Element losses from fields in conventional and conservation tillage in the Atlantic Coastal Plain, Georgia, United States

Results and Discussion

Annual rainfall and irrigation, surface runoff, and subsurface flow for the study period (2004 to 2008) have been previously described (Bosch et al. 2012, 2015). A brief summary is included here for interpretation of the element concentration and transport data (table 3). The annual total rainfall ranged from a low of 910 mm in 2007 to a high of 1,488 mm in 2005. The irrigation applied averaged 172 mm y⁻¹ and varied according to the amount of precipitation received during peak growing periods. The annual surface runoff was highest in 2005 and lowest in 2007 (table 3), accounting for 21% and 3% of the annual precipitation, respectively. Similarly, the annual lateral subsurface flow was highest in 2005 and lowest in 2007 (table 3), accounting for 25% and 9% of the annual precipitation, respectively. Over the course of the five-year study period, surface runoff was significantly greater from the CT compared to the ST plots (table 3). The average annual subsurface flow was significantly greater from the ST compared to the CT plots (table 3), due to increased infiltration as a consequence of reduced tillage (Bosch et al. 2012).

Calcium. No significant differences were observed in the surface runoff concentration of Ca between tillage treatments (table 4). On an annual basis, the concentration of Ca in surface runoff from the CT plots was highest in 2004 and decreased over the course of the study period. Elevated Ca concentrations in 2004 were likely due to the application of gypsum that year (table 1). However, elevated Ca concentrations were not observed in 2006 (table 4) despite applying two times the amount of Ca in 2006 than in 2004 (table 1). The low Ca concentrations observed in 2005 and 2007 may be due to the application of poultry litter in those years (table 1), which has been shown to reduce Ca runoff in Coastal Plain soils (Endale et al. 2014b). The reduction in Ca concentration in surface runoff may reflect its complexation with P from the poultry litter via the formation of less soluble calcium phosphates. In the lateral subsurface flow samples, the concentration of Ca was significantly higher in the CT compared to the ST plots for all years of the study period (p ≤ 0.05; table 5). Strip tillage can result in larger subsurface flow volumes compared to CT (table 3) (Bosch et al. 2012), and the lower Ca concentrations observed with this tillage practice may be caused by dilution. On an annual basis, the Ca concentration in the subsurface flow from the CT plots was highest at the end of the study period when water inputs were lowest (2007 and 2008; table 3). Calcium loads were higher in the subsurface flow compared to surface runoff for all years and regardless of tillage treatment (figure 2a). The surface runoff loads were significantly higher with CT compared to ST practices only in 2006, corresponding to greater runoff volumes with the CT practice (table 3) (Bosch et al. 2012). Elevated total (surface and subsurface) Ca loads were observed in 2005 (figure 2a), which was the year of highest Ca additions (table 1) and also the wettest year of the study period (table 3). The total Ca loss over the five-year study period was 111 and 122 kg ha⁻¹ with CT and ST management, respectively. These losses represent 8.7% and 9.6% of the Ca applied in the form of gypsum over the five-year study period (1,273.9 kg ha⁻¹; table 1) with CT and ST management, respectively.

Iron. No significant differences were observed in the surface runoff and subsurface flow concentrations of Fe between CT and ST treatments (tables 4 and 5). On an annual basis, the concentration of Fe in the surface runoff was significantly higher in 2005 compared to other years for the CT treatment only (table 4). Poultry litter, which was applied in 2005 (table 1), can contain fairly high concentrations of Fe (133 to 595 mg kg⁻¹ soil) (Jackson et al. 2003) and has been shown to increase Fe concentration in runoff from agricultural fields (Moore et al. 1998). However, elevated Fe concentrations were not observed in 2007 (table 4) when poultry litter was also applied (table 1). In the subsurface flow from the CT plots, lower Fe concentrations were observed in 2006 compared to other years (table 5), possibly due to the lack of poultry litter and fertilizer applications that year (table 1). Iron loads were greater from surface runoff compared to subsurface flow for all years and treatments (figure 2b). Elevated sediment loads associated with high runoff volumes have been observed with CT (Endale et al. 2014a). Surface soils at this site can contain up to 31 kg ha⁻¹ of Fe (unpublished data), which may suggest an association between elevated sediment and Fe runoff loads. Surface runoff Fe loads were significantly greater with CT compared to ST management in 2005 and 2006 (figure 2b) when elevated runoff volumes were also observed (table 3) (Bosch et al. 2012). Elevated total Fe loads were observed in 2006, regardless of treatment (figure 2b). The total Fe lost throughout the five-year study period was 38 and 9 kg ha⁻¹ with CT and ST management, respectively. The higher total amount of Fe lost in surface runoff compared to subsurface flow may be linked to greater sediment loads observed with this tillage practice (Endale et al. 2014a) and our use of unfiltered samples for the ICP-OES analyses.

