Excess nitrogen (N) deposition and gaseous N emissions from industrial, domestic, and agricultural sources have led to increased nitrate leaching, the loss of biological diversity, and has affected carbon (C) sequestration in forest ecosystems [1
]. Global mean N deposition is still increasing [5
] while in Europe, N deposition peaked during the mid-1980s and was slowly going down until present day as a result of emission reductions (EU28) of NOx
by 50% and NH3
by 30% between the years 1990 and 2015 [6
]. However, the amount of N deposition did not decrease in all areas in Europe [7
] and the currently legislated emission reduction targets are too low to save ecosystems and biodiversity from further effects [8
]. In order to assess the ecosystem effects of chronically high N deposition (among other air pollutants) in the forests of the Northern Limestone Alps in Austria, the long-term monitoring station “Zöbelboden” was set up in the year 1992. After 27 years of continuous observation and a number of experimental studies, we provide here an integrated view of these effects. Since half of the Austrian drinking water resources stem from karst areas [10
] such as our study catchment, detailed knowledge about the drivers of N loss in upstream karst areas in the form of the water pollutant nitrate is pivotal. The more so because it may deviate from other catchments owing to a strong heterogeneity of subsurface flow and storage characteristics [11
], and shallow, stony soils with low filtering capacity [12
]. Nitrate may therefore additionally pollute downstream water sources already affected by agricultural fertilization [14
]. An oversupply of N is thought to gradually diminish N retention in the living biomass and the soil so that surplus N leaves the ecosystem via leaching or gaseous emissions [16
]. Cross-catchment studies have however shown that N retention is high even in areas with considerable N deposition [18
], and signs of increasing oligotrophication [20
N retention in trees only decreases beyond a certain level of N deposition while it increases at lower levels [21
]. Since the deposition of N is in the range of expected growth reduction in European beech and Norway spruce [22
], the dominating tree species at Zöbelboden, we expected such an effect. However, we were not able to single out this effect from the manifold factors of tree growth. Therefore, we explored our data on nutrient concentrations in leaves and needles. The down-regulation of tree growth can be related to nutrient deficiencies as a result of previous N-driven growth enhancement, reduced fine root biomass, and a change in the composition and abundance of mycorrhizal fungi [24
Apart from the retention of deposited N in plant biomass, substantial amounts end up in the soil [26
]. Fertilization experiments in forests show N addition generally increases soil organic matter (SOM) storage [27
] but the amount of retained N depends upon the soil C:N ratio because microbial immobilization of added N may be low in soils with a C:N ratio < 20 [17
The controlling role of the soil C:N ratio on N immobilization is also reflected in its tight relationship with the leaching of dissolved inorganic N (DIN), predominately in the form of nitrate [29
]. The retention of DIN in catchments distributed across unmanaged European forest areas with no local N emission sources is high but decreased with increasing inorganic N deposition [18
]. Since our catchment has been exposed to chronic N deposition for at least three decades before the monitoring started, and because DIN leaching seems to react very fast to the addition of DIN [31
], we did not expect a long-term trend in DIN export within the timeframe of monitoring. However, apart from the long-term effects of high N deposition on DIN leaching, short-term disturbances such as forest stand-replacement, either naturally or due to management, can significantly increase nitrate loss with the seepage of water [32
] thereby elevating the catchment runoff of DIN [35
]. Knowledge about the interactive effects of long-term chronic N deposition with stand-replacing disturbances is important [38
], because in Europe’s forests, disturbances from wind, bark beetle and wildfires have already and are expected to increase further in response to climate change [39
]. Beginning during the year 2006, storms and spruce bark beetle outbreaks have caused stand replacement in 5%–10% of the study area with an immediate increase in the DIN runoff [37
]. Therefore, we expected that forthcoming disturbances will likely lead to pulses of nitrate loss in our study catchment, too.
Based on an analysis of the input-output budget of N, we explored vegetation, soil, and other catchment sinks in order to explain the likely effects of N deposition on nitrate leaching and to provide an outlook to the future. In detail, we (1) hypothesized increasing deficiency of P and K as found in other sites with high N deposition [7
], and (2) increased soil N storage [27
] and decreasing C:N ratios [17
] in the soils, which we analyzed with long-term soil inventory data supplemented with a plot-scale N fertilization experiment focusing on SOM decomposition. With climate scenario data, the application of a hydrological model, and the knowledge we have gained through analyzing the long-term data from Zöbelboden, we discuss the potential impact of the manifold drivers on nitrate discharge in the future.
