- Email: [email protected]
Potential internal loading of phosphorus in a wetland constructed in agricultural land
Water Research 37 (2003) 965–972
Potential internal loading of phosphorus in a wetland constructed in agricultural land H.K. Pant*, K.R. Reddy Soil and Water Science Department, IFAS, University of Florida, 106 Newell Hall, P.O. Box 110510, Gainesville, FL 32611-0510, USA Received 1 January 2002; received in revised form 1 August 2002; accepted 15 August 2002
Abstract Wetland construction on agricultural or dairy lands could result in solubilization of phosphorus (P) stored in soils and release to the water column. To study the extent of P flux during the start-up period of a constructed wetland, intact soil-cores from areas used for dairy operations, in Okeechobee, Florida, USA were obtained and flooded with adjacent creek water. In the first 28-day hydraulic-retention period, P concentration in the water column increased several fold due to rapid P flux from impacted soils. A continuous decrease in P flux to the water column until the third hydraulic retention cycle (initial influent P concentration 0.2 mg L1), and constant thereafter suggest that the effect of initial influent P upon long-term P flux from soils could be limited. The initial release maybe due to high concentration of labile P in impacted soils; however, slow dissolution of relatively stable P pools could maintain a steady flux, well above of that observed from non-impacted soils. Water soluble P along with double acid-extractable magnesium explained 76% of the variability in cumulative P flux to the water column. Apparently, co-occurrence of active adsorption– desorption phenomena due to independent maintenance of equilibrium by individual P compounds regulates P dynamics of the water column. The results indicated that equilibrium P concentration of the water column of the wetland would be above 1.3 mg L1, which is well above the targeted P level in the water column of the Lake Okeechobee, one of the main water bodies in the area (0.04 mg P L1). This suggests construction of wetlands in agricultural lands could result to substantial internal P loading. However, preventative measures including chemical amendments, establishment of vegetative communities or flushing the initially released P may potentially stabilize the system, and maintain P removal efficiency. r 2002 Elsevier Science Ltd. All rights reserved. Keywords: Adsorption; Desorption; Flux; Phosphorus; Wetlands
1. Introduction Phosphorus flux from soils to the overlying water column depends on various factors including physicochemical characteristics of soils and the P concentration in floodwater. Construction of wetlands on fertilized agricultural or manure-impacted lands such as those used for dairy operations could result in solubilization of P stored in soils and release into the water column. A *Corresponding author. Tel.: +1-352-392-1804; fax: +1352-392-3399. E-mail address: [email protected]fl.edu (H.K. Pant).
significant portion of the water column P can be removed by biotic [1,2], and abiotic processes [3,4]. However, during the initial stabilization period, flooded soil conditions can enhance the dissolution of Ca- and Fe/Al-bound P, and mineralization of organic P. Several studies have reported on the potential use of wetlands for removal of nutrients including P from wastewater [5– 7]. However, wetland soils can function as source or sink for P depending on the quality and quantity of native P [8]. During the stabilization period, storm water treatment areas (STAs) can potentially export P to the water column until the system reaches equilibrium. Phosphorus loading from the adjacent watershed threatens
0043-1354/03/$ - see front matter r 2002 Elsevier Science Ltd. All rights reserved. PII: S 0 0 4 3 - 1 3 5 4 ( 0 2 ) 0 0 4 7 4 - 8
966
H.K. Pant, K.R. Reddy / Water Research 37 (2003) 965–972
the water quality of Lake Okeechobee (Florida, USA) because of the primarily beef cattle ranching and dairy farming in the area. Although 80% of the P remained in soils either in stable or unstable forms, P control measures are needed to meet the target goals [9]. A constructed wetland can be used as an alternative to remove P from runoff waters; however, soil quality could be a detrimental factor in the selection of construction sites. The objectives of this study were to: (1) measure the P flux from soils of different dairy areas to the overlying water column, (2) determine the effect of floodwater P concentration on P flux from soils, and (3) estimate the P sequestration efficiency of the proposed STAs.
2. Materials and methods 2.1. Site description The Nubbin Slough STA construction is a part of the Lake Okeechobee restoration/phosphorus removal project. The proposed STA site (New Palm/Newcommer dairies) is located 2.1 km north of Lake Okeechobee and 10.5 km southeast of the city of Okeechobee, Florida, USA, and occupies an area of approximately 864 ha (Fig. 1). The soils of the proposed site are primarily Spodosols. The abandoned-intensive (areas that had active dairy operations for 34 years until 1992, but as pastures/forages at the time of sampling and is
Fig. 1. The Okeechobee Drainage Basin, FL, USA, illustrating a schematic diagram of the proposed STA. The percentages indicate the areas under the proposed STA, currently occupied by the different components.
H.K. Pant, K.R. Reddy / Water Research 37 (2003) 965–972
intensively impacted), active-intensive (areas with active dairy operations for 20 years and is intensively impacted), and native (non-impacted, forested area) areas occupy 13%, 18% and 7% of the total STA, respectively. Similarly, forage/pasture and grazing/spray field areas occupy 31% each. The adjacent sites to the STA include intensive agriculture, wetland and upland forests, and urban and reclaimed lands. 2.2. Soil-core sampling and physico-chemical analysis To study the P flux from manure-impacted areas, replicate intact soil cores of 30 cm long 6.9 cm (internal diameter) were collected from abandoned-intensive (9 cores), active-intensive (5 cores), forage/pasture (4 cores), and native areas (3 cores) using polypropylene column. To determine selected physico-chemical characteristics, soil samples were collected from the A horizons of the respective areas. Soil pH was determined using 1:2 soil to water ratio. Double acid-extractable (0.0125 M H2SO4+0.05 M HCl) calcium (DA-Ca), magnesium (DA-Mg), and P (DA-P) were extracted as described by Mehlich [10] and analyzed using Inductively Coupled Argon Plasma Emission Spectrometry (Thermo Jarrell Ash ICAP 61E, Franklin, MA). Amorphous and poorly crystalline Fe and Al (hydr)oxides were extracted with oxalate as described by Loeppert and Inskeep [11], and analyzed by the Inductively Coupled Argon Plasma Emission Spectrometry. Water soluble P (WSP) was determined by extracting 5 g soils with 25 mL deionized water for 1 h. The suspensions were centrifuged and filtered as described above. The filtrates were analyzed for total P (TP) using the automated ascorbic acid method following persulfate digestion (Method 365.1, EPA, 1993). Soil TP was determined by combusting 0.5–1.0 g finely ground dry soil at 5501C in a muffle furnace for 4 h and the ash was dissolved in 6 M HCl [12]. The digestate was analyzed for P using an automated ascorbic acid method (Method 365.1, [13]). 2.3. Soil-core flooding and incubation Each soil core was subjected to initial re-wetting by introducing diluted (containing 0.08 mg P L1) filtered adjacent Taylor Creek water (pH 7.4; initial P concentration was 0.4 mg P L1) from the bottom of the cores to avoid entrapment of air and leaching of P. Once the cores were fully saturated, floodwater depth was raised to 10 cm by the slow addition of 450 mL of the diluted filtered site water onto the top of the columns. The flooded cores were kept in the dark at 25721C to prevent algal growth, and slowly aerated to maintain aerobic conditions (oxygen level 571 mg L1). The 28day hydraulic retention cycle (i.e., 28 days of flooding) was used to mimic the general hydraulic retention time
