| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Dipartimento di Agronomia Ambientale e Produzioni Vegetali, Università di Padova, Via Romea 16, 35020 Legnaro (Pd), Italy
* Corresponding author (maurizio.borin{at}unipd.it)
Received for publication August 14, 2000.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSION AND CONCLUSIONSREFERENCES |
|---|
Abbreviations: C 0, treatment with a steady water table depth at soil level during the dormant season and fluctuating water table and possible flooding during the growing season on tufted sedge • ECw, electrical conductivity • M/S 60, treatment with a steady 60-cm water table depth on maize and sugarbeet • M/S 60–100, treatment with a water table fluctuating between 60 and 100 cm depth on maize and sugarbeet • M/Sfd, treatment with free drainage on maize and sugarbeet • P 10, treatment with a steady 10-cm water table depth during the dormant season and a fluctuating water table and possible flooding during the growing season on common reed • P 30, treatment with a steady 30-cm water table depth during the dormant season and a fluctuating water table and possible flooding during the growing season on common reed • T 0, treatment with a steady water table depth at soil level during the dormant season and a fluctuating water table and possible flooding during the growing season on common cattail • T 20, treatment with a steady 20-cm water table depth during the dormant season and a fluctuating water table and possible flooding during the growing season on common cattail • TDS, total dissolved solids • TKN, total Kjeldahl nitrogen
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSION AND CONCLUSIONSREFERENCES |
|---|
Controlled drainage + subirrigation was developed in the USA, where it has been widely applied in the last 20 yr because of its environmental benefits and increased yields (Skaggs and Brevé, 1995). One of the mechanisms by which controlled drainage reduces NO3–N losses is increased denitrification by maintaining a high water table, but the greatest reduction appears to be the result of a decrease in total drainage outflows (Gilliam et al., 1999). Subirrigation probably increases denitrification and crop N uptake, but also increases outflow volumes (Skaggs and Brevé, 1995).
Interest in the use of wetlands for wastewater treatment has grown very quickly during recent years. Few data are available on the treatment of agricultural runoff, but some experiments have shown its effectiveness (Chescheir et al., 1991a; Higgins et al., 1993; Kadlec and Knight, 1996). The primary mechanism for NO3–N removal is denitrification which, in the case of wetlands with surface flow, mainly occurs near the soil–water interface (Chescheir et al., 1991a).
The effectiveness of both practices is greatly affected by their design and management and by site conditions. Therefore, to maximize efficiency and avoid possible negative effects on ground water quality, local experimentation and good knowledge of the relations between the processes involved in pollutant removal and these practices are needed (Evans et al., 1990; Gilliam et al., 1996; Kadlec and Knight, 1996).
In the low-lying plain of the Veneto region in northeast Italy, subsurface drainage is very common and interest in this system is still increasing (Borin et al., 1997). Nitrate nitrogen losses are therefore of great concern, especially because these substances are released in sensitive areas such as the Venice Lagoon and, more in general, the coastal waters of the Adriatic (Consorzio Venezia Nuova, 1989; Giupponi and Rosato, 1999; Novotny and Olem, 1994).
Solutions such as controlled drainage + subirrigation and wetlands could be very effective in solving the problem of N pollution, but they are not as popular as in other countries. To obtain local information to encourage implementation of these practices by farmers and increase knowledge on the processes involved, extensive research has been carried out in recent years by the Padova University Department of Environmental Agronomy and Crop Production. Research included similar experiments running on different scales, from lysimeters to open field. Field activity, partially presented by Borin et al. (1998), gives information more related to real conditions and is suitable for producing guidelines for implementation, but can only consider a few variables. Lysimeters are therefore very useful tools for comparing a wider range of solutions, as in a previous experiment on controlled drainage + subirrigation (Borin and Lazzaro, 1995), and for taking more detailed measurements.