2.1. Study Site
LTER Zöbelboden has a size of 90 ha and is situated in the northern part of the national park “Northern Limestone Alps”, approximately 50 km south of Linz (N 47°50′30″, E 14°26′30″) (https://deims.org/8eda49e9-1f4e-4f3e-b58e-e0bb25dc32a6
) (Figure 1
). The altitude ranges from 550 m to 956 m a.s.l. The main rock type is Norian dolomite (Hauptdolomit), which is partly overlain by limestone (Plattenkalk). Due to the dominating dolomite, the watershed is not as heavily karstified as limestone karst systems but shows typical karst features, such as conduits and sinkholes. These conduits and sinkholes provide pathways for rapid water flow and quick response times to water input at the soil-bedrock interface. The long-term average annual temperature is 7.2 °C. The coldest monthly temperature at 900 m a.s.l. is −1 °C in January, the highest is 15.5 °C in August. Annual rainfall ranges from 1500 to 1800 mm. Monthly precipitation ranges from 75 mm (February) to 182 mm (July). Snowfall occurs between October and May with an average duration of snow cover of about 4 months. From the start of the project in 1992 onwards, forest management has been restricted to single tree harvesting just in case of bark beetle infestation. Wind throw is frequent in the area with single tree events and events affecting larger areas.
The catchment can be divided into two distinct sites: A very steep (30–70°) slope from 550–850 m a.s.l. and an almost flat plateau (850–956 m a.s.l.) on the top of the mountain. The areal coverage of each site is 50% of the watershed. At each site, one plot has been selected for intensive measurements of hydrochemical variables (Figure 1
). Intensive plot I (IP I) is located on the plateau where Chromic Cambisols and Hydromorphic Stagnosols are found. This plot was moved to a nearby location (IP III) in 2008 with the same characteristics because of forest disturbance. Intensive plot II (IP II) is located on the slope and is dominated by Lithic and Rendzic Leptosols (FAO/ISRIC/ISSS, 2006). The mean slopes are 14° at IP I and 36° at IP II. IP I is dominated by Norway spruce (Picea abies
(L.) H. Karst.) following plantation after a clear cut around the year 1910, whereas a mixed mountain forest with beech (Fagus sylvatica
L.) as the dominant species, Norway spruce, sycamore (Acer pseudoplatanus
L.), and ash (Fraxinus excelsior
L.) covers IP II.
Wind throw and bark beetle disturbances started in 2006 with the damages caused by the storm Kyrill followed by two other storms in 2008. The subsequent spruce bark beetle outbreak peaked in the year 2011 (Figure 2
). At Zöbelboden, these disturbances mostly damaged single trees and groups of trees while stand replacement of larger areas (>0.5 ha) only occurred in 5% to 10% of the catchment.
2.2. Climate and Air Pollution
Meteorological data (air temperature, precipitation, vapor pressure, wind speed, and solar radiation) were recorded half-hourly at two climate stations, at a clearing area at the plateau (950 m a.s.l., 280 m distance to the monitoring plot) and at a nearby tower (40 m height, 60 m distance to the plot). The tower can be considered to represent the climatic situation at the monitoring plot, therefore data from the clearing area was regressed to the tower, resulting in a daily record from 1993 to 2019. Data gaps were filled with nearby climate stations based on linear regression (all within a radius of 10 km). Snow depth was measured weekly with a measuring stick during the sampling campaigns. In order to explore the climatic changes since 1950, long-term temperature and precipitation data from a nearby station (Reichraming, Hydrografischer Dienst) at 360 m a.s.l was bias-corrected with on-site meteorological data. We used the monthly mean difference between temperature and precipitation in the years 1993 to 2010, when data from both stations were available and adjusted for those of the entire data series of the Reichraming station.
Bulk precipitation was collected at the clearing (non-forested) area adjacent to the climate station with 5 to 10 (from 2005 onwards) bulk collectors, each with 20 cm in diameter. Water samples were pooled, filtered, and kept cool (4 °C) until sample preparation. Weekly samples were mixed (volume-weighted) biweekly or monthly (from 2008 onwards due to financial reasons). Nitrate was analyzed by ion chromatography with conductivity detection (Dionex IC System 4000 I until 2002, thereafter with Dionex IC System Serie DX 500). Total N was determined by means of spectrophotometric analysis (Abimed TN 05). NH4+ concentrations of the weekly samples were also measured by spectrophotometry (Milton Roy Spectronic).