967
of a constructed wetland. The floodwater was sampled at 0, 1, 3, 7, 14, 21 and 28 days, and filtered through a 0.45 mm filter and analyzed for soluble reactive P (SRP), using the automated ascorbic acid method as described earlier. Total dissolved P (TDP) was also determined in the floodwater by the ascorbic acid method following persulfate digestion as described previously. The sampled floodwater was replenished with equal volume of filtered site water after each sampling; however, the reduction in floodwater level due to evaporation was replenished with distilled water prior to sampling. To estimate equilibrium P concentration of the water column (EPCw), seven more 28-day hydraulic retention cycles were repeated with diluted or spiked site water (Creek water; P level 0.4 mg L1) of 0.14, 0.20, 0.4, 1.4, 2.4, 5.4 and 10.4 mg P L1 initial concentrations, and floodwater was sampled intermittently and analyzed for SRP as described above. 2.4. Phosphorus flux estimation Since there was no significant difference between TDP and SRP, the potential rates of P flux from soil to the water column were estimated by plotting cumulative SRP released by per unit surface area of the soil column to the floodwater. The slope of P release (mg m2) vs. flooding period (days) was taken as an average potential P flux (mg m2 day1) from soil to the water column. Unless otherwise stated, experiments were carried out at least in triplicate, and statistical analysis was performed using Statgraphics plus version 3.1 [14], and significance test was performed at pp0.05 level. 2.5. Water column equilibrium phosphorus concentration To determine the EPCw, the site water (Taylor Creek water) was spiked with 1, 2, 5 and 10 mg P L1 solutions, and used for flooding the soil-cores in consecutive hydraulic-retention cycles. Phosphorus concentrations were measured as described previously, and P retention or release was calculated. The relationship between P flux or retention ðPflux Þ and the water column P concentration generally followed logarithmic pattern as described by the following equation: Pflux ¼ Ka * ln C0 þ Pk ;
ð1Þ
where Pflux is the net P flux or retention by per unit surface area of soils (mg m2), Ka is a constant representing P assimilation coefficient (L m2), C0 is the influent P concentration (mg L1), and Pk is a constant representing initial desorbed P in per unit surface area of soils (mg m2) Pflux ¼ ðC0 Ct Þ * V =A;
ð2Þ
where, Ct is the final floodwater P concentration at time ðtÞ 28 days, V is the volume of influent (L) and A is the
968
H.K. Pant, K.R. Reddy / Water Research 37 (2003) 965–972
surface area of soils (m2). By plotting the logarithmic form of Eq. (1), i.e., Pflux vs. C0 ; the (negative) slope is equal to Ka : By substituting the values of Pflux and C0 in Eq. (1), equilibrium water column P concentration (EPCw; mg L1; defined as the P concentration in the water column at which no net flux or retention of P occurs from the soils/sediments), i.e., Pflux ¼ 0; can be calculated as follows: lnðEPCw Þ ¼ Pk =Ka :
ð3Þ
3. Results and discussion The P flux from soils to the water column was mostly SRP, which indicates the direct increase in available P in the water column. Phosphorus concentration of the water column increased several fold due to flux from soils of abandoned- and active-intensive areas at the end of 28-day hydraulic-retention time of the 1st hydraulicretention cycle (Fig. 2). However, during subsequent hydraulic-retention cycles, water column P substantially decreased. Overall increase in water column P was relatively low from soils of forage/pasture areas. In general, final water column P was highest in soil-cores obtained from active-intensive area followed by abandoned-intensive, forage/pasture, and native areas, respectively. In the 1st hydraulic-retention cycle, the potential P flux to the water column varied from 15 to 93, 27 to 61 and 3 to 31 mg P m2 day1, respectively, from abandoned-intensive, active-intensive, and forage/ pasture areas. However, the flux was substantially lower from native soils ranging from 3 to 10 mg P m2 day1. The average P flux to the water column in four hydraulic retention cycles was highest from active-intensive soils (2579 mg P m2 day1) followed by abandoned-intensive (2279 mg P m2 day1), forage/pasture (1174 mg P m2 day1), and native (472 mg P m2 day1) areas, respectively. Higher P flux from soils of abandoned- and active-intensive areas was expected because of high accumulations of P resulting from intensive dairy operations. Soils of these areas have high concentrations of labile P as shown by Mehlich P (DA-P; Table 1). In addition, flooding probably resulted in release of P from solubilization of Fe- and Al-bound P. Although a high degree of variability was observed in P flux from soils to the water column among replicates of the same area, the difference in potential P flux from soils of abandoned- and active-intensive areas reduced substantially in subsequent cycles (Fig. 3) possibly due to depletion of soluble P. The potential P flux from soils of abandoned-intensive area to the water column decreased steadily up to 4th hydraulic-retention cycle. The overall decrease in P flux in the consecutive hydraulic-retention cycles may have also been affected by the initial concentrations of P in the water used to