Potassium. The concentration of K in the surface runoff from the ST plots was significantly higher than the surface runoff from the CT plots only in 2004 (table 4), likely resulting from lower surface volumes with the ST practice (table 3) (Bosch et al. 2012). On an annual basis, the concentration of K in the surface runoff was lowest in 2006, regardless of tillage treatment. In the subsurface flow, significant treatment differences in the concentration of K were observed in 2005 and 2006, with higher values from the ST plots in both years (table 5). Elevated K concentrations with ST likely correspond to higher subsurface flow reported with this tillage practice in these two years (Bosch et al. 2012). The concentration of K in the subsurface flow was not different between years with CT but was significantly higher in 2005 compared to other years with ST (table 5). The poultry litter applied in 2005 contained K (91.5 kg ha⁻¹; table 1), which may have resulted in elevated K concentrations in the subsurface flow in that year. However, elevated K concentrations were not observed in 2007 when K was also added in poultry litter (101.9 kg ha⁻¹; table 1). In 2004, elevated K loads occurred in the surface runoff, regardless of tillage treatment (figure 2c). Elevated dissolved K loads have also been observed in 2004 and 2005 at the Gibbs Research Farm and were attributed to large runoff events in those years (Bosch et al. 2015). The total K lost over the five-year study period was 79 and 116 kg ha⁻¹ with CT and ST management, respectively. This K loss represents 40.8% (CT) and 60% (ST) of the total K applied over the five-year study period (193.4 kg ha⁻¹; table 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Annual averages of surface runoff and subsurface flow loads (kg ha⁻¹) of (a) calcium, (b) iron, (c) potassium, (d) magnesium, (e) manganese, (f) sodium, (g) phosphorus, (h) sulfur, (i) silicon, and (j) zinc with conventional (CT) and strip tillage (ST) management. Significant differences (p ≤ 0.05) in surface runoff loads between tillage treatments are indicated by an asterisk (*).

Magnesium. No significant differences were observed in the surface runoff concentration of Mg between CT and ST treatments (table 4). Furthermore, the concentration of Mg in the surface runoff did not change annually, regardless of tillage treatment. In the subsurface flow samples, the concentration of Mg was higher in the CT compared to the ST plots for all years (table 5). As in the case of Ca, the lower Mg concentrations observed with ST may be caused by higher subsurface flow volumes with this tillage practice (table 3; Bosch et al. 2012) and subsequent dilution of the Mg concentration. On an annual basis, elevated Mg concentrations in the subsurface flow were observed in 2008 with CT and in 2006 with ST management. Gypsum, which can also contain Mg (Endale et al. 2014b), was applied in both 2006 and 2008 (table 1), likely resulting in elevated Mg concentrations in the subsurface flow in those years. Magnesium loads were always higher in the subsurface flow compared to the surface runoff samples, regardless of treatment and year (figure 2d). In 2006 the Mg load was significantly higher in the surface runoff from the CT compared to the ST plots. Elevated total Mg loads were observed in 2005, associated with higher water losses observed in that year (Bosch et al. 2012). Over the course of the five-year study period, the total Mg lost was 35 and 46 kg ha⁻¹ with CT and ST management, respectively.

Manganese. The concentration of Mn in the surface runoff was not affected by tillage treatments (table 4). On an annual basis, elevated Mn concentrations in the surface runoff were observed in 2005 under CT management only (table 4). High concentrations of Mn may be due to the application of poultry litter that year (table 1). Poultry litter can contain Mn (Jackson et al. 2003; Moore et al. 1995; Sims and Wolf 1994) and has been shown to increase the concentration of this element in runoff from agricultural fields (Endale et al. 2014b; Moore et al. 1998). However, and as in the case of Fe, elevated Mn concentrations were not observed in 2007 (table 4) when poultry litter was also applied (table 1). The concentration of Mn in the subsurface flow was higher under ST compared to CT for all years of the study period (table 5). Under CT management lower Mn concentrations were observed in 2008, while for ST, the values decreased in 2005 and remained constant for the remainder of the study period. The observed annual trends in Mn concentration in the subsurface flow did not appear to be related to years of soil amendment applications. Manganese loads in the surface runoff were significantly higher with CT compared to ST only in 2006, when the total loads were also the highest compared to other years (figure 2e). The total Mn lost throughout the five-year study period was 0.73 and 0.60 kg ha⁻¹ with CT and ST management, respectively.

