Module 3 Land-use and climate change impacts on organic soils and relationships to GHG emissions: new measurements
Study sites: The three sites were chosen to provide typical upland organic soil and land-use conditions but contrasting climates and N deposition rates (Table 3.1). The sites cover the proposed situations of warm & wet with two rates of atmospheric N deposition (Plynlimon - Hafren, Wales), cold and wet, with small rates of N deposition (Ullapool) and cold and dry climate with small N deposition rates (Glensaugh). At Glensaugh and Hafren the sites were the same as chosen for the cores taken for the 13C pulse labelling experiment, and first flux measurements were made within one month of the pulse labelling experiment. In addition at the Welsh site we intended to study the influence of N deposition and land use. For this purpose an unimproved land use on a wet peaty gley was chosen at altitude of 380 m asl to be compared with the similar site. Unimproved grassland, at 519 m. These sites should receive different rates of N deposition, due to the more frequent mist development at higher altitude. In addition a small forest plantation in the same valley as the improved and unimproved grassland areas and less than 100 m below the improved grassland site was selected. Deposition rates to a forest can be 2 to 3 fold larger than to shorter vegetation (Fowler et al., 2004).
Table 3.1 The study sites
1 N deposition was estimated from the CEH N deposition map and the presence of trees (Fowler et al., 2004)
Flux chambers were installed in summer/autumn 2004 and remained in position until autumn 2005. Fluxes of N 2O, CH 4 and soil respiration were measured on 7 occasions (6 at one site: Jun/Jul 04 (not Ullapool), Oct/Nov 04, Jan 05, Apr 05, Jun/Jul 05, Aug/Sep 05, Nov 05). Each time air samples were taken from the static chambers, which involved closing the chambers with a plastic dome shaped lid and taking samples after one-hour incubation. Samples were collected by syringe and transferred to Tedlar bags. Samples were analysed at CEH Edinburgh for CH 4, and N 2O by gas chromatography using a flame ionisation detector for the analysis of CH 4 and an electron capture detector for N 2O (Skiba et al., 2006). Soil respiration was measured in situ using a PP Systems EGM soil respirometer, which was placed on plots were the vegetation was cut back and on soil devoid of vegetation. These latter plots were deliberately kept bare by weeding after each measurement as preparation for the subsequent measurements a few months later. Regrowth during this period was minimal. In addition respiration from vegetated plots were measured, but not included in the data analysis. At each visit the soil temperature (at 10 cm depth) and soil surface temperature is measured. Soil cores (0-20 cm depth) were taken for subsequent analysis of soil moisture content and KCl extractable NH 4+ and NO 3- concentrations (n=3). On one occasion soil samples were taken at all sites for measurements of bulk density (n = 5), soil pH and total carbon and total nitrogen content (three sub samples from 5 soil cores 0-20 cm mixed well). Species composition in each chamber was recorded at every site visit.
3.3.1 Soil properties
Soil physical and chemical properties are summarised in Table 3.2 & 3.3. All parameters were in the expected range reported elsewhere for organic soils. The bulk density of the peat ranged from 0.08 to 0.14 g cm -3 and did not vary significantly between geographical location, and at Ullapool also not between sites. At Glensaugh and Hafren the peat bulk density was significantly smaller than for the peaty gleys and podzols. Largest bulk densities were measured for the Hafren peaty gleys (> 0.4 g cm -3). Total N, pH and bulk density increased over the gradient peat to unimproved peaty gley/podzol to improved peaty gley/ podzol. Soil C content however, decreased along this gradient at Hafren and Glensaugh, and did not change at Ullapool. Soil moisture contents were high throughout the study period, at Glensaugh and Ullapool the peat was always wetter than the peaty soils, however such difference was not observed at Hafren. Soil temperatures ranged from 3 - 15, 1 - 13 and 4 - 14 oC at Ullapool, Glensaugh and Hafren, respectively, air temperatures are shown in Table 3.2.
The KCl extractable NH 4+ concentrations ranged from 23 to 37 mg N g -1soil in Scotland, but were smaller in Wales (11 - 26 mg N g -1) and correlated with the soil extractable NH 4+ fraction (r 2=0.75) (Table 3.3 and Figure 3.1). Nitrate concentrations were below the detectable range of the analyser for all sites, suggesting very small, if any, nitrification rates.
