Denitrification rates in relation to groundwater level in a peat soil under grassland
- First Online:
- Cite this article as:
- van Beek, C.L., Hummelink, E.W.J., Velthof, G. et al. Biol Fertil Soils (2004) 39: 329. doi:10.1007/s00374-003-0685-3
- 197 Views
In this study spatial and temporal relations between denitrification rates and groundwater levels were assessed for intensively managed grassland on peat soil where groundwater levels fluctuated between 0 and 1 m below the soil surface. Denitrification rates were measured every 3–4 weeks using the C2H2 inhibition technique for 2 years (2000–2002). Soil samples were taken every 10 cm until the groundwater level was reached. Annual N losses through denitrification averaged 87 kg N ha-1 of which almost 70% originated from soil layers deeper than 20 cm below the soil surface. N losses through denitrification accounted for 16% of the N surplus at farm-level (including mineralization of peat), making it a key-process for the N efficiency of the present dairy farm. Potential denitrification rates exceeded actual denitrification rates at all depths, indicating that organic C was not limiting actual denitrification rates in this soil. The groundwater level appeared to determine the distribution of denitrification rates with depth. Our results were explained by the ample availability of an energy source (degradable C) throughout the soil profile of the peat soil.
KeywordsDenitrificationGroundwater levelPeatGrasslandNitrous oxide
Measured N losses through denitrification in peat soils appear to be of the same magnitude as N losses through denitrification in other soil types (De Klein and Logtestijn 1994; Barton et al. 1999). This is remarkable since circumstances for denitrification seem more favourable in peat soils than in other soils, considering the often anaerobic and organic matter-rich conditions. Most estimates of N losses through denitrification on different soil types, including peat soils, are based on measurements in the topsoil only (0–20 cm). The common justification for this restriction is that the main prerequisites for denitrification, i.e. the presence of NO3-, degradable C and anaerobic conditions, only occur concurrently in the topsoil (Yeomans et al. 1992; Luo et al. 1998). Yet, in peat soils high contents of degradable C are also present in the subsoil (Jörgensen and Richter 1992; Velthof and Oenema 1995) and therefore, a considerable contribution of N losses through denitrification from the subsoil can be expected in peat soils when NO3- is present under anaerobic conditions.
In agricultural soils, NO3- may directly originate from fertilizer application and, after nitrification, from animal manure, atmospheric deposition, biological N2 fixation and soil organic N. Nitrification and denitrification are oxidation-reduction processes and, in this sequence, are optimal under conditions going from high to low redox potential (Stumm and Morgan 1996). In soils with shallow groundwater levels (0–100 cm below the soil surface) groundwater fluctuations are a driving factor for changes in the redox potential. In these soils, groundwater level fluctuations are expected to play a key role in creating conditions where nitrification and denitrification can occur at relative proximity. Hence, we expect that temporal changes in the groundwater level are reflected in temporal changes in N losses through denitrification in soils with a shallow groundwater level.
In The Netherlands many peat soils are located in polders, i.e. at an elevation below mean sea level with a strictly regulated management of surface waters in ditches and canals. In summer, when evapotranspiration exceeds precipitation, the groundwater level drops below the ditch water level, and the groundwater level is deeper in the middle of the fields than near the ditches surrounding the fields. In winter, although the aim is to achieve a lower ditch water level, the soil water storage capacity is frequently exceeded and then fields become waterlogged. Hence, both in summer and in winter there is a gradient of groundwater level with distance from the surface water. Velthof et al. (1996) showed that this spatial gradient of groundwater level with distance to the ditch water was reflected in the spatial pattern of N2O emissions from an intensively managed peat soil in summer, i.e. higher N2O emissions were found near the ditch. Generally, N2O emissions and denitrification rates are related (Velthof et al. 1996; Stevens et al. 1997) and therefore, under comparable conditions, we expect similar spatial patterns for actual denitrification rates in the topsoil.
So far only a few estimates of annual denitrification losses on peat soils have been reported, whilst from the abovementioned studies it follows that mechanisms and magnitude of denitrification losses in peat soils may deviate considerably from those of other soil types. Therefore, the following three hypotheses were tested for an intensively managed grassland on peat soil: (1) the subsoil (>20 cm) makes a considerable contribution to the total N losses through denitrification, (2) temporal changes in the groundwater level are reflected in temporal changes in N losses through denitrification, and (3) spatial patterns of the groundwater level within the field are reflected in spatial patterns of denitrification rates in the topsoil (0–10 cm). In addition, N2O production rates were calculated in order to provide an estimate of total N2(O) losses from an intensively managed grassland on peat soil.