This paper presents results obtained in a 3-yr experiment on lysimeters. Some first-year results were previously presented by Borin and Bonaiti (1997). The main aim of the research was to obtain preliminary information on the effectiveness of natural systems in ameliorating water quality and reducing N losses by using various conditions of water table control. In general, increasing the proximity of the water table to the soil surface resulted in conditions favorable to enhanced denitrification. In the treatments where the water table was deeper than 60 cm, typical crops of the region were planted; in the others, wetland plants (hydrophytes; Mitsch and Richardson, 1993) were used. The results of these treatments were compared with one reference condition, in which the usual water management of crops in the open field was reproduced. For a better comparison among the treatments, fertilization was done by applying N to the hydrophytes in the same way as was done to the crops. We are aware that this method may represent an unusual way to add nutrients to a wetland, but doing so allowed us to better focus on the effects of water table control.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSION AND CONCLUSIONSREFERENCES |
|---|
Twenty drainage lysimeters, 1.5 m deep and with a surface area of 1 m2, were used to grow the study plants. Each lysimeter was connected by an underground pipe to an external tank to control the water table level by adding or removing water.
The upper 1.3 m of the lysimeters was filled with a sandy-silt-loam textured soil, homogeneous through the profile, characterized by an organic matter content of 1.1 to 1.2% and high carbonates (Table 1). This substrate was composed of 30% sand and 70% loamy-textured soil, which was collected from the surface of a fulvi-calcaric Cambisol soil according to the FAO–UNESCO classification. The bottom 0.2 m was filled with gravel about 2 cm in diameter, to allow drainage toward the pipe and avoid clogging. A PVC perforated pipe, 10 cm in diameter, was installed at the center of each lysimeter to measure water table depth and collect water samples.
|
|
Treatments and Experimental Management
Eight treatments were studied applying alternative water table managements to various crops or hydrophytes:
(i) Free drainage on maize and sugarbeet (M/Sfd), with three replicates. Maize was grown in 1996 and 1998, and sugarbeet in 1997. Irrigation was supplied during the very dry periods.
(ii) Water table fluctuating between 60 and 100 cm depth on maize and sugar beet (M/S 60–100), with three replicates. The fluctuating water table was reproduced by raising it back to 60 cm once it had dropped to 100 cm by evapotranspiration.
(iii) Steady water table depth of 60 cm on maize and sugarbeet (M/S 60), with three replicates.
(iv) Steady water table depth of 10 cm during the dormant season and fluctuating water table and possible flooding during the growing season on common reed (P 10), with three replicates. Flooding was due to the fact that no drainage was allowed during the growing season and consequently the water table could rise after heavy simulated rainfall.
(v) Steady water table depth of 30 cm during the dormant season and fluctuating water table and possible flooding during the growing season on common reed (P 30), with two replicates.
(vi) Steady water table depth at soil level during the dormant season and fluctuating water table and possible flooding during the growing season on common cattail (T 0), with three replicates.
(vii) Steady water table depth of 20 cm during the dormant season and fluctuating water table and possible flooding during the growing season on common cattail (T 20), with two replicates.
(viii) Steady water table depth at soil level during the dormant season and fluctuating water table and possible flooding during the growing season on tufted sedge (C 0), without replicates.
The lower number of replicates of treatments P 30, T 20, and C 0 was adopted to gain additional information on the behavior of common reed and common cattail and some first data on tufted sedge under different water regimes, given the lack of knowledge on these hydrophytes in our environment.
The M/Sfd treatement aimed at reproducing conditions similar to those common in cultivation in the Veneto region, where all water percolating below the root zone is considered lost and is drained from the system. The M/S 60–100 and M/S 60 treatments represent schemes of controlled drainage + subirrigation that can be adopted in field conditions. The other treatments represent wetland conditions where large amounts of water are allocated, obtaining an amelioration of its quality.
In all lysimeters, water was added once a day, producing temporary falls of the water table below the desired depth, especially in hot weather.