Table 3.2 Soil properties and air temperature at the study sites, n = 3 for total C &N and pH, n=5 for BD, n=3 replica on 7 dates for temperature and moisture
Table 3.3 Available soil NH 4+ and greenhouse gas fluxes. Average of 3 replica measurements on 7 dates
1 Median values are reported for N 2O and CH 4 fluxes, as these are very different from the average values
Figure 3.1 Relationship between soil available NH 4 and total C content in organic soils
3.3.2 Soil respiration, CH 4 and N 2O fluxes
Soil CO 2 (respiration), methane and nitrous oxide fluxes are shown for each site in Table 3.3. Temporal and spatial variations were typically large for all gases measured, as shown here for the CH 4 fluxes from peat at the three study sites and the N 2O fluxes from the fertilised improved grassland soils (Figure 3.2).
Figure 3.2 Temporal variations of N 2O fluxes from the improved peaty podzol/gleys (left graph) and of CH 4 from unimproved peat (right graph). Data are averages from 4 chambers and standard error of the mean. The SE is occasionally too small to be visible.
Nitrous oxide fluxes: All sites showed periods of N 2O emission and uptake (Table 3.3, Figure 3.2). Largest and most frequent N 2O emissions were measured from the improved peaty podzol at Glensaugh, the wood in Hafren and the unimproved peaty gley at the foot of the hills at Hafren. Uptake of N 2O occurred most often at Hafren on the peat, unimproved peaty gley, improved peaty gley. Contrary the peat at Ullapool emitted N 2O when the peaty gleys here absorbed N 2O. It was not possible to explain the switch from N 2O emission to N 2O uptake on hand of the variables measured here.
The large spatial and temporal variations in N 2O flux made it very difficult to provide correlations that would be useful to the ECOSSE model for either, the entire data set, or subsets of measurements at the same site, or for the same soil type. Average N 2O fluxes and log transformations of the fluxes only correlated well with total N and soil respiration (Figure 3.3).
Figure 3.3 Relationship of N 2O flux with total soil N (left graph) and soil respiration (right graph). Data are averages of 4 measurements on 7 occasions. In order to transform the negative N 2O fluxes, all data were increased to positive values by adding 0.15.
The range of fluxes measured in this study is similar to those measured from organic soils and similar landuse elsewhere in Britain by CEH. The 62 measurements (average values various time periods) to date (including this study) suggest an average N 2O emission of 0.25 kg N 2O-N/ha/y, ranging from -0.99 to 3.7 kg N 2O-N/ha/y (n= 62).
N deposition and nitrous oxide fluxes: The two 'extra' sites at Hafren (unimproved peaty gley at the bottom of the hill and the forest) were chosen to establish a relationship between atmospheric N deposition and N 2O emission. The forest site is adjacent to the improved grassland, both are on a slope but differ in N deposition rate, due to the scavenging effect of the forest (Table 3.1). Forest total C&N, soil available NH 4, N 2O and soil respiration rates are all larger compared to the adjacent improved grassland. Due to large spatial and temporal variability these differences are not significant. It is not possible to blame the enhanced deposition rate to the forest entirely for these increases in N & C accumulation and fluxes. The wood is the only shelter for sheep on this exposed hill site, and evidence of livestock excreta as a source of N input is plentiful.
Comparison of the unimproved peaty gleys in a hollow at the bottom of the hills, surrounded by forests, with the unimproved peaty gley on a steep slope at higher altitude, where N deposition rates are larger due to enhanced occult deposition, provided not the expected results. Total N, available N, soil respiration and N 2O fluxes were larger at the low N deposition site. The explanation for this is possible runoff of N from the above forests.
Soil respiration: Soil respiration rates were roughly of the same order at all three locations, however 4 times smaller than average annual soil respiration rates measured from improved, but unfertilised grassland on mineral soil in SE Scotland (Jones et al., 2006). There were not always four replicates per site, due to problems with the analyser and availability of bare, vegetation free patches of peat needed to make the measurements. Soil respiration rates correlated significantly with the N 2O flux (Figure 3.3) using average data for each site and also with total C& N content and the bulk density of the soil (Figure 3.4).
Figure 3.4 The dependence of soil respiration on soil N,C and bulk density, calculated by multivariate regression analysis. The data were fitted by polynomial equation shown above the graph. Data points are averages of 4 measurements .
At individual sites, good correlations between soil and plant respiration and soil temperature and soil moisture was observed at Glensaugh (ln CO 2 (plant&soil) = -0.61+(soil Temp *0.061)- (ln soil moisture *0.8), r 2 = 0.58). At Hafren a weak relationship between (the natural log of) soil respiration and soil available NH 4+ was found. For the entire study soil respiration rates were 14.89 t/ha/y (range 0 to 64.4) for bare soil and 37.18 t/ha/y (range 7 to 24.9). These respiration rates were smaller by a factor of 2 compared to those measured from bare patches of mineral soil on unfertilised and fertilised grassland (Jones et al., 2006).