flood the soil-cores resulting in alteration of P equilibrium status [15,16]. Soil-cores collected from activeintensive area had maximum cumulative P flux to the water columns followed by abandoned-intensive, forage/ pasture, and native areas, respectively (Table 2). These results from abandoned- and active-intensive areas suggest that P content of the A horizons is mainly involved in the P exchange processes. As expected, P flux declined up to the 3rd cycle because of the decrease in solubilization of stored P in soil-cores as well as the increases in initial floodwater P concentrations (Fig. 3), which suppressed the release of P from soils due to changes in the P equilibrium [17]. A continuous decrease in P flux to the water column up to 3rd cycle (initial floodwater P concentration 0.2 mg L1), and relatively constant thereafter, suggest that initial floodwater P (at o0.2 mg P L1) may have a negative effect on P flux from the soils. The total P flux was slightly higher in the 4th cycle than the 3rd, perhaps due to the slow dissolution of some of the inorganic P compounds, and mineralization of organic P [18] from active intensive area. It can be predicted that P flux to the water column could remain constant from the 3rd hydraulic-retention cycle until the relatively available (double acid-extractable) P is exhausted from soils given the initial floodwater P concentration remains in the range 0.2–0.4 mg L1. Using average P flux to the water column, and available P content in A horizon soils (DAP; calculated by dividing cumulative DA-P in A horizon by average P flux per day), it is estimated that soils from abandoned-intensive area could take an average of 3 years to stop releasing P to the water column. However, the other areas would take less than a year (9 months for active-intensive, and 6 months for forage/pasture areas). Although the variability of the data for the estimation of a time frame to exhaust available P was quite high due to the heterogeneity within the individual dairy areas, the actual time required for STAs to start to sequester P could be quite long under the given conditions. EPCw was highest for the active-intensive area (3.4 mg L1) followed by abandoned-intensive (2.6 mg L1), forage/pasture (1.8 mg L1) and native (1.3 mg L1) areas, respectively (Fig. 4). Multiple regression analysis showed that 68% of the variability in EPCw was explained by double acid-extractable Ca and Mg, and WSP as given in the model below: EPCw ¼ 0:899101 þ 0:000008½DA-Ca 0:000091½DA-Mg þ 0:001182½WSP (all values are expressed in mg m2 except EPCw in mg L1, R2 ¼ 68%, and pp0.05). This predictive model showed that double acidextractable Ca had a direct (positive) correlation with EPCw, which may indicate that P associated with Ca is responsible partly in increasing EPCw. Double
H.K. Pant, K.R. Reddy / Water Research 37 (2003) 965–972
969
Abandoned-Intensive 2000 1600 1200 800 400 0
C0= 0.08 mgP L-1
0
20
C0= 0.14 mgP L-1
40
C0= 0.20 mgP L-1
80
60
C0= 0.40 mgP L-1
100
120
Active-Intensive C0= 0.08 mgP L-1
Water column P (mg m-2)
2000 1600
C0= 0.14 mgP L-1
C0= 0.20 mgP L-1
1200 800 400 0 0
20
40
80
60
C0= 0.40 mgP L-1
100
120
Forage/Improved Pasture C0= 0.08 mgP L-1
600
C0= 0.14 mgP L-1
C0= 0.20 mgP L-1
400
C0= 0.40 mgP L-1
200 0 0
20
40
80
60
100
120
Native C0= 0.08 mgP L-1
600
C0= 0.14 mgP L-1
400
C0= 0.20 mgP L-1
C0= 0.40 mgP L-1
200 0 0
20
40
80
60
100
120
Flooding period (days) Fig. 2. Phosphorus flux to the water column from different dairy areas.
Table 1 Selected physico-chemical characteristics of A-horizon (average depth 12 cm) soils Field area
pH WSP (g m2) DA-Ca (g m2) DA-Mg (g m2) DA-P (g m2) TP (g m2) ox-Fe (g m2) ox-Al (g m2)
Abandon-intensive Active-intensive Forage/pasture Native
6.2 6.2 5.7 5.0
1.9 2.1 0.5 0.7
193 162 118 103
13.0 12.2 9.0 16.3
20.5 6.9 2.5 0.6
52.4 20.4 15.1 19.1
21.5 12.5 23.2 36.9
79.5 10.1 12.6 12.6
WSP: water soluble phosphorus, DA-Ca: double acid-extractable calcium, DA-Mg: double acid-extractable magnesium, DA-P: double acid-extractable phosphorus, TP: total phosphorus, ox-Fe: oxalate-extractable iron, and ox-Al: oxalate-extractable aluminum.
acid-extractable Mg, however, had inverse (negative) correlation with EPCw. An inverse correlation between EPCw and double acid-extractable Mg may indicate that the acid-extractable Mg sorbed the released P from the
other constituents. In general, the mass ratio of Ca to Mg was high, i.e., the systems were relatively dominated with carbonates of Ca compared to that of Mg, which left Mg available for adsorption of P. Thus, carbonates
H.K. Pant, K.R. Reddy / Water Research 37 (2003) 965–972
970
largely dictates the EPCw. Readily available P (i.e., water soluble) increases due to slow transformation of relatively stable P during the incubation period. This tends to saturate available sites for P sorption, consequently results in increase in EPCw. WSP alone could explain 54% of the variability in potential P flux during 2nd and 3rd, and 51% during 4th hydraulic-retention cycles. However, WSP only explains 21% of that in the 1st hydraulic-retention cycle (Table 3). The reason for low predictability for the potential P flux in the 1st hydraulic-retention cycle is not known. However, the possible co-occurrence of both active adsorption and desorption phenomena [19] due to independent maintenance of P equilibrium by individual P compounds [20] may have played a major role. Although decrease in potential P flux occurred with the increase in initial P concentration of the water column, no significant correlation was obtained perhaps, because of the confounding of the effect with P depletion. Water soluble P could also explain 57% and 58% of the variability in cumulative P flux to the water column in 2nd and 4th hydraulic-retention cycles. Similarly, in 3rd hydraulic-retention cycle, water soluble P along with double acid-extractable Mg could explain 76% of variability in cumulative P flux to the water column, whereas in 1st hydraulic-retention cycle, WSP could only explain 17% of the variability. It is apparent that in initial flooding, soluble P releases from soils to the water column and is adsorbed by soil constituents, which have high affinity for P including oxalate-extractable iron (oxFe) and oxalate-extractable aluminium (ox-Al), thereafter, relatively passive alternate P adsorption–desorption phenomena in soils may regulate the P dynamic of the water column.
associated Mg could play a greater role than previously thought in P sequestration or advection processes in wetland systems. Water soluble P of the surface soils
Fig. 3. Potential phosphorus flux rates to the water column from various dairy areas in different hydraulic-retention cycles. Table 2 Phosphorus flux from different dairy areas to the overlying water columns in four hydraulic retention cycles Field area
Area coverage (ha) P flux (kg ha1) % of TPa
Abandoned-intensive 112 Active-intensive 152 Forage/pasture 272 Native 64
26 31 10 6
10a 17a 6bc 3c
a Flux as percent of total phosphorus in the A-horizon soils, and values with the same letter in a column are not significantly different.
Native y = -0.47Ln(x) + 0.133 R2 = 0.88 EPCw = 1.3 mg L-1
Phosphorus retention/release (mg m-2)
2 1 0
0
4
8
2 12
0
-1
-2
-2
-4 Active-Intensive y = -2.17Ln(x) + 2.63 R2 = 0.77 EPCw = 3.4 mg L-1
12 6 0
0
48
Forage / Pasture y = -0.85Ln(x) + 0.520 R2 = 0.77 EPCw = 1.8 mg L-1
4
0
5 0
-6
-5
-12
-10
8
12
Abandoned-Intensive y = -1.99Ln(x) + 1.88 R2 = 0.91 EPCw = 2.6 mg L-1
10
12
4
0
4
8
12
Initial influent phosphorus (mg L-1) Fig. 4. Effects of initial influent phosphorus concentration on phosphorus flux to the water column from different dairy areas. Negative sign on Y-axis represents retention while positive one shows releases.