Materials and methods
Selected soil properties of the experimental field. ND Not determined
Soil layer (cm)
Clay (<2 μm)
Silt (2–50 μm)
Sand (>50 μm)
Loss on ignition
Dry bulk density
We defined potential denitrification as the denitrification rate under anaerobic conditions with abundant NO3- at 20°C, following Bijay-Singh et al. (1989) and Aulakh et al. (1992). The potential denitrification is a measure of the organic C content available to denitrifying organisms (Jörgensen and Richter 1992) and was determined once in May 2000. Samples were taken with a ring bore equipped with 100-cm3 stainless steel ring samplers (5 cm diameter). Samples were taken in triplicate and were then bulked to give a composite sample. From the topsoil (0–10 cm) 16 composite samples were collected in a regular grid of 10×100 m. Fewer samples were taken from deeper soil layers because we expected more heterogeneity in the topsoil than in the subsoil. From the soil layers deeper than 10 cm below the soil surface, two composite samples were taken per 10 cm soil layer until the groundwater level was reached (70 cm below the soil surface). The composite samples were placed in PVC containers (894 ml) equipped with a screw cap with rubber sealing rings and rubber septa. Approximately 0.1 mg N g-1 dry soil as KNO3 was applied to each container. To remove O2 from the containers, the containers were pumped under vacuum for 1 min and flushed with N2 for 5 min.
Measurements of actual denitrification started in September 2000 and lasted until August 2002. Soil samples were collected every 3– 4 weeks with a ring bore yielding 24 sampling events in total. During the first year of measurements, our emphasis was on spatial relationships between the groundwater level and denitrification rates in the topsoil. However, after we found that the subsoil was a major contributor to the N losses through denitrification the emphasis shifted more to the subsoil in the second year. In the first year the sampling procedure for actual denitrification measurements was the same as that for potential denitrification measurements. Note that the number of sampled layers fluctuated with the groundwater level resulting in a minimum of one sampled layer (occasionally in winter) to a maximum of nine sampled layers (in summer). In the second year the field was split into eight plots of 20×100 m and one ring sample was taken per soil layer of 10 cm from every plot until the groundwater level was reached. Each ring sample was incubated individually. Additionally, from February to August 2002 (eight sampling events) separate sampling rings were collected in triplicate from each 10-cm soil layer to determine N2O production rates. On all occasions samples were taken until the groundwater level was reached, because we were not able to take undisturbed soil samples below the groundwater level. In both seasons samples for actual denitrification measurements and samples for N2O production measurements were incubated at the prevailing air temperature.
where, D is the N2O-N production in an incubation container (g ha-1 day-1) per 10-cm soil layer, ρ N2O-N is the density of N in N2O, which was linearly dependent on temperature (range 1.15–1.25 g l-1), Δ N2O is the maximum N2O production between two measurements corrected for the internal volume of the TGA, V g is the gaseous volume of the incubation containers corrected for the sample volume (l), V m is the soil volume in the incubation pots (l), α is the Bunsen adsorption coefficient used to correct for the amount of N2O dissolved in water (Moraghan and Buresh 1977), θ is the volumetric water content, A is the surface area of the samples (m2), Δt is the time difference between two measurements (days) and 2×104 a conversion factor (ha m-2) to convert from g N per m2 per 5-cm soil layer to g N ha-1 per 10-cm soil layer. Note that in this paper denitrification rates imply actual denitrification rates, unless stated otherwise.
Soil and water analyses
The samples in the rings containing exactly 100 ml undisturbed soil, were weighed, dried at 105°C for 24 h and re-weighed. Dry bulk densities were calculated per sample from the volumetric water content. In the 2001–2002 season, each time denitrification was measured separate samples were taken to determine NO3- and NH4+ contents (three replicates per soil layer of 10 cm). Twenty grams of each replicate of fresh soil was transferred to a flask and 50 ml of 1 M KCl was added. The flasks were shaken for 1 h, the contents filtered and NO3- and NH4+ concentrations analysed by segmented flow analysis (Houba et al. 1995).
Groundwater tubes were installed at 0, 1, 5, 10, 20 and 22 m from the ditch, where 0 m equals the ditch water at one end of the transect and 20 m equals the middle of the field. The ditch water level at the other end of the transect (i.e. at 40 m) was set equal to the ditch water level at 0 m, because the ditches surrounding the field were connected. Groundwater levels were measured biweekly. In addition, an automatic groundwater probe was placed at 7 m from the ditch, which recorded groundwater levels every 5 h.
For each sampling event, denitrification rates were integrated over depth, resulting in predicted total N losses through denitrification. Subsequently, annual N losses through denitrification were calculated by integrating N losses through denitrification over time. Because of the different sampling strategies for the 2000–2001 and the 2001–2002 seasons, separate calculation procedures were followed for each season. Briefly, for the 2000–2001 season average denitrification rates per soil layer were summed per sampling event and then linearly integrated over time. For the 2001–2002 season, annual N losses through denitrification per plot were calculated and then averaged. SEs were calculated from replicates and annual N losses through denitrification were calculated with a 95% confidence level. Consistency of spatial patterns of denitrification rates in the topsoil (0–10 cm) was tested by calculating Spearman rank correlation coefficients (r s) of successive measurements for 16 (2000–2001) or eight (2001–2002) plots (Davis 1986). Also, the depths at which highest denitrification rates were found per sampling event were compared with the prevailing groundwater level.
