Long Term Effects of Forest Liming on the Acid-Base Budget

Abstract

In Rhineland-Palatinate (Germany), a high percentage of the forest area is located on poor soils with low buffering capacity. Extensive liming applications were performed to compensate for the negative consequences of acid deposition. In 1988, three experimental sites with untreated control plots and different liming treatments were established in coniferous stands to investigate the effectiveness of liming on acidification and its effect on forest ecosystems. Measuring deposition and seepage waters for 24 years allowed for calculating long-term acid-base budgets. The original approach was expanded by data from a detailed sampling of the forest stand and mineral weathering rates. Without liming, the acid load exceeded the buffer capacity by base cation release from silicate weathering during the whole observation period. As a result, there was a high release of aluminum. After liming seepage water output of organic anions, nitrate and sulfate increased in some cases, leading to a higher acid load. However, the carbonates of dolomitic limestone compensated for a higher acid load, resulting in less aluminum released compared to the control plots. Until sulfate output by seepage water declines and nitrogen emissions are reduced, liming and restricted biomass harvesting are required for forest stands on base poor soils to prevent further acidification, decline of nutrient stocks, and the destruction of clay minerals.

1. Introduction

Since the late 19th century, an increased acid input from anthropogenic activities was observed in Europe and North America [1]. Forest ecosystems were especially affected when their position was exposed and because of their large intercepting canopy surface [2]. Since the 1980s, acid atmospheric deposition has been decreasing, mainly due to reduced sulfur dioxide emissions [3,4,5]. However, the input of nitrogen components still remains at a high level (cf. [6]) and is currently the main source for the acid load entering the forest ecosystems [7,8].

In addition, centuries of intensive litter and timber harvesting contributed to large scale soil acidification [9,10]. In Rhineland-Palatinate, which is located in southwestern Germany (Figure 1), a high percentage of forested areas are located on base-poor soils with small nutrient reserves and low buffer capacity against acidity [11]. As a result, mobilization of aluminum and heavy metals, reduction of base cation reserves, and destabilization of clay minerals could, and still can, be observed for many forest sites [12,13,14,15]. To compensate for the negative consequences of the acid deposition, extensive liming actions with dolomitic lime were performed at an early stage [16,17]. The application of 3 to 4 tons per hectare every 10 years was recommended [2,18].

In 1988, three experimental sites (Figure 1) with different liming treatments were established. The base-poor forest sites were treated once to evaluate the forest management practice of liming and to investigate its effectiveness and impacts on forest ecosystems [19]. Long term input-output element and acid-base budgets were calculated (cf. [14]) to characterize the forest ecosystems of the three experimental sites by the different processes causing the fluxes of acidity and to quantify soil acidification without and with different dosages of dolomitic limestone. Therefore, we investigated if the mobilization of aluminum could be observed in the control plots without liming, which would indicate that proton production processes exceeded the proton consumption by reaction associated with M_b cations (=Ca, K, Mg, Na). In this case, proton consumption is carried out partially by the weathering of Al, Mn and Fe (M_a cations) oxides causing the destabilization of clay minerals [20]. Further, important questions include did the liming treatments counter the effects of acidification in the long term without the negative consequences of nitrate mobilization? These questions will be answered by input-output budgeting.

2. Materials and Methods

2.1. Sites

Each of the three experimental sites, Adenau (AD), Idar-Oberstein (IO), and Hochspeyer (HS) (

Table 1. Information about the three experimental sites. More detailed Information is available in Greve [22].

Study Areas	Adenau (AD)	Idar-Oberstein (IO)	Hochspeyer (HS)
Elevation above mean sea level	580–630 m	540–550 m	385–400 m
Coordinates (ETRS 1989 UTM32N)	X: 364340 Y: 5588220	X: 371450 Y: 5512000	X: 421560 Y: 5476010
Slope (Degree)	7°	4°	3°
Mean annual temperature	7.6 °C	8.3 °C	8.7 °C
Mean annual temperature of the vegetation period	12.6 °C	13.3 °C	14.5 °C
Mean annual precipitation	850 mm	1065 mm	770 mm
Seepage (60 cm)	275 mm	310 mm	180 mm
Parent material	Diluvial loam above devonic quarzite	Diluvial loam above devonic quarzite	Sandstone of the bunter sandstone
Soil Taxonomy (WRB)	Cambisol	Stagnic cambisol	Podzol
Humus form	Mor humus	Mor humus	Raw humus
Soil texture	Clay loam	Clay loam	Loamy Sand
pH(CaCl2): 0–10/20–30 cm	2.9/3.9	3.0/4.0	2.9/3.6
Base saturation: 0–10/20–30 cm	7.2%/2.8%	6.3%/4.6%	10.3%/5.5%
Cation exchange capacity [µeq g−1]: 0–10/20–30 cm	139/46	137/47	103/19
C content [g kg−1]: 0–10/20–30 cm	65.8/12.7	75.0/12.1	72.8/10.7
N content [g kg−1]: 0–10/20–30 cm	3.1/1.3	3.7/1.1	2.7/0.5
S content [g kg−1]: 0–10/20–30 cm	1.49/0.38	1.21/0.46	0.58/0.19
Tree species	Picea abies	Picea abies	Pinus sylvestris mixed with Fagus sylvatica from natural regeneration
Stand age (2016)	81	97	90/96 (Pinus sylvestris)

), was subjected to five liming treatments, ranging from 3 to 15 t ha⁻¹ dolomitic limestone with untreated control plots (

Table 2. Liming treatments (LT) of the three study areas. Lime and fertilizer were applied in December 1988. Patentkali is a potash fertilizer containing potassium sulfate and magnesium sulfate.