H.K. Pant, K.R. Reddy / Water Research 37 (2003) 965–972
971
Table 3 Multiple regression analysis of potential and cumulative phosphorus flux from soils to the water column in different hydraulicretention cycles, and selected independent variables Hydraulic-retention cycle
C0
Potential phosphorus flux to the water column 1st 0.08 2nd 0.14 3rd 0.20 4th 0.40
Fitted model equation P P P P
flux¼ 17:7 þ 0:01[WSP] flux¼ 9:5 þ 0:01[WSP] flux¼ 5:6 þ 0:01[WSP] flux¼ 3:8 þ 0:004[WSP]
Cumulative phosphorus flux to the water column 1st 0.08 CP flux¼ 580 þ 0:21[WSP] 2nd 0.14 CP flux¼ 228 þ 0:24[WSP] 3rd 0.20 CP flux¼ 185 þ 0:15[WSP]20:01[DA-Mg] 4th 0.40 CP flux¼ 136 þ 0:14[WSP]
R2 21 54 54 51
17 57 76 58
SE 20.8 11.4 7.7 6.9
729 327 138 187
DW 1.7 2.5 2.1 2.0
1.7 2.3 2.1 1.9
C0 : initial influent P concentration (mg L1), R2 : equivalent to variability explain (%), SE: standard error of estimation, DW: Durbin– Watson statistic (value >1.4 indicates no auto-correlation in residuals), P flux: potential phosphorus flux (mg m2 day–1), WSP: water soluble phosphorus (mg kg1), CP flux: cumulative phosphorus flux (total flux in 28-day cycle, mg m2), and DA-Mg: double acidextractable magnesium (mg kg1).
4. Conclusions This study indicated that highly soluble P accumulated from dairy operations may be released during first 28 days of flooding, but the slow dissolution of relatively stable pools of P (Fe-, Al-, Ca- and Mg-bound P) could maintain a long-term steady flux, well above that observed from non-impacted (native) areas. Moreover, the equilibrium P concentration of the water column of all the dairy areas were over 1.3 mg L1, which means the storm water treatment areas (STA) will not be able to remove any amounts of P as long as influent P level remains equal or below 1.3 mg L1. Since the targeted P level in the water column of the Lake Okeechobee is well below to 0.04 mg L1, it is virtually impossible to have any of the STA constructed in these lands for reduction of P influx to the lake. Thus, P flux potentials of soils should be given serious consideration prior to STA establishment in agricultural or manure-impacted lands, otherwise massive internal P loading could undermine the effective use of the wetlands. Preventative measures such as treating soils with chemical amendments, establishing vegetative communities or flushing the initially released P, however, may potentially stabilize the system, and maintain P removal efficiency.
Acknowledgements The authors acknowledge the help from J.E. Colson, I.C. Torres, H.M. Spencer, V.D. Nair, and D.A. Graetz, University of Florida, Soil and Water Science Department, Gainesville, Florida. This research was supported by the Florida Agricultural Experiment Station and a
grant from US Army Corps of Engineers, Florida, and approved for publication as Journal Series No. R-08183.
References [1] Newbold JD, Elwood JW, O’Neil RV, Sheldon AL. Phosphorus dynamics in a woodland stream ecosystem: a study of nutrient spiraling. Ecology 1983;64(5):1249–63. [2] Greenway M, Woolley A. Constructed wetlands in Queensland: performance efficiency and nutrient bioaccumulation. Ecol Eng 1999;12(1–2):39–55. [3] Diaz OA, Reddy KR, Moore PA. Solubility of inorganic P in stream water as influenced by pH and Ca concentration. Water Res 1994;28(8):1755–63. [4] Tanner CC, Sukias JPS, Upsdell MP. Relationships between loading rates and pollutant removal during maturation of gravel-bed constructed wetlands. J Environ Qual 1998;27(2):448–58. [5] Reddy KR, Smith WH. Aquatic plants for water treatment and resource recovery. Orlando, FL: Magnolia Publishing Inc., 1987. p. 1032. [6] Kadlec RH, Knight RL. Treatment wetlands. Ann Arbor, MI: Lewis Publishers, 1995. [7] Pant HK, Nair VD, Reddy KR, Graetz GA, Villapando RR. Influence of flooding on phosphorus solubility in manure-impacted soils. J Environ Qual 2002;4:1399–405. [8] Richardson CJ. Mechanisms controlling phosphorus retention capacity in freshwater wetlands. Science 1985;228(4706):1424–7. [9] Flaig EG, Reddy KR. Fate of phosphorus in the Lake Okeechobee watershed, Florida USA: overview and recommendations. Ecol Eng 1995;5(2+3):127–42. [10] Mehlich A. Determination of P, Ca, Mg, K, Na, and NH4. In: Soil testing div. pub., vols. 1–53. Raleigh, NC: North Carolina Dep. Agric., 1953.
972
H.K. Pant, K.R. Reddy / Water Research 37 (2003) 965–972
[11] Loeppert RH, Inskeep WP. Iron. In: DL Sparks et al., editors. Methods of Soil Analysis, Part 3. Madison, WI: Soil Science Society of America Inc., and American Society of Agronomy Inc., 1996. p. 639–64. [12] Andersen JM. An ignition method for determination of total phosphorus in lake sediments. Water Res 1976; 10(4):329–31. [13] EPA (US Environmental Protection Agency) Methods of chemical analysis of water and wastes. Cincinnati, OH: Environ. Monit. Support Lab., 1993. [14] Statistical Graphics. Statgraphics plus version 3.1. Rockville, MD: Statistical Graphics Corp, 1997. [15] Mayer LM, Gloss SP. Buffering of silica and phosphate in a turbid river. Limnol Oceanogr 1980;25(1):12–22.
[16] Froelich PN. Kinetic control of dissolved phosphate in natural rivers and estuaries: a primer on the phosphate buffer mechanism. Limnol Oceanogr 1988;33(4):649–68. [17] Pant HK, Reddy KR, Lemon E. Phosphorus retention capacity of root bed media of sub-surface flow constructed wetlands. Ecol Eng 2001a; 17(4):345–55. [18] Sanchez PA. Properties, management of soils in the tropics. NY: Wiley, 1976. [19] Barrow NJ. A mechanistic model for describing the sorption and desorption of phosphate by soil. J Soil Sci 1983;34(4):733–50. [20] Pant HK, Reddy KR. Phosphorus sorption characteristics of estuarine sediments under different redox conditions. J Environ Qual 2001;30(4):1474–80.