Maize and sugarbeet were cultivated as is usual in the region, but digging was used instead of plowing. All hydrophytes were planted in spring 1996, according to the indications of Hawke and José (1996): common cattail was planted as rhizome cuttings with one to three buds, common reed and tufted sedge were planted as clumps with some developed plants. The aboveground biomass was harvested at the end of fall each year.
Nitrogen fertilization of plants was done in all the lysimeters as follows: at the beginning of the experiment, in April 1996, 1500 g m-2 of poultry manure was applied, for a total amount of 60 g m-2 of organic N, before maize planting, and a further 2.3 g m-2 of mineral N was applied in June; in March 1997, 10 g m-2 of mineral N was applied before sugarbeet planting and 20 g m-2 of mineral N was applied in July; in May 1998, 15 g m-2 of mineral N was applied before maize planting and 15 g m-2 of mineral N was applied in June. This amount of poultry manure, which is higher than the conventional applications to crops, was applied to increase the organic matter and N content of the soil, which was quite poor at the start of the experiment. After this application, N fertilization was conducted according to crop requirements, always using urea.
All added and drained water was measured, and samples were collected monthly from the water table and daily from drainage water, when present. After freezing the samples, NO3–N concentrations, total Kjeldahl nitrogen (TKN), pH, and electrical conductivity (ECw) were determined in the laboratory. The TKN concentrations were only determined in the first year. The procedure of Cataldo et al. (1975) was used for measuring NO3–N, TKN was measured following the procedure indicated by the SISS (1985), and portable devices were used for pH and ECw, measured at 25°C. Assuming that water samples contained a mixture of salts and that the measured ECw was always <10000 µS cm-1, the following linear relationship (Shainberg and Oster, 1978) was used to calculate the amount of total dissolved solids (TDS) leached from the lysimeters using the measured ECw data:
 |
Statistical Analysis
Since data distributions of NO3–N concentrations, ECw, and pH were not normal, nonparametric statistical tests were adopted. Medians were used to describe the central trends of the data, and were compared by a median test. A log-probit transformation (Cavalli-Sforza, 1961) was carried out to analyze data dispersion with respect to medians. In this way, data distributions were linearized and a regression line was calculated and tested by a chi square test. The slope of the regression line was calculated for each treatment (the higher the slope, the lower the data dispersion around the median value) and compared with the others by analysis of variance and an SNK test. A dispersion index (C16–C84) was also calculated to show the range of concentrations containing two-thirds of the values. In the case of NO3–N concentrations, log-probit analysis also accounted for those samples in which concentrations were below the detection limit (Travis and Land, 1990).
Cumulative data on drainage volumes, N, and salt losses were compared using analysis of variance and Duncan's test.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSION AND CONCLUSIONSREFERENCES |
|---|
Nitrate Nitrogen and Total Kjeldahl Nitrogen Concentrations
Medians of NO3–N concentrations were significantly different (p = 0.05) among M/Sfd (33 mg L-1), M/S 60 (2.6 mg L-1), M/S 60–100 (1.6 mg L-1), and the wetland treatments (0.5–0.7 mg L-1).
The time pattern of values also differed among treatments (Fig. 2). In M/Sfd, the seasonal pattern was characterized by maxima in spring and minima at the end of fall. Values were all very high and single detectable data ranged between 8 and 77 mg L-1. Great variability with time was observed in M/S 60–100 and M/S 60, in which single detectable data (i.e., data coming from samples with concentrations above the detection limit) ranged between 0.1 and 36 mg L-1, with peaks in winter and minima at the end of fall. The highest concentrations were reached quickly and were followed by a slow drop during the rest of the year. Very different behavior was observed in wetland conditions, in which single detectable data ranged between 0.1 and 23 mg L-1 and no seasonal pattern was evident. In all treatments, high concentrations were observed especially at the beginning of the experiment, due to the initial very high fertilizer dose.