Sodium. The concentration of Na in the surface runoff was not different between tillage treatments (table 4). On an annual basis, elevated Na concentrations were observed in 2005, regardless of tillage practice (table 4). Once again, elevated Na concentrations in 2005 may be due to elevated water losses that year. The concentration of Na in the subsurface flow was significantly higher in the plots managed with CT compared to ST for all years of the study period (table 5). As in the case of Ca and Mg, lower Na concentrations with ST may be due to dilution caused by increased water volumes with this tillage practice (table 3) (Bosch et al. 2012). Elevated Na concentrations in the subsurface flow were observed in 2008, regardless of tillage treatment (table 5). Sodium loads were significantly higher in the surface runoff from the CT plots compared to the ST plots only in 2006 (figure 2f). The total Na loads were highest in 2005 compared to other years (figure 2f). Over the course of the five-year study period, the total Na lost was 50 kg ha⁻¹ with both tillage treatments.

Phosphorus. The concentration of P in surface runoff was significantly higher in the ST compared to the CT plots only in 2004 (table 4). This trend may be due to lower surface volumes with ST compared to CT (table 3) (Bosch et al. 2012). On an annual basis, lower P concentrations were observed in the surface runoff from the CT plots in 2004 compared to other years. Elevated P concentrations in 2005 and 2007 (table 4) are likely due to the application of poultry litter (table 1), which has been shown to enhance P losses through surface runoff (Endale et al. 2014b). Previous to the study period, poultry litter was applied to all the plots annually from 1999 to 2002. The repeated application of poultry litter can cause P accumulation in agricultural soils (Kingery et al. 1994) and create a potential for enhanced P losses from croplands. Data from the Gibbs Research Farm show increasing P content in the top 2 cm of soil with repeated poultry litter applications (unpublished data). However, elevated P concentrations were not observed at the beginning of the current study, suggesting that this accumulated P is complexed and fixed in the soil. In the subsurface flow, significantly higher P concentrations were observed with CT compared to ST management only in 2008 (table 5). On an annual basis, elevated P concentrations were observed in 2007 and 2008 (table 5). The majority of P loads occurred in the surface runoff compared to the subsurface flow for all years and treatments (figure 2g), likely due to the association of P with the greater amount of sediment particles in the surface runoff. Significantly higher P loads were observed in the surface runoff from the CT compared to the ST plots in 2006 (figure 2g). Over the five-year study period, the total P lost was 2.8 (3.0% of the total P applied) and 3.4 kg ha⁻¹ (3.6% of the total P applied) with CT and ST practices, respectively.

Sulfur. Tillage had no significant effect on the concentration of S in the surface runoff (table 4). On an annual basis, elevated S concentrations were observed in 2004 for the surface runoff from the CT plots only. Gypsum was applied to all the plots in 2004 (table 1), likely resulting in increased S concentrations. However, elevated S concentrations were not observed in other years of gypsum amendments (2006 and 2008; table 1). The concentration of S in the subsurface flow was significantly higher with CT compared to ST for all years (table 5). As in the case of P, elevated S concentrations were observed at the end of the study period (2007 and 2008) regardless of treatment. In fact, gypsum was applied in equal amounts to plots under both tillage treatments. However, the observed annual differences in S concentration for both the surface runoff and subsurface flow samples do not reflect gypsum additions (table 1), suggesting that S concentrations may be related to hydrology. The majority of the S loads occurred in the subsurface flow, regardless of year and tillage treatment (figure 2h). The S load in the surface runoff was significantly higher with CT compared to ST only in 2006 (figure 2h). The highest total S loads were observed in 2005 regardless of tillage treatment, and the total S lost over the course of the five-year study period was 56 and 46 kg ha⁻¹ with CT and ST management, respectively. These losses represent 8.2% and 6.7% of the total S applied in the form of gypsum and fertilizer over the course of the five-year study period (682.3 kg ha⁻¹; table 1) from the CT and ST plots, respectively.