LT	Lime and Fertilizer Application	Mg [kg ha−1]	Ca [kg ha−1]	K [kg ha−1]	P [kg ha−1]	S [kg ha−1]	ANC [keq ha−1]	Additional Informations
0	-	-	-					control plot, no treatment
1	Dolomite: 3000 kg ha−1	349	603				51	particle size 0–2 mm
3	Dolomite: 3000 kg ha−1	349	603				53	particle size 0–2 mm
	Hyperphos: 330 kg ha−1	6	96	6	37			Soft ground rock phosphate
6	Dolomite: 5000 kg ha−1	582	1005				85	particle size 0–2 mm
7	Dolomite: 9000 kg ha−	1048	1809				153	particle size 0–2 mm
	Patentkali: 340 kg ha−1	21		85		58		as K2SO4 and MgSO4
	Kieserite: 660 kg ha−1	107				145		as MgSO4
8	Mixture of Dolomite and Hyperphos: 15,000 kg ha−1	1441	3804	25	145		304	particle size 0–0.09 mm mixed with soft ground rock phosphate

). The area of each liming treatment is 2000 m² separated into two subplots of 1000 m² (Figure 1). The control treatment has three subplots of 2125 m² each. The experimental sites are fenced, and managed forest stands are thinned regularly. Lime and fertilizers were spread by hand in December 1988.

Since the establishment of the study, the sampling of seepage water occurred at depths of 60 cm and 10 cm using suction cups and directly below the humus layer by funnel lysimeters. Four suction cups per depth and five funnel lysimeters were installed on each subplot. Six continuously open bulk samplers with a total collection area of 1885 cm² were used to collect the bulk deposition at a nearby clearing and the throughfall on two of the control subplots. The water samples were collected every two weeks and kept in cold storage. For each subplot, the collected water was analyzed once every three months as a mixed sample for each of the different sampling types. The analyses were performed according to the methods of the Handbuch Forstliche Analytik [21].

2.2. Element Fluxes

The soil water fluxes for each experimental site were calculated by a calibrated COUPMODEL, as described in Karl et al. [23]. To calculate the element fluxes, the element concentrations in the seepage water are multiplied with the soil water fluxes, for each depth (cf. [24,25]). The total deposition (TD) was calculated for each experimental site using the bulk deposition and throughfall measurements as inputs into the canopy budget model after Ulrich [26] and Draaijers and Erisman [27]. For the total annual N deposition, the maximum from both canopy budget models plus the input of organic N of the bulk deposition was used, because total N deposition derived from canopy budget models is typically underestimated [8,28].

Additionally, weathering rates, calculated by PROFILE (cf. [29]), were used to derive the proton consumption from base cation release for each experimental site. The updated PROFILE version 4.4 was used, which includes typical minerals found in German soils [30]. A quantitative mineral analysis was performed in 1997 and the surface area of the mineral soil was calculated based on the soil texture, the coarse soil, and the dry bulk density (DBD) after Becker [30]. The long term mean air temperature was used for the soil temperature (cf. [31]). The plant available soil water content (water content reduced by the non-plant available water content below the permanent wilting point) was calculated by the COUPMODEL and used for soil moisture input to the PROFILE model. The influence of CO₂ was removed from the calculations for all minerals by setting the “Rate Constant_CO2” to 30, because of its weak effect on mineral dissolution (cf. [32,33,34]).

The acid load by biomass increment was also calculated. The single tree-based stand simulator SILVA [35], which was adjusted with allometric relations for Rhineland-Palatinate [36,37], was used to calculate the above ground biomass of the different tree compartments for the years 1988 and 2011 (IO, HS), respectively 1988 and 2013 (AD) for each subplot. As input data, the diameter at breast height (DBH) was measured for each tree and the height for every third tree on the experimental site in winter 1988/89 and 2011/12 (IO, HS) respectively 1988/89 and 2013/14 (AD).