We calculated the total deposition of N for the two intensively measured sites as the sum of throughfall and canopy exchange. The latter was based on a canopy exchange model according to Staelens et al. [42
] with sodium as the tracer ion, bulk precipitation and throughfall, a yearly time step, and relative uptake efficiency of NH4+
of 6. In order to get a long-term deposition of N, we scaled reconstructed deposition from 1880 to 2000 [43
] to the measurements using the mean bias during the years when both were available as a correction factor. Throughfall was collected at each intensively measured site with 15 regularly distributed bulk deposition samplers (∅ = 20 cm). From 2006–2008, an additional two to five deposition samplers collected throughfall in the small bark beetle gap at IPI. While throughfall measurements ceased at IPI in September 2009 due to small-scale wind throw events and bark beetle infestations, measurements started at IPIII in August 2008. For chemical analyses, throughfall samples were pooled for each individual monitoring plot, except for bark beetle gap samples at IPI, which comprised individual samples. Subsequent throughfall sample preparation and analyses correspond to the described method for bulk precipitation.
2.3. Foliage Nutrient Concentrations
The dominant tree species, Norway spruce and European beech were sampled from a subsample of the forest and soil inventory plots distributed in a 100 × 100 m grid covering the entire study area (90 ha) and from the intensively monitored plots at the plateau (IP I and IP III) and the slope (IP II). Sampling began in the year 1992 for spruce needles and in 1993 for beech leaves. Between 36 and 52 dominant or predominant spruce trees and 16 to 17 beech trees were sampled annually until 2003. Thereafter sampling took place in the years 2004 (subsample of spruce trees), 2006, 2008, 2011, 2014, and 2017. In 2015, 2016, 2018, and 2019, only leaves and needles of the trees at IP I and IP III were collected and analyzed. All sample trees of the inventory plots were located as close as possible but outside the 10 (8) m circle of the forest inventory and the 100 m² vegetation plot. We chose new sample trees as close as possible to prior ones in case of damage or decreasing vitality. During each sampling campaign, we collected a 4 l plastic bag of mature leaves from the upper third of the north-west exposed crown of each beech tree. From spruce trees, the current and one-year-old needles were collected separately in a 0.5 L bag from as close as possible to the 7th whirl (counted from the top) of the north-west exposed crown. In some years until 2003, leaves and needles were collected separately from all four cardinal directions. Sampling took place in late August and September and in late September and October, respectively for beech and spruce. All specimen were labeled and then stored at < 4 °C until pretreatment and analysis in the lab.
The specimens were oven-dried (30 °C) until constant weight, separated from the twigs, and grounded with an ultra-centrifugal grinding mill (ZM1, Retsch, D). Dry mass at 105 °C was determined with a subsample of 100 undamaged spruce needles. Another subsample was digested (HNO3/HClO4) in glass vessels on a heating block (SMA 20 A, Gerhardt, D) to measure total residue with gravimetric vapor sorption at 105 °C. Subsequent determination of Ca, K, Mg, Mn, and total P was carried out with inductively coupled plasma optical emission spectrometry (ICP-OES; Perkin-Elmer Optima 3000 XL, 3000 DV, and 7300 DV). Total N was measured by means of potentiometric titration (Kjeldatherm Vapotest 4S Gerhardt).
Foliage deficiency thresholds were taken from Mellert and Göttlein [44
]. Temporal trends of single nutrients and ratios (N:Ca, N:Mg, N:K, and N:P) were tested using a Wilcoxon rank-sum test after outlier deletion comparing the first 4 years with the last 4 years (boxplot function).