Potential internal loading of phosphorus in a wetland constructed in agricultural land H.K. Pant*, K.R. Reddy Soil and Water Science Department, IFAS, University of Florida, 106 Newell Hall, P.O. Box 110510, Gainesville, FL 32611-0510, USA Received 1 January 2002; received in revised form 1 August 2002; accepted 15 August 2002
Abstract Wetland construction on agricultural or dairy lands could result in solubilization of phosphorus (P) stored in soils and release to the water column. To study the extent of P flux during the start-up period of a constructed wetland, intact soil-cores from areas used for dairy operations, in Okeechobee, Florida, USA were obtained and flooded with adjacent creek water. In the first 28-day hydraulic-retention period, P concentration in the water column increased several fold due to rapid P flux from impacted soils. A continuous decrease in P flux to the water column until the third hydraulic retention cycle (initial influent P concentration 0.2 mg L1), and constant thereafter suggest that the effect of initial influent P upon long-term P flux from soils could be limited. The initial release maybe due to high concentration of labile P in impacted soils; however, slow dissolution of relatively stable P pools could maintain a steady flux, well above of that observed from non-impacted soils. Water soluble P along with double acid-extractable magnesium explained 76% of the variability in cumulative P flux to the water column. Apparently, co-occurrence of active adsorption– desorption phenomena due to independent maintenance of equilibrium by individual P compounds regulates P dynamics of the water column. The results indicated that equilibrium P concentration of the water column of the wetland would be above 1.3 mg L1, which is well above the targeted P level in the water column of the Lake Okeechobee, one of the main water bodies in the area (0.04 mg P L1). This suggests construction of wetlands in agricultural lands could result to substantial internal P loading. However, preventative measures including chemical amendments, establishment of vegetative communities or flushing the initially released P may potentially stabilize the system, and maintain P removal efficiency. r 2002 Elsevier Science Ltd. All rights reserved. Keywords: Adsorption; Desorption; Flux; Phosphorus; Wetlands
1. Introduction Phosphorus flux from soils to the overlying water column depends on various factors including physicochemical characteristics of soils and the P concentration in floodwater. Construction of wetlands on fertilized agricultural or manure-impacted lands such as those used for dairy operations could result in solubilization of P stored in soils and release into the water column. A *Corresponding author. Tel.: +1-352-392-1804; fax: +1352-392-3399. E-mail address: [email protected]fl.edu (H.K. Pant).
significant portion of the water column P can be removed by biotic [1,2], and abiotic processes [3,4]. However, during the initial stabilization period, flooded soil conditions can enhance the dissolution of Ca- and Fe/Al-bound P, and mineralization of organic P. Several studies have reported on the potential use of wetlands for removal of nutrients including P from wastewater [5– 7]. However, wetland soils can function as source or sink for P depending on the quality and quantity of native P [8]. During the stabilization period, storm water treatment areas (STAs) can potentially export P to the water column until the system reaches equilibrium. Phosphorus loading from the adjacent watershed threatens
0043-1354/03/$ - see front matter r 2002 Elsevier Science Ltd. All rights reserved. PII: S 0 0 4 3 - 1 3 5 4 ( 0 2 ) 0 0 4 7 4 - 8
966
H.K. Pant, K.R. Reddy / Water Research 37 (2003) 965–972
the water quality of Lake Okeechobee (Florida, USA) because of the primarily beef cattle ranching and dairy farming in the area. Although 80% of the P remained in soils either in stable or unstable forms, P control measures are needed to meet the target goals [9]. A constructed wetland can be used as an alternative to remove P from runoff waters; however, soil quality could be a detrimental factor in the selection of construction sites. The objectives of this study were to: (1) measure the P flux from soils of different dairy areas to the overlying water column, (2) determine the effect of floodwater P concentration on P flux from soils, and (3) estimate the P sequestration efficiency of the proposed STAs.
2. Materials and methods 2.1. Site description The Nubbin Slough STA construction is a part of the Lake Okeechobee restoration/phosphorus removal project. The proposed STA site (New Palm/Newcommer dairies) is located 2.1 km north of Lake Okeechobee and 10.5 km southeast of the city of Okeechobee, Florida, USA, and occupies an area of approximately 864 ha (Fig. 1). The soils of the proposed site are primarily Spodosols. The abandoned-intensive (areas that had active dairy operations for 34 years until 1992, but as pastures/forages at the time of sampling and is
Fig. 1. The Okeechobee Drainage Basin, FL, USA, illustrating a schematic diagram of the proposed STA. The percentages indicate the areas under the proposed STA, currently occupied by the different components.
H.K. Pant, K.R. Reddy / Water Research 37 (2003) 965–972
intensively impacted), active-intensive (areas with active dairy operations for 20 years and is intensively impacted), and native (non-impacted, forested area) areas occupy 13%, 18% and 7% of the total STA, respectively. Similarly, forage/pasture and grazing/spray field areas occupy 31% each. The adjacent sites to the STA include intensive agriculture, wetland and upland forests, and urban and reclaimed lands. 2.2. Soil-core sampling and physico-chemical analysis To study the P flux from manure-impacted areas, replicate intact soil cores of 30 cm long 6.9 cm (internal diameter) were collected from abandoned-intensive (9 cores), active-intensive (5 cores), forage/pasture (4 cores), and native areas (3 cores) using polypropylene column. To determine selected physico-chemical characteristics, soil samples were collected from the A horizons of the respective areas. Soil pH was determined using 1:2 soil to water ratio. Double acid-extractable (0.0125 M H2SO4+0.05 M HCl) calcium (DA-Ca), magnesium (DA-Mg), and P (DA-P) were extracted as described by Mehlich [10] and analyzed using Inductively Coupled Argon Plasma Emission Spectrometry (Thermo Jarrell Ash ICAP 61E, Franklin, MA). Amorphous and poorly crystalline Fe and Al (hydr)oxides were extracted with oxalate as described by Loeppert and Inskeep [11], and analyzed by the Inductively Coupled Argon Plasma Emission Spectrometry. Water soluble P (WSP) was determined by extracting 5 g soils with 25 mL deionized water for 1 h. The suspensions were centrifuged and filtered as described above. The filtrates were analyzed for total P (TP) using the automated ascorbic acid method following persulfate digestion (Method 365.1, EPA, 1993). Soil TP was determined by combusting 0.5–1.0 g finely ground dry soil at 5501C in a muffle furnace for 4 h and the ash was dissolved in 6 M HCl [12]. The digestate was analyzed for P using an automated ascorbic acid method (Method 365.1, [13]). 2.3. Soil-core flooding and incubation Each soil core was subjected to initial re-wetting by introducing diluted (containing 0.08 mg P L1) filtered adjacent Taylor Creek water (pH 7.4; initial P concentration was 0.4 mg P L1) from the bottom of the cores to avoid entrapment of air and leaching of P. Once the cores were fully saturated, floodwater depth was raised to 10 cm by the slow addition of 450 mL of the diluted filtered site water onto the top of the columns. The flooded cores were kept in the dark at 25721C to prevent algal growth, and slowly aerated to maintain aerobic conditions (oxygen level 571 mg L1). The 28day hydraulic retention cycle (i.e., 28 days of flooding) was used to mimic the general hydraulic retention time