In the winter of 2011/12, nine trees of the control plots (treatment 0) and six trees per liming treatment 1 (3 t ha⁻¹), 3 (3 t ha⁻¹ +P), and 8 (15 t ha⁻¹ +P) were felled on each experimental site to obtain information about the incorporated nutrients. These trees were divided into needles, twigs, branches, bark, and wood according to the method described by Pretzsch et al. [37]. Additionally, the wood of the Scots pine trees was separated into sapwood and heartwood. A mixed sample of each compartment per tree was analyzed separately. Under the assumption, the element concentrations gained of the trees from the control plot were identical to the element concentrations at the beginning of the experiment, they were used to calculate the above ground element stocks for all liming treatments and the control at the start of the liming trial in 1988. The above ground element stocks of the years 2011 (IO, HS) respectively 2013 (AD) were calculated by the element concentrations of the trees felled from the according liming treatments. The element concentrations of the liming treatments 6 and 7 were interpolated by linear regression, because the needle and litterfall samples of all liming treatments indicate a significant correlation between liming dosage and element concentration [22]. The differences of the element stocks between these two dates represent the amount of elements incorporated since the beginning of the liming trial. The results were adjusted to account for the period of the other element fluxes by calculating the mean annual incorporation and multiplying it by 24 years.

2.3. Calculation of Acid-Base Budgets

The input-output budgets for the forest soil were calculated based on element fluxes (cf. [14,26]) on a yearly basis for 24 years, from 1989 to 2012. The inputs to the soil-internal element cycle are total deposition and elements released by mineral weathering. The element loss by seepage water is an output from the forest soil. Lime and fertilizers are treated as part of the soil and not as an additional input.

The acid load based on the budgets of NH₄⁺, H⁺, Mn²⁺, Al³⁺, Fe³⁺, SO₄²⁻, NO₃⁻, organic anions (Org⁻), Ca²⁺, K⁺, Mg²⁺, and Na⁺ was calculated for each subplot after Ulrich [26,38]. Dissolved organic carbon (DOC) was measured in the water samples and converted to Org⁻, as described in Mosello et al. [39] with the updated conversion factor of ICP Forests [40]. P was not included because of uncertainties in the input-output budgeting. The element fluxes at 60 cm depth were taken as an output from the ecosystem and subtracted from the accordant element input by TD.

In the original approach by Ulrich [26], mineral weathering and the acid load by biomass increment are not calculated separately but instead included in the element budgets. Positive budgets of NH₄⁺, H⁺, Mn²⁺, Al³⁺, or Fe³⁺ account for a net input of acidity into the ecosystem because these ions count as potential proton donors [14]. Moreover, positive budgets of Ca²⁺, K⁺, Mg²⁺, and Na⁺ add to the acid load, because it is assumed that these M_b cations are incorporated into the biomass increment or are bound to exchange sites of the mineral soil and releasing M_a cations that lead to proton production. Negative budgets of SO₄²⁻, NO₃⁻ or Org⁻ represent the dissolution of aluminum sulfates, nitrification, or the dissociation of dissolved organic acids and contribute also to the acid load. Opposing results of the element budgets lead to proton consumption by acid base reactions like dissolution of M_b and M_a oxides, cation exchange, mineralization, or the formation of aluminum sulfates [26].

Because of the decision to treat the added lime and fertilizer as part of the soil, the M_b cation budgets of the liming treatments are more negative in comparison to the control plots. Lime and fertilizer could also be treated as an additional input, which would lead to positive M_b cation budgets and therefore adding to the acid load for all liming treatments. This acid load originating from the M_b cation input by liming would have to be balanced by protonation of HCO₃⁻, which would complicate the calculations as well as the presented tables, without adding additional insights.

2.4. Modification of the Original Approach

In our modified approach, we use the biomass sampling of forest stand and the PROFILE calculations to get a detailed look at the budget of M_b cations in the original calculation.

M_b cations are also taken up by the forest stand and are incorporated into the biomass, which leads to proton production. For the calculation of the acid load, the incorporation of elements in the aboveground biomass (net element uptake by the forest stand that is needed for long-term growth) is taken into account. The element uptake to supply roots, leaves, and needles, which is not removed by harvesting, or returned to the forest floor by fine root turnover or litterfall, is part of the ecosystem-internal element cycle and was not included in the input-output budgets. The cation excess equals the proton production by biomass increment [26] and was calculated based on the analyzed elements:

$${\mathrm{H}}^{+}={\mathrm{K}}^{+}+{\mathrm{Ca}}^{2+}+{\mathrm{Mg}}^{2+}+{\mathrm{Na}}^{+}+{\mathrm{Mn}}^{2+}+{\mathrm{Fe}}^{3+}+{\mathrm{Al}}^{3+}-{\mathrm{SO}}_4^{2-}-{\mathrm{H}}_2{\mathrm{PO}}_4^{-}\left[{\mathrm{keq}\mathrm{ha}}^{-1}\right]$$

In this calculation, the influence of the utilized N form is not taken into account, because of the uncertainty of their availability to and uptake by the forest stand. It is assumed that both N forms are taken up in equal shares. A higher NH₄⁺ uptake would increase the acid load whereas a higher NO₃⁻ uptake would lead to a lower acid load by biomass increment [41].