|
|
|
Total Kjeldahl N concentrations in water were quite low in all treatments, even after poultry manure application. Median values were 0 mg L-1 in M/Sfd and M/S 60–100, 0.002 mg L-1 in M/S 60, and 0.05 mg L-1 in wetland treatments (the latter merged). Taking all 174 samples analyzed for this parameter, 83% were lower than 1 mg L-1. In our experimental conditions, the low TKN concentrations in water may be explained by the following processes: (i) burial of organic N with organic matter produced, (ii) adsorption of ammonia into soil (clay), and (iii) enhanced nitrification in the oxygenated environment provided by the fluctuating water table (Novotny and Olem, 1994). Because of these low values, we decided to stop this analysis after 1 yr of monitoring and not to consider TKN concentrations when computing N losses from lysimeters.
Electrical Conductivity and pH
Median values of ECw were very low and of the same order of magnitude in all treatments except M/Sfd, which exceeded 1400 µS cm-1 (Table 3). A tendency to lower ECw values with a shallower water table was also observed and statistically confirmed. With the water table at the soil surface (T 0 and C 0), the data showed lower variability; M/Sfd and M/S 60–100 had higher variability. Considering all the lysimeters, maximum values ranged between 410 and 2170 µS cm-1.
|
The pH measured in water samples was of the same order of magnitude in all treatments, with medians close to 8 (Table 4). Nevertheless, the pH values in T 0 and P 10 treatments were significantly higher (p = 0.05) than in the crop treatments; the other wetland treatments had intermediate pH values.
|
Balance of Water Volumes, Nitrate Nitrogen, and Total Dissolved Solids
Water Volumes
Water was applied to simulate rainfall and to keep the water table at the desired depths by subirrigation. Total cumulated volumes are given in Fig. 4.
|
In the first growing season, common cattail required higher subirrigation volumes than the other two hydrophytes because it was the only plant that spread to occupy the entire available space. In the rest of the experimental period, common cattail and tufted sedge received about the same amounts of water, which were much less than those added to common reed. Within the same species, treatments with a shallower water table required more water in the period June through October.
Total water volumes added for the entire period were significantly different (p = 0.05) among all treatments (Table 5). Compared with the average volumes added to P treatments, T treatments received on average 73% of that water; C 0, 66%; M/S 60, 52%; M/S 60–100, 49%; and M/Sfd, 29%. Within the same species, treatments with deeper water tables required less water: P 30 required 93% of P 10, T 20 required 97% of T 0, and M/S 60–100 required 94% of M/S 60.
|
Within crop treatments, as observed for the volumes of water added, M/Sfd always had the lowest values. In M/S 60, continuous subirrigation caused higher drainage than in M/S 60–100, in which the possibility of the water table dropping to 100 cm created space to contain part of the next rain, thus reducing drainage. This process was more noticeable during the warm months, when evapotranspiration was higher.
Drainage within wetland treatments was very similar, and close to that of controlled drainage + subirrigation treatments. This may have been due to the low temperature typical of drainage periods, which dramatically reduced evapotranspiration and caused soil saturation in all treatments. In cold weather, the role of vegetation in consuming water became negligible and almost all rainfall was lost by drainage. However, two observations can be made for the hydrophytes. First, in both 1996 and 1997, drainage at the end of fall was lower in P treatments than in the others, especially in October and November 1997. This may be due to still active transpiration in common reed, which maintained green leaves until the beginning of winter. Second, during spring 1997 and 1998, treatments T 0 and C 0 had lower drainage values than the other treatments, especially in March 1997 and March–April 1998. We believe that the air temperature during these months (peaks between 16 and 23°C) allowed evaporation of water available at the soil surface of these two treatments. This may also explain why only within common cattail did the different water table management schemes affect drainage volumes. Overall, common cattail had the highest drainage volumes.
Total volumes drained over the entire period were significantly different (p = 0.05) within treatments, except among T 0, P 10, and P 30, which were also similar to those obtained from C 0 (Table 5). Compared with M/S 60, M/S 60–100 drained 14% less and M/Sfd 69% less.