967
of a constructed wetland. The floodwater was sampled at 0, 1, 3, 7, 14, 21 and 28 days, and filtered through a 0.45 mm filter and analyzed for soluble reactive P (SRP), using the automated ascorbic acid method as described earlier. Total dissolved P (TDP) was also determined in the floodwater by the ascorbic acid method following persulfate digestion as described previously. The sampled floodwater was replenished with equal volume of filtered site water after each sampling; however, the reduction in floodwater level due to evaporation was replenished with distilled water prior to sampling. To estimate equilibrium P concentration of the water column (EPCw), seven more 28-day hydraulic retention cycles were repeated with diluted or spiked site water (Creek water; P level 0.4 mg L1) of 0.14, 0.20, 0.4, 1.4, 2.4, 5.4 and 10.4 mg P L1 initial concentrations, and floodwater was sampled intermittently and analyzed for SRP as described above. 2.4. Phosphorus flux estimation Since there was no significant difference between TDP and SRP, the potential rates of P flux from soil to the water column were estimated by plotting cumulative SRP released by per unit surface area of the soil column to the floodwater. The slope of P release (mg m2) vs. flooding period (days) was taken as an average potential P flux (mg m2 day1) from soil to the water column. Unless otherwise stated, experiments were carried out at least in triplicate, and statistical analysis was performed using Statgraphics plus version 3.1 [14], and significance test was performed at pp0.05 level. 2.5. Water column equilibrium phosphorus concentration To determine the EPCw, the site water (Taylor Creek water) was spiked with 1, 2, 5 and 10 mg P L1 solutions, and used for flooding the soil-cores in consecutive hydraulic-retention cycles. Phosphorus concentrations were measured as described previously, and P retention or release was calculated. The relationship between P flux or retention ðPflux Þ and the water column P concentration generally followed logarithmic pattern as described by the following equation: Pflux ¼ Ka * ln C0 þ Pk ;
ð1Þ
where Pflux is the net P flux or retention by per unit surface area of soils (mg m2), Ka is a constant representing P assimilation coefficient (L m2), C0 is the influent P concentration (mg L1), and Pk is a constant representing initial desorbed P in per unit surface area of soils (mg m2) Pflux ¼ ðC0 Ct Þ * V =A;
ð2Þ
where, Ct is the final floodwater P concentration at time ðtÞ 28 days, V is the volume of influent (L) and A is the
968
H.K. Pant, K.R. Reddy / Water Research 37 (2003) 965–972
surface area of soils (m2). By plotting the logarithmic form of Eq. (1), i.e., Pflux vs. C0 ; the (negative) slope is equal to Ka : By substituting the values of Pflux and C0 in Eq. (1), equilibrium water column P concentration (EPCw; mg L1; defined as the P concentration in the water column at which no net flux or retention of P occurs from the soils/sediments), i.e., Pflux ¼ 0; can be calculated as follows: lnðEPCw Þ ¼ Pk =Ka :
ð3Þ
3. Results and discussion The P flux from soils to the water column was mostly SRP, which indicates the direct increase in available P in the water column. Phosphorus concentration of the water column increased several fold due to flux from soils of abandoned- and active-intensive areas at the end of 28-day hydraulic-retention time of the 1st hydraulicretention cycle (Fig. 2). However, during subsequent hydraulic-retention cycles, water column P substantially decreased. Overall increase in water column P was relatively low from soils of forage/pasture areas. In general, final water column P was highest in soil-cores obtained from active-intensive area followed by abandoned-intensive, forage/pasture, and native areas, respectively. In the 1st hydraulic-retention cycle, the potential P flux to the water column varied from 15 to 93, 27 to 61 and 3 to 31 mg P m2 day1, respectively, from abandoned-intensive, active-intensive, and forage/ pasture areas. However, the flux was substantially lower from native soils ranging from 3 to 10 mg P m2 day1. The average P flux to the water column in four hydraulic retention cycles was highest from active-intensive soils (2579 mg P m2 day1) followed by abandoned-intensive (2279 mg P m2 day1), forage/pasture (1174 mg P m2 day1), and native (472 mg P m2 day1) areas, respectively. Higher P flux from soils of abandoned- and active-intensive areas was expected because of high accumulations of P resulting from intensive dairy operations. Soils of these areas have high concentrations of labile P as shown by Mehlich P (DA-P; Table 1). In addition, flooding probably resulted in release of P from solubilization of Fe- and Al-bound P. Although a high degree of variability was observed in P flux from soils to the water column among replicates of the same area, the difference in potential P flux from soils of abandoned- and active-intensive areas reduced substantially in subsequent cycles (Fig. 3) possibly due to depletion of soluble P. The potential P flux from soils of abandoned-intensive area to the water column decreased steadily up to 4th hydraulic-retention cycle. The overall decrease in P flux in the consecutive hydraulic-retention cycles may have also been affected by the initial concentrations of P in the water used to