To calculate the M_b cation exchange and dissolution of calcium and magnesium carbonates included in the dolomite, the amount of M_b cations incorporated into the biomass and lost by seepage water output are subtracted from M_b cation release by mineral weathering and deposition input. For the control plots, this difference represents the proton consumption or production by M_b cation exchange processes. On the liming trials, the dissolution of carbonates is also included in this difference.

Differences between the budgets of the subplots within an experimental site are the result of differences in element loss by seepage, as well as the incorporation of elements in the forest stand, which were measured separately for each subplot. Deposition and mineral weathering were assumed to be identical for all subplots of an experimental site.

3. Results

3.1. Site Characteristics

We use the original approach to characterize the three sites because the complex processes are summarized to a greater extent. The acid load and the acid-base reactions, based on the element budgets of the control plots, show clear differences in their composition between the three experimental sites (

Table 3. Total proton production and proton consumption by the different element budgets accumulated over 24 years (1989 to 2012) for the control subplots and liming treatments of the three experimental sites.

			Proton Production [keq ha−1 24a−1]										Proton Consumption [keq ha−1 24a−1]							Portion
Site	Liming Treatment	Subplot	H+	NH4+	NO3−	SO42−	Org−	Ma	Mb	∑	Diff.	∑	H+	NH4+	NO3−	SO42−	Org−	Ma	Mb	Ma	Mb
AD	0	1	5.4	31.1	0.2	5.8	6.6	0.1	5.5	54.6	0.6	54.0	1.2	0.0	18.5	2.0	0.0	29.9	2.5	55%	5%
	0	2	6.5	31.1	1.8	4.0	4.3	0.1	5.6	53.4	−0.7	54.1	0.7	0.0	6.5	3.6	0.0	40.1	3.2	74%	6%
	1	1	6.0	31.4	4.5	13.8	7.2	0.1	4.5	67.5	−1.2	68.7	0.7	0.0	8.6	0.4	0.0	34.0	24.9	50%	36%
	1	2	6.3	31.0	7.9	22.3	6.0	0.1	3.9	77.6	−2.5	80.1	0.8	0.0	1.7	0.5	0.0	39.3	37.7	49%	47%
	3	1	6.4	30.7	2.4	9.5	6.9	0.1	4.4	60.3	0.1	60.2	1.6	0.0	7.7	1.2	0.0	35.9	13.9	60%	23%
	3	2	5.6	31.4	1.5	3.7	10.2	0.0	3.7	56.1	−0.5	56.5	1.0	0.0	7.5	3.4	0.0	29.3	15.4	52%	27%
	6	1	6.3	31.5	9.2	11.3	7.1	0.1	4.2	69.7	0.2	69.5	0.6	0.0	1.3	1.9	0.0	35.8	29.9	52%	43%
	6	2	5.0	31.0	22.3	15.5	11.0	0.0	3.2	88.0	−2.3	90.4	1.2	0.0	1.3	0.9	0.0	38.6	48.4	43%	54%
	7	1	6.0	31.4	26.0	26.6	11.1	0.0	3.6	104.8	−5.5	110.3	0.9	0.0	0.0	0.6	0.0	44.5	64.3	40%	58%