In general, differences in total drained volumes among treatments were smaller than those observed for total water applied, due to the higher evapotranspiration of the hydrophytes, which received more water.
Nitrate Nitrogen
Nitrate nitrogen input with water added ranged between 3 g m-2 in M/Sfd and 11.8 g m-2 in P 10, that is, 2.5 to 10% of the total N added with fertilization (Table 5). Differences among treatments (significant at p = 0.05) were very similar to those discussed for the water volumes added, since the water used was the same for all lysimeters and its quality varied very little over the year.
Cumulative NO3–N losses with drainage water are shown in Fig. 5. Looking at the pattern for crop treatments, losses were higher in the first drainage season, mainly because of the higher NO3–N concentrations caused by the rich initial fertilizer applications.
|
The differences in losses between M/Sfd and controlled drainage + subirrigation treatments were greatest in spring, corresponding to maximum differences in NO3–N concentrations. Differences between M/S 60–100 and M/S 60 are explained by higher drainage and NO3–N concentration in M/S 60.
In the wetland treatments, NO3–N losses were lower than in the crop treatments, due to very low concentrations. Since NO3–N concentrations in these lysimeters varied only slightly over time, the pattern was very close to that for drainage volumes.
Total NO3–N losses were strongly influenced by treatments (differences significant at p = 0.05), showing the following ranking: conventional drainage > controlled drainage + subirrigation > wetlands (Table 5). Compared with M/Sfd, NO3–N losses were reduced by 46% in M/S 60, 63% in M/S 60–100, and 95% in the average of wetland treatments.
With respect to total N input, NO3–N losses with drainage were about 9% in M/Sfd, 3 to 5% in controlled drainage + subirrigation treatments, and 0.4% in the average of wetland treatments. This assumes particular importance considering that treatments with lower losses were those that received the higher total N input (fertilization + irrigation water).
Total Dissolved Solids
Although TDS data were estimated from ECw, we believe that the cumulative values obtained at the end of the experiment are interesting for an indicative comparison of treatments. A significant difference (p = 0.05) was found between the average total losses of crop treatments (253 g m-2) and those of wetland treatments (175 g m-2).
|
DISCUSSION AND CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSION AND CONCLUSIONSREFERENCES |
|---|
In spite of the controlled environment, drainage measured in the lysimeters with conventional drainage reproduced field conditions quite well regarding total discharged volumes and time patterns of discharge, confirming that drainage occurs mainly at the end of winter and early spring (Borin et al., 1998). Instead, the order of magnitude of NO3–N concentrations in the lysimeters was slightly higher than that usually monitored in field drainage water. As a consequence, measured losses may have overestimated real losses in open fields. This may be explained with the high fertilization applied to the lysimeters. Nevertheless, the ratio of N losses in drained water to N input, which was lower than 10%, was similar to that observed in the field. For these reasons, we believe that our results may be extrapolated to open-field conditions.
Even though conventional drainage had high overall water use efficiency (ratio of evapotranspiration to total water added around 88%), it had a higher risk from the standpoint of ground water contamination, because NO3–N concentrations in water were consistently and dramatically higher than 11 ppm, which is the European Union safe drinking water standard. In spite of the very high N removal, total losses were 11.1 g m-2.
When compared with conventional drainage, the controlled drainage + subirrigation treatments discharged higher volumes of water because of the subirrigation volumes applied, but had higher evapotranspiration rates due to the continuous presence of available water. The overall water use efficiency, on the average of the two treatments, was 79%. In spite of the higher drainage, NO3–N losses were almost halved due to the great reduction of NO3–N water concentrations. This reduction was probably due to various processes. One of these could be the enhanced evapotranspiration, which increased plant absorption, as suggested by the fact that, at the end of the cycle, the plant biomass and N uptake were around 20% higher than plants under conventional drainage (data not presented). As a consequence it can be expected that there was a lower amount of NO3–N in the soil at the end of the growing season in the controlled drainage + subirrigation lysimeters, which could be lost by percolation during the following winter. Another reason for reduced concentration in water could be related to the anaerobic conditions due to the presence of a water table, which probably limited the production of nitrates from organic N and caused denitrification. Controlled drainage + subirrigation with a changing water table depth reduced N losses by 63%, whereas control of the water table at a constant depth of 60 cm, which is more difficult to achieve in field conditions, reduced them by 46%. The better performances of the fluctuating water table were mainly due to the reduction of drainage. In the practical implementation of controlled drainage + subirrigation, the amount of water required to raise the water table may be satisfied by using low-quality water, contributing toward solving the problem of its allocation and treatment.