Annual N losses through denitrification
NO3- and NH4+ contents and groundwater level
NO3- contents of the soil fluctuated between 1 and 30 mg N kg-1, with higher NO3- contents during the growing season. Sometimes decreasing NO3- contents with depth were found, but there was no consistent pattern (Fig. 3). NH4+ contents of the soil occasionally showed a strong increase with depth (e.g. May, 2001), but usually were rather constant with depth and were higher in summer than in winter (Fig. 3).
Spatial patterns of denitrification in the topsoil
For most situations the Spearman Rank correlation coefficient │rs│ between successive measurements in the topsoil was <0.4, indicating that the temporal stability of spatial patterns was poor (data not shown). Furthermore, the hypothesis that spatial variations in groundwater levels are reflected in spatial patterns of denitrification rates in the topsoil was tested. In the 2000–2001 season, for each sampling event per grid distance perpendicular to the ditch, average denitrification rates were calculated (n =4) and compared with measurements of groundwater level at seven distances from the ditch. The spatial dependency of denitrification rates in the topsoil with distance from the ditch water showed contrasting results. For some measurements (e.g. 10 October 2000 and 20 February 2001) denitrification and groundwater level followed a comparable spatial pattern, but for other events such relations were absent (Fig. 4).
Groundwater level and depth of maximum denitrification
Mean annual N losses through denitrification equalled 87 kg N ha-1. In the 2000–2001 season, the N losses through denitrification of two sampling events accounted for 78% of the total annual N loss through denitrification (Fig. 2). There were no extraordinary circumstances during these events; probably a combination of favourable meteorological, agricultural and hydrological conditions occurred. In comparison to other soil types, the measured N losses through denitrification found in this study are high (Barton et al. 1999), and are considerable in terms of N inputs and N surpluses at farm level. N surpluses at farm level consisting of inputs via fertilizers and feed and outputs via sale of milk, manure and cattle, averaged 280 kg N ha-1 year-1 of which 220 kg N ha-1 year-1 was applied as mineral fertilizer (Van Beek et al., in press). Moreover, annually 263 kg N ha-1 was released by mineralization of peat (C.L. van Beek, unpublished data), thus N losses through denitrification accounted for 16% of the N surplus at the farm-level and for 40% of the mineral fertilizer application.
N2O production rates, which included N2O produced by nitrification and N2O produced by denitrification, sometimes exceeded denitrification rates (Fig. 2) and yielded in total 126 kg N ha-1 year-1. N2O produced by denitrification and N2O produced by nitrification cannot be separated easily (Wrage 2003; Aulakh et al. 1992), but the high N2O production in comparison with denitrification indicates that N2O produced during nitrification was an important source of N2O emission. The N emission measured with and without C2H2 partly overlapped (viz the N2O produced during denitrification) and consequently we could not calculate total N2+N2O emission by nitrification and denitrification. Instead, minimum and maximum ranges of total N2 + N2O emission were calculated by assuming all N2O production originated from denitrification and by assuming all N2O production originated from nitrification, respectively. This estimate yielded 126–213 kg N ha-1 year-1.
In the literature, only a few estimates of annual N losses through denitrification in managed grasslands on peat soils have been reported. De Klein and Logtestijn (1994) and Berendse et al. (1994) reported N losses through denitrification for grassland on peat soils of 4–16 kg N ha-1 year-1 and 17 kg N ha-1 year-1, respectively, and in both studies measurements were limited to the topsoil (0–10 cm below the surface). In the present study annually about 12 kg N ha-1 originated from the top 10 cm, which is consistent with the results of De Klein and Logtestijn (1994) and Berendse et al. (1994). However, De Klein and Van Logtestijn (1994) and Berendse et al. (1994) may have considerably underestimated total denitrification rates, because we observed here that 69% of the annual N loss through denitrification originated from soil layers deeper than 20 cm (e.g. Fig. 3). To our knowledge, the only other denitrification study on managed peat soil at greater depth (0–40 cm) was performed by Koops et al. (1996), who reported N losses through denitrification from intensively managed grasslands of 70 kg N ha-1 year-1, which agrees reasonably well with our results (in the present study 58 kg N ha-1 year-1 originated from the upper 40 cm). Thus, considerable N losses through denitrification may originate from the subsoil (>20 cm) of grassland on peat soil.