	7	2	6.7	31.3	10.9	29.1	9.4	0.0	5.1	92.5	−4.8	97.2	0.9	0.0	5.9	1.0	0.0	47.0	42.5	48%	44%
	8	1	7.1	31.6	7.1	14.7	10.0	0.0	3.5	74.0	−2.6	76.6	0.4	0.0	1.9	1.7	0.0	26.0	46.6	34%	61%
	8	2	6.5	31.3	9.9	14.2	9.0	0.1	3.3	74.2	−1.7	75.9	0.5	0.0	3.7	1.2	0.0	25.7	44.7	34%	59%
IO	0	1	13.8	13.7	0.0	18.3	2.5	0.2	5.7	54.1	−5.7	59.8	0.0	0.0	20.8	0.3	0.0	35.0	3.7	58%	6%
	0	2	13.8	13.7	0.0	34.7	3.3	0.2	3.0	68.7	−6.7	75.4	0.0	0.0	17.6	0.0	0.0	46.5	11.3	62%	15%
	1	1	15.0	13.7	0.0	18.9	2.9	0.1	3.9	54.6	−6.5	61.1	0.0	0.0	21.4	0.5	0.0	30.2	9.1	49%	15%
	1	2	14.9	13.7	0.0	32.2	3.6	0.1	2.3	66.8	−6.2	73.0	0.0	0.0	14.3	0.0	0.0	31.6	27.2	43%	37%
	3	1	13.8	13.7	7.3	34.0	3.7	0.1	1.6	74.2	−7.7	81.9	0.0	0.0	14.7	0.1	0.0	39.8	27.3	49%	33%
	3	2	15.8	13.7	0.0	13.8	1.6	0.1	4.1	49.1	−6.9	56.0	0.0	0.0	21.1	0.6	0.0	27.7	6.6	49%	12%
	6	1	14.7	13.5	2.2	29.6	9.7	0.1	0.8	70.6	−10.4	81.0	0.0	0.0	12.4	0.0	0.0	27.3	41.3	34%	51%
	6	2	15.3	13.7	0.0	30.6	3.1	0.1	1.9	64.7	−7.1	71.8	0.0	0.0	17.3	0.3	0.0	26.9	27.3	38%	38%
	7	1	15.2	13.8	0.1	29.1	2.5	0.1	2.3	63.2	−6.9	70.2	0.0	0.0	17.4	0.0	0.0	30.9	21.9	44%	31%
	7	2	15.4	13.5	0.3	38.3	6.7	0.1	0.7	75.1	−6.1	81.1	0.0	0.0	13.0	0.0	0.0	25.4	42.7	31%	53%
	8	1	15.8	13.8	0.1	25.7	2.7	0.2	2.5	60.8	−6.5	67.3	0.0	0.0	15.3	0.2	0.0	23.0	28.8	34%	43%
	8	2	15.9	13.8	0.0	31.0	4.0	0.1	1.4	66.2	−5.5	71.8	0.0	0.0	18.7	0.6	0.0	20.5	31.9	29%	45%
HS	0	1	2.9	11.1	0.0	0.5	16.5	0.1	2.8	33.8	−3.8	37.6	0.5	0.0	14.3	2.0	0.0	18.2	2.6	48%	7%
	0	2	5.9	11.2	0.0	3.3	3.1	0.2	3.3	27.0	−2.8	29.7	0.0	0.0	14.3	0.1	0.0	13.4	1.9	45%	6%
	1	1	5.1	11.2	0.0	1.0	16.4	0.1	1.5	35.2	−6.3	41.5	0.0	0.0	14.3	1.6	0.0	20.0	5.6	48%	13%
	1	2	4.6	11.1	0.0	2.8	16.4	0.1	2.4	37.4	−8.0	45.4	0.2	0.0	14.3	1.1	0.0	22.8	7.0	50%	15%
	3	1	5.9	11.2	0.0	8.4	6.9	0.1	1.5	33.9	−3.3	37.2	0.0	0.0	14.2	0.4	0.0	9.2	13.5	25%	36%
	3	2	5.6	11.2	0.0	2.9	9.3	0.1	3.5	32.7	−5.0	37.7	0.0	0.0	14.3	1.5	0.0	16.4	5.5	44%	15%
	6	1	5.5	11.2	0.0	2.2	12.2	0.1	2.2	33.3	−10.6	43.9	0.0	0.0	14.1	0.5	0.0	20.0	9.3	46%	21%
	6	2	6.0	11.2	0.0	5.0	11.3	0.1	2.5	36.0	−4.3	40.3	0.0	0.0	14.1	0.2	0.0	12.7	13.3	32%	33%
	7	1	5.5	11.2	0.0	9.4	14.6	0.1	2.1	42.9	−6.6	49.5	0.0	0.0	14.3	1.5	0.0	19.8	13.9	40%	28%
	7	2	4.2	11.1	0.0	14.8	10.0	0.1	1.8	42.0	−11.7	53.8	0.2	0.0	14.3	0.3	0.0	21.9	17.0	41%	32%
	8	1	5.6	11.2	0.0	5.6	14.2	0.1	2.5	39.2	−4.9	44.1	0.0	0.0	14.2	0.4	0.0	16.0	13.4	36%	30%
	8	2	6.1	11.1	1.3	2.8	25.9	0.1	1.7	49.1	−9.6	58.6	0.0	0.0	11.9	1.7	0.0	22.9	22.1	39%	38%