Methane: Average CH 4 fluxes for each site showed that Ullapool was a net source for CH 4, but Hafren was a net sink and at Glensaugh the peat was a source and the peaty podzols a sink (Table 3.3). Fluxes were very variable and frequently changed between uptake and emission. For the Hafren soils this variability meant that the above general conclusion very much depends on calculating median or averages of the 7 measurement dates (Table 3.3, Figure 3.5). The 7 measurements are averages of flux measurements from 4 chambers, which fortunately were similar to median values.
The woodland at Hafren was a sink for CH 4 only in June 2005 (- 624 µg CH 4 m -2 h -1), a small source (+2.7 to 10.8 µg CH 4 m -2 h -1) in January, March, Sept and Nov 2005, and a substantial source in July and Sept 2004. The September 2004 data are a bit dubious, the 4 chamber fluxes are all positive, but some are unusually large (22060, 9820, 2591 and 63 µg CH 4 m -2 h -1), and therefore are removed from the general interpretation of the summary data (Figure 3.5). However these data show that under certain conditions forests can be a source rather than a sink of CH 4.
Figure 3.5 Temporal variability in CH 4 flux at the Hafren sites
Methane fluxes depend on soil wetness. Soil wetness was expressed here as water filled pores space, which was calculated from the volumetric moisture content (the product of gravimetric moisture content, measured every time fluxes were measured, and the bulk density, measured once at each site) and the porosity. Water filled pore space ( WFPS) was calculated by dividing the volumetric water content by total soil porosity. Total soil porosity was calculated according to the relationship: soil porosity = (1- soil bulk density/1.4), assuming a particle density for peat of 1.4 g cm -3 (Rowell, 1994). It is interesting that at a WFPS < 50% CH 4 flux was very small and that at WFPS > 50% both CH 4 uptake and emission occurred (Figure 3.6). Maximum uptake occurred at the WFPS range of 70-90 %, and maximum emission at the range 90-100%. These differences can partly be explained by differences in soil available NH 4 concentration. Inclusion of these in a multi-regression equation provided a good fit of the CH 4 data (Figure 3.7).
This is in agreement with the relationship of soil available NH 4 and total C content (Figure 3.1). The larger the C content, the wetter the soils, the smaller the chance for nitrification, but the larger the opportunity for CH 4 oxidation to occur.
The magnitude of the CH 4 fluxes reported here reflects the range of those measured by CEH in previous studies. Combined average CH 4 fluxes from organic soils are: 3.9 kg CH 4/ha/y (range -3.95 to 62.4, n = 44).
Figure 3.6 Average, minimum and maximum CH4 concentrations at various ranges of WFPS. The data are calculated from average fluxes from 4 chambers measured on 6/7 dates at all sites.
Figure 3.7 The dependence of the (the natural log of) CH4 flux on soil available NH4 concentration and water filled pore space of the soil. Data are averages of 4 measurements . In order to transform the negative CH4 fluxes, all data were increased to positive values by adding 1.5.
1) Typical large spatial and temporal variability was observed for all fluxes measured.
2) Methane and N 2O fluxes were bi-directional, fluxes were of the same order of magnitude as observed in previous studies.
3) Soil respiration rates were smaller than those measured from bare patches on mineral unfertilised and fertilised grasslands soils. Respiration rates measured from vegetated plots were larger than from bare plots, it is therefore important to exactly report the measurement surface to modellers so that this can be accounted for when using the results for calibration or model testing.
4) Relationships of fluxes with measured parameters were better for the averages of the entire dataset than for individual sites or dates.
5) The influence of N deposition on N 2O fluxes was not clear, due to possible overriding influence of other variables (livestock and runoff).
6) N 2O fluxes from organic soils correlated with total N content in soil and soil respiration rates.
7) Soil respiration could be calculated from the total soil C and N content and the bulk density of the soil.
8) Methane fluxes were depended on the WFPS and could be calculated from the soil available NH 4 concentration and the WFPS.
3.5 13CO 2 pulse labelling
Methods: 13CO 2 pulse labelling of cores was carried out at CEH Lancaster to ensure sufficient enrichment of photoassimilate (carbon fixed from the atmosphere by the plant by photosynthesis) to produce a strong 13C rhizosphere C flow pulse into the soil and enable chasing of the pulse to characterise turnover. After pulse labelling, sampling of the cores was carried out at 0, 3, 16h, 6d, 1 month and 3 months to determine 13C fate in vegetation, soil pools and microbial biomass All 13C analyses were carried out using an Infra-Red Mass Spectrometer ( IRMS).