In the present study, samples were taken up to groundwater level, because we were not able to take undisturbed soil samples from saturated peat. However, there are indications that denitrification rates may be significant also below the groundwater level (Well et al. 2001). We assumed that denitrification rates below the groundwater level were small at our site, because of the decreasing NO3- contents with depth (Fig. 3). To check this assumption, NO3- removal from the soil was calculated by summing reductions in NO3- contents between two successive measurements (e.g. Fig. 2). Annual NO3- removal from the unsaturated zone yielded 135 kg NO3--N ha-1, which included crop uptake, leaching and denitrification. Because N2+N2O losses could completely account for the NO3- removal from the unsaturated zone, we believe that we did not miss large peaks in denitrification.
N losses through denitrification appeared to be largely controlled by NO3- contents of the soil profile (Fig. 5). During one event (19 June 2001) NO3- contents were exceptionally high in the topsoil (Fig. 3) as a result of fertilization, and did not coincide with high N losses through denitrification (Fig. 5, indicated by an arrow). At the same time, the groundwater level was relatively deep (61 cm below the soil surface, e.g. Fig. 2). Apparently, the depth where favourable conditions occurred for denitrification (i.e. close to the groundwater level) did not correspond with the depth with the highest NO3- contents. Hence, relatively large (vertical) distances between the groundwater level and NO3 caused results of denitrification in relation to NO3- contents of the soil to deviate. Vice versa, relatively high N losses through denitrification are expected when relatively high NO3- contents (e.g. after fertilization) coincide with shallow groundwater levels.
In low-land areas, groundwater levels tend to determine the magnitude of N losses via denitrification. In these areas, rising groundwater levels often coincide with increasing denitrification rates (Steenvoorden 1983; Kliewer and Gilliam 1995; Abdirashid et al. 2002). However, Koops et al. (1996) found no significant differences in N losses through denitrification for two intensively managed grasslands on peat soil with different mean groundwater levels. In the present study no clear relation between temporal changes in groundwater level and N losses through denitrification was found (Fig. 5). Highest denitrification rates were predominantly found 0–40 cm above the groundwater level (Fig. 6). Just above the groundwater level, aerobic and anaerobic niches may occur in relative proximity (Grundmann et al. 1995; Abbasi and Adams 2000) resulting in optimal conditions for simultaneous nitrification and denitrification. Consequently, in shallow soils (like the present peat soil) temporal changes in groundwater level are expressed in changes in the depth where denitrification occurs, and to a lesser extent in changes in the magnitude of N losses through denitrification, providing that degradable C is not limiting denitrification. In the present study, potential denitrification rates (as a measure of degradable C availability) were always more than twice the actual denitrification rate at all depths, indicating that actual denitrification rates were not limited by C throughout the soil profile (Figs. 1, 3). Hence, as the depth at which the denitrification rates were highest approximately followed the groundwater level, the magnitude of N losses through denitrification was largely controlled by the NO3- contents of the soil (Fig. 5).
During some events, spatial patterns of denitrification rates in the topsoil were related to spatial patterns of groundwater level, but these relationships were not consistent in time (Fig. 4). In the literature, contrasting effects of distance to surface water on denitrification rates in the topsoil are reported. Schnabel et al. (1996) and Clément et al. (2002) reported decreasing denitrification rates with distance from surface water, while Pinay et al. (1993), Schipper et al. (1993) and Lowrance (1992) found increasing denitrification rates with distance from surface water. Clément et al. (2002) showed that spatial patterns of denitrification rates with distance to surface water were often weak (i.e. not significant) and were not consistent for different vegetation covers and time periods. The absence of a consistent relation between denitrification rates in the topsoil and distance to surface water was ascribed to the high intersite variability. The present study and the study of Clément et al. (2002) indicate that spatial patterns of denitrification are not consistent in time. Therefore, conclusions about spatial relations between denitrification rates and groundwater level should be temporally validated, before being regarded as general patterns.
In conclusion, annual N losses through denitrification from an intensively managed grassland on peat soil yielded 87 kg N ha-1, which equalled about 40% of the fertilizer N input and about 16% of the N surplus at the farm level, including mineralization of peat. Almost 70% of the total N loss through denitrification originated from soil layers deeper than 20 cm below the soil surface, showing that the subsoil should be taken into account when measuring N losses through denitrification from grassland on peat soils. Because degradable organic C was not limiting actual denitrification at any depth in our soil, favourable conditions for denitrification could develop throughout the soil profile. As a consequence, NO3- contents of the soil largely governed the magnitude of N losses through denitrification, while groundwater level acted as a mechanism determining the depth where denitrification occurred.
We thank Marius Heinen of Alterra for his assistance in data interpretation and for his comments on former drafts of this paper. Furthermore, we thank Gé van den Eertwegh of the Rijnland Water Board for cooperation in the project. This study was part of the DOVE-peat project and was funded by the Dutch Ministry of Agriculture, Nature and Food Quality (LNV programme 398-II).