). At AD, the main source of proton production is the N budget because of high NH₄⁺ input. In some years, the NO₃^- output exceeds the NO₃⁻ input, also leading to proton production. In IO, the input of NH₄⁺ is lower and almost all the N is retained in the ecosystem so that the combined N budget of NH₄⁺ and NO₃⁻ still adds to the proton consumption. Instead of N, the release of stored SO₄²⁻ and the retention of H⁺ contribute primarily to the acid load of this forest site. In HS, the main source of proton production is the loss of organic acids in combination with the input of NH₄⁺. Like in IO, almost all NO₃⁻ is retained in the ecosystem, noticeably contributing to the proton consumption.

The proton consumption for all control plots of the three study areas occurs mostly by M_a cation exchange or weathering, especially of aluminum. Except for subplot 2 in AD, the accumulation of NO₃⁻ contributes noticeably to the proton consumption. In IO and HS, higher proton consumption is calculated compared to the proton production (Difference unequal zero). This could be caused by an underestimation of the flux of organic acids (cf. [26]). Additionally, natural variation inside the experimental plots may contribute to the errors.

3.2. Effect of Liming Treatments

The liming treatments in AD show a higher acid load compared to the control plots. The low-dose as well as the high-dose liming treatments increased the dissolution of sulfates and the nitrification, leading to a greater loss of S and N by seepage water flux [42]. Especially the lower retention of NO₃⁻ reduces the proton consumption by the acid-base reactions of the N budget in comparison to the control plot. The high acid load of treatment 7 is caused by the additional application of sulfur bound K and Mg fertilizers (cf. Table 2), which leads to a high sulfate flux by seepage water accompanied by cation loss.

The forest ecosystem in IO shows a different reaction to the liming treatment. Acid load does not increase on the different liming treatments in general. There are plots of the liming treatments with similar, higher, or lower acid load compared to the control plots. In HS, some liming treatments show higher acid load due to higher flux of organic acids with the seepage water, though the acid load in HS for all treatments is lower than even the acid load of the control plots in AD and IO.

For all three experimental sites, the bigger part of the acid load on the control plots is compensated by M_a cation release. Although the acid load increases for most liming treatments, the proportion of proton consumption by M_a cations decreases with increasing dosage for all study areas. In AD and HS, the absolute portion [keq ha⁻¹] remains on a similar level to the control plots and in IO both absolute portion and relative proportions decrease. This shows clearly that the additional acid load of reactions triggered by the liming is compensated by the dissolution of calcium and magnesium carbonates included in the dolomite.

The contribution of the different element budgets to H⁺ buffering or production is shown more clearly when proton production and consumption cations (which are shown separately for the control plots in Table 3) are summed up (

Table 4. Columns a-f and h contain the net proton production (>0)/proton consumption (<0) accumulated over for 24 years (1989 to 2012) [keq ha⁻¹ 24a⁻¹] which is shown separately in Table 3. Column g contains remaining acid load when the M_b budget (h) is not included. Columns i-n give a detailed look at the processes involved in the M_b budget. M_b accumulation or release (n): k + l − j − m; Acid/base reactions through M_b budget (h): j + n − k.