2.4. Soil Chemistry
We collected mixed soil samples at 64 plots between July and August 2014 to compare with the soil survey data of the years 1992 and 2004. These plots were 100 m apart covering the entire catchment. For one mixed soil sample, three individual soil cores with 4 cm diameter were taken from the upper mineral soil layer (0–10 cm) after litter removal. The soil cores were taken with a 2.5 m distance to the 1992 and 2004 sampling sites. In 2004, the samples were taken 5 m apart from the samples in 1992. In all surveys, soil samples were dried at approximately 30 °C, coarse aggregates crushed and dried again until constant weight. Then they were sieved through a 2 mm sieve. Soil suspensions were made by dissolving 5 g of each soil sample in 12.5 mL of 0.01 M CaCl2 solution and pH was measured electrochemically with a pH electrode (Metrohm 654 pH meter in 1992 and 2004 and Argus X pH meter in 2014). For the determination of the C:N-ratio, the soil samples were ground and further decarbonized using 3M HCl solution. The organic carbon content was measured after dry combustion using isotope-ratio mass spectroscopy (IRMS). These methods deviate from the analysis in 1992 and 2004 where the total content of organic C (TOC) was calculated by subtracting the total content of CaCO3 from the total content of C (TC-TIC). The total content of C (TC) was analyzed by dry combustion (1300 °C). Released CO2 was detected coulometrically (Ströhlein Coulomat 702 and Si 111/6). The total content of CaCO3 (TIC) was measured via the addition of HCl and volumetrically determination of the released CO2 (Scheibler). The total content of N of the year’s samples was determined by a modified Kjeldahl method. The organic N was converted to NH4+ by digestion with H2SO4 and a catalyst (Kjeldahlterm KT8 Gerhardt). The accumulated NH4+ was converted to NH3 (distillation) and measured by potentiometric titration. To account for N oxygen compounds, salicylic acid was added prior to digestion. For the determination of total N stocks, the forest floor was sampled once with a 30 × 30 cm frame and the mineral soil with a metal pole with a 70 mm diameter at 3 points within a distance of two meters. Differences between years were tested using a paired Wilcoxon test.
2.5. Catchment Hydrology and N Measurements
Artificial tracer experiments, carried out in our study area, showed the large heterogeneity of the hydrologic system ranging from the retardation of 1 day to 10 years [45
]. Water age dating with CFCs, 3
H, and 3
He even showed mean residence times of up to 20 years at one spring. Estimating the water balance of the larger system demonstrated that major parts of the rainfall input are transformed into the intermediate flow and deep percolation within the dolomite leaving the system as diffuse runoff instead of surface runoff (Humer and Kralik 2008). For this study, our main source of information was the gauging station (number 551) discharge and N concentration data between 2000 and 2018. Damped δ18
O variations at this spring indicated a delayed flow component, both fast and slow flow paths, and a considerable fraction of intermediate flow [46
]. We disregarded the discharge data before and after this period because we deemed this data too uncertain. The discharge was calculated based on the gauging station’s rating curve and 15 min water level sensor data. Missing values were gap-filled based on regression with nearby gauging station data. The recharge area of this spring was estimated based on a model calibration described in detail in [37
Weekly observations of NO3−-N, NH4+-N, and total N were available at the gauging station 551. From 2010 onwards, samples (ISO 5677–6) were filtered (0.45 μm) before the analysis. NH4+ concentrations were measured by spectrophotometry (Milton Roy Spectronic). Weekly NO3− and total N samples were pooled to provide volume-weighted biweekly (until March 2009) and monthly (thereafter) samples. NO3− concentrations were determined by ion chromatography with conductivity detection. DIN input was then calculated as the sum of NO3−-N and NH4+-N. Dissolved organic nitrogen (DON) was calculated as the difference between total N and DIN. In order to calculate DIN discharge, we filled all missing 15 minutes values with the last weekly measurement value.
In 2018 and 2019, a spectral sensor probe (S:CAN spectro::lyser) with a UV-Vis 220–720 nm detector (15 mm path length) was installed in the measuring weir to obtain 15 minutes NO3- measurements.
2.6. Hydrochemical Modelling
We use the VarKarst model [47
] for the discharge simulations, which was already performed satisfactorily in a previous study at this site [37
]. The VarKarst model considers the variability of a karstic system reflected in the variability of (i) soil and epikarst depths, (ii) concentrated and diffuse recharge to the groundwater storage, and (iii) the epikarst and groundwater hydrodynamics. Spatial heterogeneity is accounted for by a set of N model compartments, which represent varying system properties over space. The influence of karst conduits is included by simulating concentrated recharge and fast conduit groundwater discharge. The VarKarst model is primarily running on a daily resolution. Details of the model can be found in [47
For discharge projections, we perform the model simulations with the calibrated model parameters at this same site from the previous study [37
] using different climate projections.
We used eight RCP 8.5 scenarios of bias-adjusted regional climate model (RCM) data from the EURO-CORDEX initiative (Table 1
), the European branch of the Coordinated Regional Downscaling Experiment (CORDEX) project [49
], available through the data nodes of the Earth System Grid Federation (ESGF) model data dissemination system [51
]. RCP 8.5 assumes emissions to rise throughout the 21st century. Because the actual altitude of a site did not match with the altitude of the closest RCM grid element, the 2 m air temperature was height corrected using a hypsometric lapse rate of 0.65 K per 100 m before temporal averaging was done.