flood the soil-cores resulting in alteration of P equilibrium status [15,16]. Soil-cores collected from activeintensive area had maximum cumulative P flux to the water columns followed by abandoned-intensive, forage/ pasture, and native areas, respectively (Table 2). These results from abandoned- and active-intensive areas suggest that P content of the A horizons is mainly involved in the P exchange processes. As expected, P flux declined up to the 3rd cycle because of the decrease in solubilization of stored P in soil-cores as well as the increases in initial floodwater P concentrations (Fig. 3), which suppressed the release of P from soils due to changes in the P equilibrium [17]. A continuous decrease in P flux to the water column up to 3rd cycle (initial floodwater P concentration 0.2 mg L1), and relatively constant thereafter, suggest that initial floodwater P (at o0.2 mg P L1) may have a negative effect on P flux from the soils. The total P flux was slightly higher in the 4th cycle than the 3rd, perhaps due to the slow dissolution of some of the inorganic P compounds, and mineralization of organic P [18] from active intensive area. It can be predicted that P flux to the water column could remain constant from the 3rd hydraulic-retention cycle until the relatively available (double acid-extractable) P is exhausted from soils given the initial floodwater P concentration remains in the range 0.2–0.4 mg L1. Using average P flux to the water column, and available P content in A horizon soils (DAP; calculated by dividing cumulative DA-P in A horizon by average P flux per day), it is estimated that soils from abandoned-intensive area could take an average of 3 years to stop releasing P to the water column. However, the other areas would take less than a year (9 months for active-intensive, and 6 months for forage/pasture areas). Although the variability of the data for the estimation of a time frame to exhaust available P was quite high due to the heterogeneity within the individual dairy areas, the actual time required for STAs to start to sequester P could be quite long under the given conditions. EPCw was highest for the active-intensive area (3.4 mg L1) followed by abandoned-intensive (2.6 mg L1), forage/pasture (1.8 mg L1) and native (1.3 mg L1) areas, respectively (Fig. 4). Multiple regression analysis showed that 68% of the variability in EPCw was explained by double acid-extractable Ca and Mg, and WSP as given in the model below: EPCw ¼ 0:899101 þ 0:000008½DA-Ca 0:000091½DA-Mg þ 0:001182½WSP (all values are expressed in mg m2 except EPCw in mg L1, R2 ¼ 68%, and pp0.05). This predictive model showed that double acidextractable Ca had a direct (positive) correlation with EPCw, which may indicate that P associated with Ca is responsible partly in increasing EPCw. Double
H.K. Pant, K.R. Reddy / Water Research 37 (2003) 965–972
969
Abandoned-Intensive 2000 1600 1200 800 400 0
C0= 0.08 mgP L-1
0
20
C0= 0.14 mgP L-1
40
C0= 0.20 mgP L-1
80
60
C0= 0.40 mgP L-1
100
120
Active-Intensive C0= 0.08 mgP L-1
Water column P (mg m-2)
2000 1600
C0= 0.14 mgP L-1
C0= 0.20 mgP L-1
1200 800 400 0 0
20
40
80
60
C0= 0.40 mgP L-1
100
120
Forage/Improved Pasture C0= 0.08 mgP L-1
600
C0= 0.14 mgP L-1
C0= 0.20 mgP L-1
400
C0= 0.40 mgP L-1
200 0 0
20
40
80
60
100
120
Native C0= 0.08 mgP L-1
600
C0= 0.14 mgP L-1
400
C0= 0.20 mgP L-1
C0= 0.40 mgP L-1
200 0 0
20
40
80
60
100
120
Flooding period (days) Fig. 2. Phosphorus flux to the water column from different dairy areas.
Table 1 Selected physico-chemical characteristics of A-horizon (average depth 12 cm) soils Field area
pH WSP (g m2) DA-Ca (g m2) DA-Mg (g m2) DA-P (g m2) TP (g m2) ox-Fe (g m2) ox-Al (g m2)
Abandon-intensive Active-intensive Forage/pasture Native
6.2 6.2 5.7 5.0
1.9 2.1 0.5 0.7
193 162 118 103
13.0 12.2 9.0 16.3
20.5 6.9 2.5 0.6
52.4 20.4 15.1 19.1
21.5 12.5 23.2 36.9
79.5 10.1 12.6 12.6
WSP: water soluble phosphorus, DA-Ca: double acid-extractable calcium, DA-Mg: double acid-extractable magnesium, DA-P: double acid-extractable phosphorus, TP: total phosphorus, ox-Fe: oxalate-extractable iron, and ox-Al: oxalate-extractable aluminum.
acid-extractable Mg, however, had inverse (negative) correlation with EPCw. An inverse correlation between EPCw and double acid-extractable Mg may indicate that the acid-extractable Mg sorbed the released P from the
other constituents. In general, the mass ratio of Ca to Mg was high, i.e., the systems were relatively dominated with carbonates of Ca compared to that of Mg, which left Mg available for adsorption of P. Thus, carbonates
H.K. Pant, K.R. Reddy / Water Research 37 (2003) 965–972
970
largely dictates the EPCw. Readily available P (i.e., water soluble) increases due to slow transformation of relatively stable P during the incubation period. This tends to saturate available sites for P sorption, consequently results in increase in EPCw. WSP alone could explain 54% of the variability in potential P flux during 2nd and 3rd, and 51% during 4th hydraulic-retention cycles. However, WSP only explains 21% of that in the 1st hydraulic-retention cycle (Table 3). The reason for low predictability for the potential P flux in the 1st hydraulic-retention cycle is not known. However, the possible co-occurrence of both active adsorption and desorption phenomena [19] due to independent maintenance of P equilibrium by individual P compounds [20] may have played a major role. Although decrease in potential P flux occurred with the increase in initial P concentration of the water column, no significant correlation was obtained perhaps, because of the confounding of the effect with P depletion. Water soluble P could also explain 57% and 58% of the variability in cumulative P flux to the water column in 2nd and 4th hydraulic-retention cycles. Similarly, in 3rd hydraulic-retention cycle, water soluble P along with double acid-extractable Mg could explain 76% of variability in cumulative P flux to the water column, whereas in 1st hydraulic-retention cycle, WSP could only explain 17% of the variability. It is apparent that in initial flooding, soluble P releases from soils to the water column and is adsorbed by soil constituents, which have high affinity for P including oxalate-extractable iron (oxFe) and oxalate-extractable aluminium (ox-Al), thereafter, relatively passive alternate P adsorption–desorption phenomena in soils may regulate the P dynamic of the water column.
associated Mg could play a greater role than previously thought in P sequestration or advection processes in wetland systems. Water soluble P of the surface soils
Fig. 3. Potential phosphorus flux rates to the water column from various dairy areas in different hydraulic-retention cycles. Table 2 Phosphorus flux from different dairy areas to the overlying water columns in four hydraulic retention cycles Field area
Area coverage (ha) P flux (kg ha1) % of TPa
Abandoned-intensive 112 Active-intensive 152 Forage/pasture 272 Native 64
26 31 10 6
10a 17a 6bc 3c
a Flux as percent of total phosphorus in the A-horizon soils, and values with the same letter in a column are not significantly different.
Native y = -0.47Ln(x) + 0.133 R2 = 0.88 EPCw = 1.3 mg L-1
Phosphorus retention/release (mg m-2)
2 1 0
0
4
8
2 12
0
-1
-2
-2
-4 Active-Intensive y = -2.17Ln(x) + 2.63 R2 = 0.77 EPCw = 3.4 mg L-1
12 6 0
0
48
Forage / Pasture y = -0.85Ln(x) + 0.520 R2 = 0.77 EPCw = 1.8 mg L-1
4
0
5 0
-6
-5
-12
-10
8
12
Abandoned-Intensive y = -1.99Ln(x) + 1.88 R2 = 0.91 EPCw = 2.6 mg L-1
10
12
4
0
4
8
12
Initial influent phosphorus (mg L-1) Fig. 4. Effects of initial influent phosphorus concentration on phosphorus flux to the water column from different dairy areas. Negative sign on Y-axis represents retention while positive one shows releases.