Wetland treatments discharged volumes of the same order of magnitude as the controlled drainage + subirrigation treatments but, since they received even more irrigation water, they had even higher evapotranspiration rates, with an average overall water use efficiency of 90%. Nitrogen loads were further reduced (about 95% of the conventional drainage treatment). In fact, NO3–N concentrations in water were extremely low, without differences related to hydrophytes and water table management. It is likely that the anaerobic conditions created in these lysimeters involved the entire mass of the soil with the effect of consuming the available nitrates almost completely. Plant uptake also played an important role, with about 10% more uptake than plants with conventional drainage (data not presented). As previously noted, N loads to hydrophytes were managed in the same way as for crops, so that this result cannot be directly assimilated to a wetland receiving agricultural waters, because N loads from agricultural drainage are more diluted over time (Chescheir et al., 1991a; Borin et al., 2000). Nevertheless, considering the entire period (31 mo), our wetland assimilated about 1330 kg ha-1 of N, with no sign of "breakthrough". As a consequence, an assimilation rate of at least 500 kg ha-1 yr-1 could be expected. Given that nitrate N losses for the M/Sfd treatment were roughly 40 kg ha-1 yr-1, we could conservatively predict that a wetland one-tenth the size of a conventionally drained field (e.g., M/Sfd) could assimilate the nitrate N leached from the field, if the drainage water could be collected and routed to the surface. Since about 10% of the water added to M/Sfd as simulated rainfall was leached, applying leached water from 10 ha of cropland to 1 ha of wetland would roughly equal the amount of water that would have fallen on the wetland by natural rainfall, thereby doubling the water input to the wetland, while adding 400 kg NO3–N yr-1. In addition, the fact that even after heavy loads (up to 20 g m-2 of mineral N or 60 g m-2 of organic N) there were very low losses suggests that the nitrate removal rate can be higher, as found by Horne et al. (1995). The very good results obtained with wetland treatments in these particular conditions indicate the great potential of wetlands for treating nutrient loads from agricultural drainage water and, eventually, for acting as sites where drainage peaks and urban storm water could be allocated. In these applications, care should be put on the management of winter input water volumes, when wetland evapotranspiration is low. In this case the possibility of flooding the surface, which we did not do to prevent ice damage to the few plants, might be considered. Finally, the particular hydrology of our systems must be stressed: even though the movement of water is vertical, conditions of almost permanent soil saturation were produced, which are typical of horizontal flow wetlands.
Lastly, as also found by other researchers (e.g., Bavor et al., 1988; Chescheir et al., 1991b), wetland treatments reduced the ECw of drainage water and consequently the losses of total dissolved solids. This reduction, observed also in the treatments with controlled drainage + subirrigation, could be explained only partially with the greater flushing through the lysimeters. In fact, with respect to the control, in wetland treatments volumes of water almost three times higher were added, but the reduction of ECw was almost five times. Similarly, in controlled drainage + subirrigation treatments the reduction of ECw (about four times) was higher than the increase of added volumes (around twice).
The use of lysimeters allowed us to quantify the order of magnitude of the NO3–N and salt balance in waters for the different treatments of water table management considered. In particular, we emphasized the differences that alternative solutions of controlled drainage + subirrigation and wetlands can cause to the reduction of N losses, discussing the potential and limits of implementation.
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSION AND CONCLUSIONSREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| 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 |
