of moths (Eddleman et al, 1980; Platt et al, 2001). Given the documented specific wildlife needs, there is significant interest in giant cane restoration among federal and state natural resource agencies.

Recently, the water quality benefits of giant cane have been investigated. Research has shown that this species can significantly reduce nitrogen, phosphorus, and sediment in surface runoff and nitrogen in ground water (Schoonover and Williard, 2003; Schoonover et al., 2005). The objective of this study was to determine ground water DRP attenuation capabilities of relatively narrow (12 m) giant cane and forest riparian buffers in southern Illinois, where stream phosphorus levels are relatively high (Table 1). It was hypothesized that ground water phosphate concentrations would be significantly reduced in both giant cane and forest riparian buffers.


Three sites (blocks) containing both giant cane and mixed hardwood riparian vegetation were selected along Big Creek, Cypress Creek, and the Cache River, all within the Cache River drainage basin in southern Illinois (Figure 1). The Cache River watershed contains wetland complexes that have received the internationally significant wetlands designation by the United Nations Education, Science, and Cultural Organization (UNESCO) (Mankowski et al., 1997). At each study site, the giant cane and forest buffers were adjacent to the same cultivated field, and all buffers were at least 12 m wide. All sites were similar with respect to elevation, slope, adjacent land use, and underlying geology (Table 1). Differences in the study sites included nutrient management of the adjacent crop fields, soil type, dominant forest species, and depth to the water table (Table 1).


A transect of monitoring wells was installed in each forest and cane buffer at the three sites at successive distances from the crop field margin (0, 1.5, 3, 6, 9, and 12 m), perpendicular to the stream channel. Monitoring wells were composed of 3.2 cm diameter polyvinyl chloride pipe with attached sand filtration sleeves and were installed to a depth of 4.3 to 5.2 m with a 7.5 cm mud auger. Ground water DRP (expressed as PO^sub 4^-P), chloride, and pH were measured from the six well locations in each buffer every two weeks from May 2002 through May 2003. A stream grab sample was also taken on every sampling date from the three adjacent streams. The sampling protocol for each well consisted of measuring the water table elevation with a stainless steel tape, emptying the well with a bottom filling bailer, allowing it to refill, sampling the resulting fresh ground water, transferring the sample to a clean labeled storage bottle, and transporting the sample on ice to the Forestry Laboratory of Watershed Research at Southern Illinois University at Carbondale. In the laboratory, samples were stored at 4°C until analysis within 48 hours. Prior to analysis, all samples were filtered with 0.45-µm nitrocellulose filter papers. Dissolved reactive phosphate concentrations were measured using a Hach DR/4000V spectrophotometer via the ascorbic acid method (Greenberg et al., 1992). Chloride concentrations were measured on a Dionex 4000i ion chromatograph (Dionex Corporation, Sunnyvale, California). Chloride was used as a conservative tracer to assess dilution effects by upwelling ground water or concentration effects by evapotranspiration (Lowrance, 1992; Devito et al., 2000). Ground water pH was measured with an Accumet AR20 pH/conductivity meter (Fisher Scientific, Pittsburgh, Pennsylvania).

Vegetation surveys of the study sites were conducted in May 2003. All trees within 6 m of the center ground water monitoring well in each forest buffer plot were inventoried for diameter at breast height (DBH), species, and location (distance to the center well). Three tree cores were taken from trees of different sizes within the 6 m radius plot to determine the mean age of each forested stand. The mean stand densities of the cane buffers were obtained by taking six random measurements of the number of cane stems per square meter along each monitoring well transect.


Mean monthly data were analyzed according to a randomized complete block design with repeated measures in space because distances could not be randomized within each transect. The mixed model procedure in SAS (SAS Institute, Inc., 2000) was used to test for fixed effects of distance from field edge, vegetation composition (treatments), and their interaction. Blocks were assigned as a random effect in the model. Annual means were compared using the least significant difference procedure in SAS at α = 0.05 (SAS Institute Inc., 2000).


Chloride concentrations were not significantly different among the sampling locations within the buffers, indicating that dilution or concentration of ground water was negligible (Figure 2). Therefore, any changes in DRP concentrations were attributed to biological or sorption processes within the riparian buffers.