Silicon. The concentration of Si in the surface runoff was significantly higher in the plots managed with CT compared to ST only in 2005 (table 4). The higher Si concentration in the runoff from the CT plots may be due to increased transport of eroded material with this tillage practice (Endale et al. 2014a). Elevated Si concentrations were observed with CT in 2005 (table 4) when higher water volumes have been reported (Bosch et al. 2012). In the subsurface flow, the concentration of Si was not different between tillage treatments (table 5). Regardless of treatment, the annual average Si concentration in the subsurface flow was lowest in 2006 compared to other years (table 5). The Si loads were significantly higher in the surface runoff from the CT compared to the ST plots only in 2006 (figure 2i). On an annual basis, elevated total Si loads were observed in 2005 and 2006 for ST and CT, respectively. The total Si lost over the five-year study period was 45 and 31 kg ha⁻¹ with CT and ST management, respectively.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Principal component analysis of the element load data including (a) the score plot and (b) the loading plot. The first principal component (PC1) explains 60.4% of the variance and separates the element load data based on sample type (surface runoff vs. subsurface flow). The second principal component (PC2) explains 19.5% of the variance and separates the data based on years with high flow (2004 to 2005) and low flow (2006 to 2008).

Zinc. The concentration of Zn in the surface runoff was significantly higher with ST compared to CT management only in 2006 (table 4). On an annual basis, a lower Zn concentration in the surface runoff was observed with CT in 2007 compared to other years (table 4). In the subsurface flow, the concentration of Zn was higher with CT compared to ST management only in 2008 and was highest for both treatments in that same year (table 5). The annual trends in Zn concentration in the surface runoff and subsurface flow did not appear to be related to year of soil amendment application and may be more dependent on hydrology. Zinc loads were higher in the surface runoff compared to subsurface flow in the first three years of the study, regardless of treatment (figure 2j). The five-year Zn loss was 0.55 and 0.75 kg ha⁻¹ with CT and ST management, respectively.

Patterns of Element Losses via Surface Runoff and Subsurface Flow. To investigate potential patterns in element losses due to tillage treatment (CT versus ST) and time (years), a PCA was performed on the element loads (figure 3). The data points representing average element loads were broadly distributed along PC1 (60.4% of variance) and PC2 (19.5% of variance; figure 3a). As shown, the PC1 axis represents variations in the sample type with subsurface flow on the positive scale and surface runoff on the negative scale (figure 3b). Calcium, Mg, and S clustered together on the positive PC1 axis of the PCA (figure 3a) were more strongly correlated to subsurface runoff (positive scale of PC1) and had higher loads in the subsurface flow compared to the surface runoff (figure 2). The relationship between these three elements can be explained as follows. Once applied to the soil, gypsum (CaSO₄) dissociates with moisture into free Ca and SO₄. The resulting SO₄ can bind to excess Mg in the soil complex forming a soluble compound that can leach out of the crop rooting zone. In addition, Ca and Mg behave similarly in soils: both elements have a 2+ charge, both are in the alkaline earth metal group of elements in the periodic table, and both are found together in many minerals. Thus the PCA results are likely driven by differences in water flow path (i.e., surface runoff versus subsurface flow).

The PC2 axis provided an additional degree of separation regarding year with samples from 2006 to 2008 on the positive scale and samples from 2004 to 2005 on the negative scale (figure 3b). The separation along the PC2 axis was likely due to the varying amount of precipitation and flow observed in the various years. For example, 2004 and 2005 (located on the negative scale of PC2) represent years when high flows for both surface runoff and subsurface flow were observed (table 3) (Bosch et al. 2012). Iron, Mn, and Si clustered together on the positive PC2 axis of the PCA (figure 3a) with greatest losses of these elements observed in surface runoff from 2006 with the CT practice (figure 2). This result is likely due to the high concentration of these elements in the soils at this site (up to 31 and 41 kg ha⁻¹ of Fe and Mn, respectively; unpublished data) and their association with the reported elevated sediment loads with CT (Endale et al. 2014a). Finally, K, Na, and Zn clustered together on the negative PC2 axis of the PCA (figure 3a), and the greatest losses of these elements were generally observed during the years with the greatest total water loss (figure 2). Sodium and K are usually weakly held on cation exchange sites and their movement out of the soil profile occurs readily depending on precipitation patterns.

The PCA did not indicate variation in element loss due to tillage treatment. The PCA also did not indicate variation in element loss based on years of soil amendments (poultry litter or gypsum). The discrepancy in poultry litter or gypsum influencing the loss of an element in one year and not in another is likely related to the interval between application and runoff/drainage event date. Furthermore, peanut harvest disturbs the soil, and the interval between harvest and runoff/drainage event might have some impact on the observed differences in element loss and water quality. Thus, the weightings of the element loads on the first two principal components suggest that variation in sample type (surface runoff versus subsurface flow; PC1) and amount of water flow (PC2) had the largest influence on element losses in agricultural fields of the Coastal Plain. This finding has implications on surface and subsurface elemental transport pathways for model-based projections of precipitation patterns for the Southeast (Ingram et al. 2013).