Results: Plynlimon data only are presented here as similar patterns were found at both sites. When vegetation on 3 soil types was pulse labelled with 13CO 2, a weak 13C signal could be detected in the soil within a few hours of the pulse, and this was detectable up to three months (the last sampling point). This agrees with earlier data from pulse labelling at Sourhope on the NERC Soil Biodiversity programme (Rangel-Castro et al., 2004; Rangel-Castro et al., 2005a,b). Only the improved podzol, with its high above ground biomass and its associated high photosynthetic activity, showed a marked peak in 13C in the soil (Figure 3.8).
Figure 3.8 Soil 13C levels for Plynlimon
Figure 3.9 13C soil microbial biomass levels for Glensaugh
The greatest enrichment was seen for 13C soil microbial biomass at Glensaugh for the podzols with unimproved treatments (Fiigure 3.9). This may be due to the slower turnover of 13C in the microbial biomass of the unimproved versus the improved podzol, rather than greater assimilation of 13C. The peat microbial 13C showed the low levels typical of such low productivity systems.
It is interesting to note that the microbial biomass demonstrated very rapid assimilation of 13C. This is a strong aspect of this study as it has often been missed by other studies because the sampling has been too slow to catch the very rapid pulse of 13C from the plant roots, primarily through exudation. Stable isotope probing of the nucleic acid pools of the Glensaugh microbial biomass is currently underway and will be published during 2007.
Figure 3.10 13C soil microbial biomass levels for Plynlimon
In contrast to Glensaugh, higher enrichments of 13C in the soil microbial biomass were associated with the peat at Plynlimon (Figure 3.10). This is a surprising result as one would expect greater supply of 13C from the vegetation associated with the podzols. It could be, however, that the relatively high substrate assimilation efficiency of an almost entirely fungal population (often around 40-50% compared to around 20% for many bacteria) explains the greater enrichment in the peat. Further microbial investigation, based upon selective probes of DGGE (Denaturant gradient gel electrophoresis) also revealed a high actinomycete population in the peat and this could also be a factor in the high 13C enrichment compared to the podzols at Plynlimon.
Mean residence time calculations for shoots (Table 3.4) revealed a broadly similar turnover across the different soil/vegetation systems, with a mean residence time of between one and two months.
Table 3.4 Mean residence times - shoots
Table 3.5 Mean residence times - soils
Mean residence times for the plant fixed 13C pulse in soils highlighted the very slow turnover in peat compared to the improved podzol (Table 3.5). The data in this matrix are not yet complete due to problems in obtaining reliable natural abundance data for some systems.
The final strand of the research in this work package involved using a combined stable isotope and molecular approach (stable isotope probing of the nucleic acid and fatty acid pools of the soil microbial communities of the study sites) to identify the key players in terms of carbon processing in the soil. In the two figures (stable isotope probing of denaturation gradient gel electrophoresis) below, the incorporation of 13C into soil actinomycete RNA from pulse labelling of the vegetation sward on the improved and unimproved podzols at Glensaugh is illustrated as an example. The figures firstly demonstrate the complexity of the active microbial community (or, in this case, a sub-set of it) in terms of carbon processing. They also illustrate, because of the changing band profile at time 0 and time 3h, that there is a broader community response to the low molecular weight carbon compounds (dominated by glucose and simple organic acids and amino acids) first 13C labelled than there is with time when there is activation of a more restricted, specialised microbial community degrading the higher molecular weight, labelled carbon. In terms of a comparison between unimproved and improved podzols, the figures show that the active soil community is less complex in the unimproved compared to the improved podzol. This agrees with the findings of Rangel-Castro et al. (2005a, b) for soil bacterial and fungal communities at a similar site in the Scottish borders. Furthermore, the active microbial community at time 0 is simpler in the unimproved compared to the improved podzol.
These findings of the nature of the active microbial communities processing carbon in upland, organic soils using state of the art 13C stable isotope probing are exciting and new and are currently being more fully evaluated through confirmatory application of 13C phospholipid fatty acid probing.
Figure 3.11 SIP- DGGE of actinomycete community analysis - podzols at Glensaugh, unimproved (left) and improved (right)
Conclusions: The 13C data presented in this report are particularly exciting and reveal a great deal about C cycling and turnover in upland acid systems for the first time. They highlight the rapid capture of the plant derived 13C in the soil microbial biomass and the slow turnover of this material in the acid soils of this study, particularly in the blanket peats. The 13C SIP data reveal the active components of the soil microbial community in terms of carbon processing, and highlight that a greater proportion of the community is involved in processing of simpler C compared to more complex C and how the complexity of the active community changes under soil improvement. This shows that carbon is captured quickly in upland systems, but that when in the soil, carbon can take a long time to turn over. These data will be used to calibrate the processes of C capture and soil C turnover in versions of the model that have an explicit representation of plant growth, which are scheduled for development in future projects.