H.K. Pant, K.R. Reddy / Water Research 37 (2003) 965–972
971
Table 3 Multiple regression analysis of potential and cumulative phosphorus flux from soils to the water column in different hydraulicretention cycles, and selected independent variables Hydraulic-retention cycle
C0
Potential phosphorus flux to the water column 1st 0.08 2nd 0.14 3rd 0.20 4th 0.40
Fitted model equation P P P P
flux¼ 17:7 þ 0:01[WSP] flux¼ 9:5 þ 0:01[WSP] flux¼ 5:6 þ 0:01[WSP] flux¼ 3:8 þ 0:004[WSP]
Cumulative phosphorus flux to the water column 1st 0.08 CP flux¼ 580 þ 0:21[WSP] 2nd 0.14 CP flux¼ 228 þ 0:24[WSP] 3rd 0.20 CP flux¼ 185 þ 0:15[WSP]20:01[DA-Mg] 4th 0.40 CP flux¼ 136 þ 0:14[WSP]
R2 21 54 54 51
17 57 76 58
SE 20.8 11.4 7.7 6.9
729 327 138 187
DW 1.7 2.5 2.1 2.0
1.7 2.3 2.1 1.9
C0 : initial influent P concentration (mg L1), R2 : equivalent to variability explain (%), SE: standard error of estimation, DW: Durbin– Watson statistic (value >1.4 indicates no auto-correlation in residuals), P flux: potential phosphorus flux (mg m2 day–1), WSP: water soluble phosphorus (mg kg1), CP flux: cumulative phosphorus flux (total flux in 28-day cycle, mg m2), and DA-Mg: double acidextractable magnesium (mg kg1).
4. Conclusions This study indicated that highly soluble P accumulated from dairy operations may be released during first 28 days of flooding, but the slow dissolution of relatively stable pools of P (Fe-, Al-, Ca- and Mg-bound P) could maintain a long-term steady flux, well above that observed from non-impacted (native) areas. Moreover, the equilibrium P concentration of the water column of all the dairy areas were over 1.3 mg L1, which means the storm water treatment areas (STA) will not be able to remove any amounts of P as long as influent P level remains equal or below 1.3 mg L1. Since the targeted P level in the water column of the Lake Okeechobee is well below to 0.04 mg L1, it is virtually impossible to have any of the STA constructed in these lands for reduction of P influx to the lake. Thus, P flux potentials of soils should be given serious consideration prior to STA establishment in agricultural or manure-impacted lands, otherwise massive internal P loading could undermine the effective use of the wetlands. Preventative measures such as treating soils with chemical amendments, establishing vegetative communities or flushing the initially released P, however, may potentially stabilize the system, and maintain P removal efficiency.
Acknowledgements The authors acknowledge the help from J.E. Colson, I.C. Torres, H.M. Spencer, V.D. Nair, and D.A. Graetz, University of Florida, Soil and Water Science Department, Gainesville, Florida. This research was supported by the Florida Agricultural Experiment Station and a
grant from US Army Corps of Engineers, Florida, and approved for publication as Journal Series No. R-08183.
References [1] Newbold JD, Elwood JW, O’Neil RV, Sheldon AL. Phosphorus dynamics in a woodland stream ecosystem: a study of nutrient spiraling. Ecology 1983;64(5):1249–63. [2] Greenway M, Woolley A. Constructed wetlands in Queensland: performance efficiency and nutrient bioaccumulation. Ecol Eng 1999;12(1–2):39–55. [3] Diaz OA, Reddy KR, Moore PA. Solubility of inorganic P in stream water as influenced by pH and Ca concentration. Water Res 1994;28(8):1755–63. [4] Tanner CC, Sukias JPS, Upsdell MP. Relationships between loading rates and pollutant removal during maturation of gravel-bed constructed wetlands. J Environ Qual 1998;27(2):448–58. [5] Reddy KR, Smith WH. Aquatic plants for water treatment and resource recovery. Orlando, FL: Magnolia Publishing Inc., 1987. p. 1032. [6] Kadlec RH, Knight RL. Treatment wetlands. Ann Arbor, MI: Lewis Publishers, 1995. [7] Pant HK, Nair VD, Reddy KR, Graetz GA, Villapando RR. Influence of flooding on phosphorus solubility in manure-impacted soils. J Environ Qual 2002;4:1399–405. [8] Richardson CJ. Mechanisms controlling phosphorus retention capacity in freshwater wetlands. Science 1985;228(4706):1424–7. [9] Flaig EG, Reddy KR. Fate of phosphorus in the Lake Okeechobee watershed, Florida USA: overview and recommendations. Ecol Eng 1995;5(2+3):127–42. [10] Mehlich A. Determination of P, Ca, Mg, K, Na, and NH4. In: Soil testing div. pub., vols. 1–53. Raleigh, NC: North Carolina Dep. Agric., 1953.
972
H.K. Pant, K.R. Reddy / Water Research 37 (2003) 965–972
[11] Loeppert RH, Inskeep WP. Iron. In: DL Sparks et al., editors. Methods of Soil Analysis, Part 3. Madison, WI: Soil Science Society of America Inc., and American Society of Agronomy Inc., 1996. p. 639–64. [12] Andersen JM. An ignition method for determination of total phosphorus in lake sediments. Water Res 1976; 10(4):329–31. [13] EPA (US Environmental Protection Agency) Methods of chemical analysis of water and wastes. Cincinnati, OH: Environ. Monit. Support Lab., 1993. [14] Statistical Graphics. Statgraphics plus version 3.1. Rockville, MD: Statistical Graphics Corp, 1997. [15] Mayer LM, Gloss SP. Buffering of silica and phosphate in a turbid river. Limnol Oceanogr 1980;25(1):12–22.
[16] Froelich PN. Kinetic control of dissolved phosphate in natural rivers and estuaries: a primer on the phosphate buffer mechanism. Limnol Oceanogr 1988;33(4):649–68. [17] Pant HK, Reddy KR, Lemon E. Phosphorus retention capacity of root bed media of sub-surface flow constructed wetlands. Ecol Eng 2001a; 17(4):345–55. [18] Sanchez PA. Properties, management of soils in the tropics. NY: Wiley, 1976. [19] Barrow NJ. A mechanistic model for describing the sorption and desorption of phosphate by soil. J Soil Sci 1983;34(4):733–50. [20] Pant HK, Reddy KR. Phosphorus sorption characteristics of estuarine sediments under different redox conditions. J Environ Qual 2001;30(4):1474–80.