			a	b	c	d	e	f	g	h	i	j	k	l	m	n
Site	Liming Treatment	Subplot	Acid/Base Reactions through Budget of						Acid Load	Budget of	Acid Load by above Ground Biomass Increment		Mineral Weathering	Deposition	Seepage Water	Accum./Release
			H+	NH4+	NO3−	SO42−	Org−	Ma	∑	Mb	Total	Only Mb
	0	1	4.1	31.1	−18.3	3.8	6.6	−29.9	−2.5	3.1	17.5	15.9	8.9	29.1	26.0	−4.0
	0	2	5.8	31.1	−4.8	0.4	4.3	−40.0	−3.1	2.4	16.8	15.3	8.9	29.1	26.6	−4.0
	1	1	5.3	31.4	−4.1	13.4	7.2	−33.9	19.3	−20.5	24.6	24.6	8.9	29.1	49.5	−36.2
	1	2	5.5	31.0	6.2	21.8	6.0	−39.2	31.3	−33.8	21.9	22.0	8.9	29.1	62.9	−46.9
	3	1	4.8	30.7	−5.3	8.3	6.9	−35.8	9.5	−9.5	25.9	24.9	8.9	29.1	38.5	−25.4
AD	3	2	4.5	31.4	−5.9	0.3	10.2	−29.2	11.2	−11.6	26.9	26.0	8.9	29.1	40.7	−28.8
	6	1	5.7	31.5	7.9	9.4	7.1	−35.7	25.9	−25.7	23.9	24.4	8.9	29.1	54.7	−41.2
	6	2	3.9	31.0	21.0	14.5	11.0	−38.5	42.9	−45.2	23.4	23.9	8.9	29.1	74.3	−60.2
	7	1	5.2	31.4	26.0	26.0	11.1	−44.5	55.2	−60.7	25.0	25.9	8.9	29.1	89.8	−77.7
	7	2	5.8	31.3	5.0	28.0	9.4	−47.0	32.6	−37.4	28.9	30.2	8.9	29.1	66.5	−58.7
	8	1	6.8	31.6	5.2	13.0	10.0	−26.0	40.6	−43.2	24.7	26.5	8.9	29.1	72.3	−60.8
	8	2	6.0	31.3	6.2	13.0	9.0	−25.7	39.7	−41.4	30.1	32.0	8.9	29.1	70.5	−64.5
	0	1	13.8	13.7	−20.8	18.0	2.5	−34.8	−7.7	1.9	17.2	13.7	11.4	18.2	16.2	−0.3
	0	2	13.8	13.7	−17.6	34.7	3.3	−46.3	1.6	−8.3	20.5	16.3	11.4	18.2	26.5	−13.3
	1	1	15.0	13.7	−21.4	18.3	2.9	−30.0	−1.4	−5.1	16.4	15.5	11.4	18.2	23.3	−9.2
	1	2	14.9	13.7	−14.3	32.2	3.6	−31.5	18.6	−24.8	24.0	22.6	11.4	18.2	43.0	−36.1
	3	1	13.8	13.7	−7.3	33.9	3.7	−39.7	18.0	−25.6	17.3	17.2	11.4	18.2	43.8	−31.4
IO	3	2	15.8	13.7	−21.1	13.2	1.6	−27.6	−4.4	−2.5	22.9	22.3	11.4	18.2	20.7	−13.4
	6	1	14.7	13.5	−10.3	29.6	9.7	−27.3	30.0	−40.5	25.9	26.2	11.4	18.2	58.6	−55.3
	6	2	15.3	13.7	−17.3	30.3	3.1	−26.8	18.3	−25.4	22.1	22.8	11.4	18.2	43.6	−36.8
	7	1	15.2	13.8	−17.2	29.1	2.5	−30.7	12.7	−19.6	28.3	29.3	11.4	18.2	37.8	−37.5
	7	2	15.4	13.5	−12.7	38.3	6.7	−25.3	35.9	−42.0	25.0	26.4	11.4	18.2	60.1	−57.0
	8	1	15.8	13.8	−15.2	25.5	2.7	−22.8	19.7	−26.3	26.1	28.1	11.4	18.2	44.4	−43.0
	8	2	15.9	13.8	−18.7	30.4	4.0	−20.4	25.0	−30.5	26.2	29.0	11.4	18.2	48.7	−48.1
	0	1	2.4	11.1	−14.3	−1.5	16.5	−18.1	−3.9	0.1	10.9	11.1	2.7	11.6	11.5	−8.3
	0	2	5.9	11.2	−14.3	3.2	3.1	−13.2	−4.1	1.3	12.5	12.6	2.7	11.6	10.3	−8.6
	1	1	5.0	11.2	−14.3	−0.6	16.4	−19.9	−2.2	−4.1	18.5	18.9	2.7	11.6	15.7	−20.3
	1	2	4.4	11.1	−14.3	1.7	16.4	−22.7	−3.4	−4.6	13.1	13.5	2.7	11.6	16.2	−15.4
	3	1	5.9	11.2	−14.2	8.1	6.9	−9.0	8.8	−12.0	15.6	15.9	2.7	11.6	23.6	−25.2
HS	3	2	5.6	11.2	−14.3	1.5	9.3	−16.3	−3.1	−1.9	11.3	11.3	2.7	11.6	13.5	−10.5
	6	1	5.5	11.2	−14.1	1.7	12.2	−19.9	−3.5	−7.1	11.4	11.7	2.7	11.6	18.7	−16.1
	6	2	6.0	11.2	−14.1	4.8	11.3	−12.6	6.6	−10.8	14.1	14.5	2.7	11.6	22.4	−22.6
	7	1	5.5	11.2	−14.3	7.9	14.6	−19.6	5.2	−11.8	18.3	19.1	2.7	11.6	23.4	−28.3
	7	2	3.9	11.1	−14.3	14.5	10.0	−21.7	3.5	−15.3	16.6	17.1	2.7	11.6	26.8	−29.6
	8	1	5.6	11.2	−14.2	5.2	14.2	−15.9	6.0	−10.9	16.5	17.1	2.7	11.6	22.5	−25.3
	8	2	6.0	11.1	−10.6	1.1	25.9	−22.7	10.8	−20.4	19.1	20.1	2.7	11.6	32.0	−37.8

). In this table, we also included the additional data of mineral weathering and biomass production to examine the effects of the liming treatments and the involved processes on the M_b budget in more detail.

For the liming treatments, the acid load through biomass increment increases compared to the control plot because of higher incorporation of Ca and Mg in all biomass compartments and because of increase in biomass increment [22]. With regard to this calculation, it is assumed that NH₄⁺ and NO₃⁻ are taken up with the same proportion of 50%. In HS and AD an uptake of N only as NH₄⁺ could almost double the acid load compared to a proportion of 50% NH₄⁺. In the case of IO an uptake of N only in the form of NO₃⁻ would not only lead to a lower acid load, but result in proton consumption by biomass increment for the liming treatments (Figure 2).

The M_b cation input by mineral weathering and deposition is on all control plots not high enough to compensate the loss by seepage water and uptake by the forest stand. This indicates that whole-tree harvesting is not nutrient sustainable for these forest stands without any nutrient return. HS has an especially low mineral weathering rate because of the nutrient and base poor parent material. The input of M_b cation depends almost solely on deposition rates.

4. Discussion

Liming forests in Rhineland-Palatinate was performed under the assumption of rapid acidification of soils taking place by air pollutants from anthropogenic sources [43]. During the 24 years, the control plots show that the acid load noticeably exceeds the proton consumption by M_b cations. Without liming, M_b cations compensated for less than 15% of the acid load (Table 3). As a result M_a cations, especially Al and Mn, are released, which can destabilize clay minerals [15,44], disturb plant nutrition [45,46,47], damage fine root systems [48], and reduce the activity of soil biota [49]. With regard to the three study plots, liming partially increased the acid load because of (1) a higher output of anions (SO₄²⁻, NO₃⁻, Org⁻) with the seepage water and (2) a higher proton production through the incorporation of more cations into the biomass. However, the reduction of M_a cation release below the level of the control plot shows that the higher acid load was compensated by the carbonates of the dolomitic limestone. The clay minerals were stabilized on the plots of the liming treatments whereas on the control plots an ongoing destabilization could be observed. However, this stabilization effect started to diminish for the low-dose treatments after approximately 20 years and was limited to the upper 5 cm of the mineral soil [50,51,52].

On the three sites, the loss of anions is linked with a loss of metal cations, which is equivalent to quantitative acidification of the ecosystem [14]. Negative consequences are the depletion of the soil nutrient reserves if M_b cations are leached together with anions, or the pollution of spring water and other adjacent fresh water if the anions are relocated together with M_a cations like Al³⁺. The high loss of SO₄²⁻ and aluminum in AD caused by the liming treatments indicates dissolution of aluminum sulfates due to changed pH value (cf. [53]). To a lower extent, this was also observed in IO at a soil depth of 10 cm [22]. In IO the SO₄²⁻ output of the control plot is already on a high level whereby liming induced none or only a slight increase. In HS, the missing effect of the liming treatments on the SO₄²⁻ output is caused by the soil type. Sandy soils have a lower storage capacity for SO₄²⁻ than more cohesive soils with a higher percentage of clay [54]. Because of the lower S pools in HS [22], there is a smaller potential of a high long-term SO₄²⁻ release, as observed in AD and IO.

The liming treatment 7 with the additional sulfur bound K and Mg fertilizers (Table 2) has the highest acid load due to the high SO₄²⁻ output. The loss of SO₄²⁻ led to a coupled output of Mg and Al. On the other hand, K was retained in the ecosystem. Higher K fluxes compared to the control plots could only be observed in soil depth of 10 cm, but not below the rooting zone in 60 cm soil depth. Also, the forest stands show a tendency for higher K concentrations in older needles and in needle litter compared to the other high-dose liming treatment 5 and 8 [22].

The N budget in AD is the main source for the proton production on the control plot (Table 4). The loss of more than 5 kg NO₃⁻ ha⁻¹ a⁻¹ [22] indicates N saturation on a low level (cf. [55]). Liming increases the leaching of NO₃⁻ and therefore the acid load. This is an example of the risks which are associated with liming of N saturated forest areas. However, on the control plot, NO₃⁻ is for the most part accompanied by Al which could be a risk for biota living in adjacent freshwater ecosystems and can cause higher costs and increased technical effort to remove aluminum from drinking water [56]. After liming the absolute portion and relative proportion of Al is reduced and instead of Al more M_b cations (especially Mg) are transported together with NO₃⁻ to the outside of the ecosystem.

To improve the accuracy of the acid load by biomass increment it is important to know in which form N is taken up by the forest stand. The sample calculations for the three study areas show that uptake of N only in the form of NO₃⁻ reduces the acid load considerably, in the case of IO even below zero, which means net proton consumption. On the other hand, a high proportion of NH₄⁺ could almost double the acid load. Experiments indicate for Picea abies that a large proportion of its N uptake from the soil is NH₄⁺ [57,58,59]. Based on the literature, it is more likely that there is an underestimate in the acid load by biomass increment when the influence of the utilized N form is not taken into account.

For the long-term, it is unknown if liming leads to a loss of a greater amount of N if a reduction of N deposition occurs. It could be hypothesized, that the amount of N lost with seepage water is the same. Under current deposition, limed ecosystems could show a more rapid increase of N in seepage waters. However, with reduced N deposition the limed areas may begin to store N and evade N saturation at an earlier point in time. Liming increases tree growth and additional N can be retained by the vegetation or removed by harvesting.

In IO and HS, liming leads also to a slight increase of the acid load from the N budget but is less important to the other proton sources. Most of the N input is retained in the ecosystem of these two study areas even on the high-dose treatment 8 with 15 t/ha. In the case of IO liming could prevent the ecosystem to reach N saturation as there was significantly more N taken up and stored in the stem wood of the trees on the liming trials [22]. Possible reasons for this could be a higher number of living cells (cf. [60]) or that N is taken up to a greater extent as NO₃⁻ instead of NH₄⁺, which is stored in the wood (cf. [61]). Why this effect could not be observed for the spruce forest stand in AD is not clear.

This study clearly shows the importance of minimizing the input of N into forest ecosystems as long as most of the N is still retained and not lost as nitrate by seepage water flux. Otherwise, nitrate loss would likely continue over decades similar to the sulfate outputs (cf. [3,62]).

The lower amount of anions in the seepage water in HS compared to AD and IO leads to stronger retention of the applied M_b cations (Mg and Ca). Therefore, sandy soils should not be excluded automatically from liming treatments. Instead, the period between liming treatments can be longer for sandy soils with similar conditions to HS. The lower cation exchange capacity in HS [22] does not cause higher outputs of Mg and Ca by the seepage water.